WO2023134364A1 - 引导车辆行驶的方法、地图生成方法及相关系统 - Google Patents

引导车辆行驶的方法、地图生成方法及相关系统 Download PDF

Info

Publication number
WO2023134364A1
WO2023134364A1 PCT/CN2022/138461 CN2022138461W WO2023134364A1 WO 2023134364 A1 WO2023134364 A1 WO 2023134364A1 CN 2022138461 W CN2022138461 W CN 2022138461W WO 2023134364 A1 WO2023134364 A1 WO 2023134364A1
Authority
WO
WIPO (PCT)
Prior art keywords
lane
road
topological
curve
curves
Prior art date
Application number
PCT/CN2022/138461
Other languages
English (en)
French (fr)
Inventor
张叶青
陆星阳
程思源
王晨远
万滢
Original Assignee
华为技术有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 华为技术有限公司 filed Critical 华为技术有限公司
Publication of WO2023134364A1 publication Critical patent/WO2023134364A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
    • G01C21/34—Route searching; Route guidance
    • G01C21/3453—Special cost functions, i.e. other than distance or default speed limit of road segments
    • G01C21/3492—Special cost functions, i.e. other than distance or default speed limit of road segments employing speed data or traffic data, e.g. real-time or historical
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
    • G01C21/34—Route searching; Route guidance
    • G01C21/3407—Route searching; Route guidance specially adapted for specific applications
    • G01C21/343—Calculating itineraries, i.e. routes leading from a starting point to a series of categorical destinations using a global route restraint, round trips, touristic trips
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
    • G01C21/34—Route searching; Route guidance
    • G01C21/3446—Details of route searching algorithms, e.g. Dijkstra, A*, arc-flags, using precalculated routes
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
    • G01C21/34—Route searching; Route guidance
    • G01C21/36—Input/output arrangements for on-board computers
    • G01C21/3626—Details of the output of route guidance instructions
    • G01C21/365—Guidance using head up displays or projectors, e.g. virtual vehicles or arrows projected on the windscreen or on the road itself

Definitions

  • the present application relates to the field of vehicle technology, and in particular to a method for guiding a vehicle to travel, a method for generating a map of an intersection, a related system, and a storage medium.
  • the automatic driving system needs to conduct reasonable road topology analysis and provide corresponding Topological navigation guidance information for vehicles to perform intention prediction, trajectory prediction, lane decision-making and motion planning, etc.
  • High-quality road topology analysis requires not only that lane trajectories conform to human driving habits, but also that the number of lane topologies that can be passed reasonably and their navigation guidance information should also be consistent with human driving experience, so as to improve the human-likeness and accuracy of autonomous vehicle trajectories. Intelligent, it is also more accurate to predict the intention and trajectory of other vehicles.
  • the existing road topology analysis methods can be roughly divided into two categories: road topology generation based on high-precision map mapping or road topology generation based on vehicle trajectory clustering.
  • the road topology generation method based on high-precision map drawing mostly relies on the drawing method to obtain the endpoint or direction vector of the road entering and exiting the road, calculate the angle between the direction vector, and then perform curve fitting on the endpoint or direction vector, so the generated
  • the line type is single and the generalization is insufficient, which cannot cover the differentiated scenes in the real scene and requires manual intervention; while the road topology generation method based on vehicle trajectory clustering relies on big data technology to collect human driving data or crowdsourced trajectories, and then Tracks are screened and clustered to generate lane topology curves, which will inevitably bring a lot of data preprocessing work, and the quality of road topology generation is closely related to the quality of collected data and cannot be guaranteed.
  • the direction vector is calculated based on the endpoints or direction vectors of the entry and exit roads at the intersection by selecting the reference line endpoints of the intersection lanes from the nodes at the intersection on the boundary of the two converging roads and determining their boundary line endpoints Finally, curve fitting is performed on the end point or direction vector, and then the virtual lane reference line and boundary in the intersection can be automatically generated.
  • the reference line generated by this scheme has a single line type, and it is necessary to manually adjust the parameters to ensure the output quality for different intersections; and for scenes with non-direct straight intersections, flower beds, curbs, fences and other obstacles, it cannot be automatically generated.
  • the reference curve or the generated reference curve trajectory is not human-like; and for complex many-to-many intersections, the generated road topology is not complete, and there may also be unreasonable road topologies, requiring manual intervention to select a reasonable topology connection relationship.
  • This application discloses a method for guiding vehicles to drive, a method for generating a map of an intersection, a related system, and a storage medium, which can provide corresponding navigation and guidance information for vehicles during automatic driving, and effectively improve the human-likeness of the trajectory of the vehicle passing through the intersection. traffic efficiency.
  • an embodiment of the present application provides a method for guiding a vehicle to travel, including: generating M lane topological curves according to the intersection, the entry road of the intersection, the obstacles in the exit road of the intersection, and the lane lines , the lane topological curve is a curve with the end of the entering lane of the entering road and the starting point of the exiting lane of the exiting road as endpoints; performing rationality detection processing on the M lane topological curves , to obtain K' lane topological curves, wherein K' is not greater than M; when the vehicle is located in the first lane of the incoming road, determine the target path from the K' lane topological curves , the target path includes a lane topological curve whose end point is the end of the first incoming lane among the K′ lane topological curves.
  • the lane topological curves are generated based on the intersection, the entry road of the intersection, the obstacles in the exit road of the intersection, and the lane lines, and then K' lane topological curves are obtained through rationality detection processing.
  • the target path is determined from the K' lane topological curves.
  • the complete and reasonable lane topology curve generated by this solution based on the actual scene is more in line with human driving habits, more reasonable, without manual intervention, has good generalization and human-like nature, and can provide corresponding navigation guidance information for vehicles during automatic driving, effectively Improve the trajectory and traffic efficiency of vehicles passing through the intersection.
  • the generating M lane topological curves according to the intersection, the entry road of the intersection, the obstacles in the exit road of the intersection, and the lane line includes: according to the intersection, the Obstacles in the entry road of the intersection, the exit road of the intersection, and the lane lines that cannot be crossed to obtain the hard boundary constraints of the lane topology curve; according to the intersection, the entry road of the intersection, and the exit road of the intersection can cross the lane line to obtain the soft boundary constraint of the lane topological curve; obtain K virtual boundary constraints of the lane topological curve according to the hard boundary constraint of the lane topological curve and the soft boundary constraint of the lane topological curve; according to the hard boundary constraint of the lane topological curve
  • the constraints, the lane topological curve soft boundary constraints and the K lane topological curve virtual boundary constraints generate the M lane topological curves.
  • This method obtains soft and hard boundary constraints and virtual boundary constraints based on high-precision maps and real-time perception of road obstacles and lane lines, which ensures the human-like nature of virtual lane trajectories and the universality of vehicle types, and improves the ability to generate lane topology curves. reliability.
  • the K lane topological curve virtual boundary constraints correspond to the K lane topologies
  • any lane topology curve virtual boundary constraint A in the K lane topological curve virtual boundary constraints is obtained by placing the leftmost lane of the intersection
  • the hard boundary constraint and/or soft boundary constraint on the left side of the topology is shifted to the right by a first preset distance
  • the hard boundary constraint and/or soft boundary constraint on the rightmost lane topology right side of the intersection is shifted to the left by a first distance
  • the preset distance is obtained, wherein the first preset distance is determined according to the lane order of the lane topology A', or the first preset distance is determined according to the vehicle preset passing width, lane width determined by at least one item of the lane topology A' and the lane position order of the lane topology A', the lane topology curve virtual boundary constraint A corresponds to the lane topology A'
  • the K lane topologies include the leftmost side of the intersection lane topology and the
  • the scheme considers the interference effects of other traffic trajectories, the indirect constraints of soft and hard boundaries on the gradual weakening of lanes in the same direction, etc., and generates virtual boundary constraints to ensure the human-like nature of virtual lane trajectories and the universality of vehicle types.
  • the generating the M lane topological curves according to the lane topological curve hard boundary constraints, the lane topological curve soft boundary constraints and the K lane topological curve virtual boundary constraints includes : Angle sampling is performed on the end of each incoming lane in the incoming road to obtain at least one starting point pose vector in the incoming road, and the angle of each outgoing lane in the outgoing road is respectively Performing angle sampling at the starting point to obtain at least one end point pose vector in the exiting road; performing curve sampling on the at least one starting point pose vector and the at least one end point pose vector to obtain the entering road and A plurality of curves between the outgoing roads; performing screening processing on the plurality of curves according to the hard boundary constraints of the lane topological curves, the soft boundary constraints of the lane topological curves and the virtual boundary constraints of the K lane topological curves , to obtain the topological curves of the M lanes.
  • the screening process may be, for example, to first screen curves satisfying the above constraints, and then to screen until there is at most one optimal curve between each incoming lane and each outgoing lane.
  • each curve is adaptively adjusted, and then the M lane topological curves are obtained.
  • This processing manner is only an example, and it may also be other manners, which are not specifically limited in this solution.
  • This scheme performs angle sampling based on the end of the entering lane and the starting point of exiting the lane, and then generates multiple lane topological curves. Based on the soft and hard boundary constraints and virtual boundary constraints obtained above, the multiple curves are screened, and then Get M lane topological curves. By adopting this method, the human-likeness, flexibility and vehicle type universality of the trajectory of the virtual lane are guaranteed, and the traffic conflict with other lanes is reduced.
  • the performing curve sampling on the at least one start point pose vector and the at least one end point pose vector to obtain a plurality of curves between the entry road and the exit road includes: A plurality of control points are generated between the end of each incoming lane of the incoming road and the starting point of each outgoing lane of the outgoing road; according to the at least one starting point pose vector, the at least A terminal pose vector and the plurality of control points generate the plurality of smooth curves.
  • this scheme not only performs angle sampling based on the end of the entering lane and the starting point of exiting the lane, but also based on control point sampling, which makes the number of generated curves more and improves the flexibility of curve generation.
  • the pose vector of at least one starting point in the entering road is obtained by extending the end of each entering lane by a second preset distance and performing sampling.
  • the sampling points of the beginning and end attitudes of the virtual intersection side trajectory are reasonably extended to the outside, so as to improve the quality of the generated trajectory, enhance the human-like nature of the trajectory, and avoid the trajectory caused by high-precision map drawing It is unreasonable, and it also avoids wrong screening due to unreasonable trajectory generation in topology screening, and ensures the completeness of road topology analysis.
  • performing rationality detection processing on the M lane topological curves to obtain K' lane topological curves includes: according to the direction vector of the entering road, the direction vector of the exiting road
  • the direction vector obtains the projection line between the entering road and the exiting road
  • the projection line is the straight line where the bisector of the angle obtained by the intersection of the direction vector of the entering road and the direction vector of the exiting road is located , or, the projection line is a straight line perpendicular to the direction vector of the exiting road and passing through the starting point of the exiting road
  • the alignment coefficient between each exit lane wherein, the alignment coefficient between each entry lane and each exit lane is the ratio between the first parameter and the second parameter, and the first
  • the parameter is the overlapping length between two line segments obtained by extending the lane edge of each entering lane and the lane edge of each exiting lane respectively to the projection line
  • the second parameter is the The lane sideline of each entering lane and the lane sideline of each exiting lane are respectively
  • the K' lane topological curves include the end of the leftmost inbound lane of the inbound road and the starting point of the leftmost outbound lane of the outbound road.
  • the curve whose starting point is the end point also includes the curve whose end points are respectively the end of the rightmost inbound lane of the inbound road and the starting point of the rightmost outbound lane of the outbound road.
  • the K' lane topological curves include the end of the incoming lane X in the incoming road and the left lane of the outgoing lane Y in the outgoing road.
  • the lane topological curve with the starting point as the end point, and the lane topological curve with the end point of the entering lane X and the starting point of the right lane of the exiting lane Y as the end point also includes the The end point and the starting point of the exit lane Y are lane topological curves, wherein there are lanes on the left and right sides of the exit lane Y; or, the K' lane topological curves include the The end point of the entry lane X in the entry road and the starting point of the left lane or the right lane of the exit lane Y in the exit road are the lane topological curves with the endpoints, and also include the The end of the entering lane X and the starting point of the exiting lane Y are the lane topological curves, wherein the exiting
  • the maximum curvature of each lane topological curve in the K' lane topological curves is not greater than a second preset threshold, and each lane topological curve is identical to the lane topological curve
  • the distance between the soft boundary constraint and the hard boundary constraint of the lane topological curve is not less than a third preset distance, and the distance between any two lane topological curves is not less than a fourth preset distance.
  • the method further includes: calculating an evaluation value of each lane topological curve in the K' lane topological curves, and the evaluation value is related to the curvature, curvature change rate, and slope of the lane topological curves.
  • Determining the target path from the topological curves of the K' lanes when the first entering lane of includes: when the vehicle is located in the first entering lane of the entering road, according to the K' lanes
  • the evaluation value of each lane topology curve in the topology curve determines the target path, and the target path includes the lane topology curve whose end point is the end of the first driving-in lane among the K' lane topology curves with the highest evaluation value The topological curve of the lane.
  • this method combined with global navigation information, macro traffic flow and other information, provides a global vision, conducts navigation recommendation evaluation on the lane-level topology, provides a global vision for vehicles during driving, avoids high-risk lane topologies in advance, and improves the efficiency of self-vehicle traffic , to reduce the risk of self-driving.
  • the intersection includes at least one of an intersection, a roundabout, an intersection in a waiting area, a small S-curve, an elevated entrance and exit, a multi-lane road section without lane markings, a continuous turning intersection, and a narrow road U-turn intersection.
  • the present application provides a method for generating a map based on an intersection, including: generating M lane topological curves according to the intersection, the entry road of the intersection, the obstacles in the exit road of the intersection, and the lane lines , the lane topological curve is a curve with the end of the entering lane of the entering road and the starting point of the exiting lane of the exiting road as endpoints; performing rationality detection processing on the M lane topological curves , to obtain K' topological curves of lanes, wherein K' is not greater than M; a map of the intersection is generated according to the topological curves of K' lanes in the intersection.
  • the lane topological curves are generated based on the intersection, the entry road of the intersection, the obstacles in the exit road of the intersection, and the lane lines, and then K' lane topological curves are obtained through rationality detection processing, and then the intersection is generated The K' lane topological curves.
  • the map generation of this solution based on the actual scene, generates a complete and reasonable lane topology curve that is more in line with human driving habits and is more reasonable. It does not require manual intervention and has good generalization and human-like characteristics. It can provide corresponding navigation for vehicles during automatic driving. Guide information to effectively improve the trajectory and efficiency of vehicles passing through the intersection.
  • the generating M lane topological curves according to the intersection, the entry road of the intersection, the obstacles in the exit road of the intersection, and the lane line includes: according to the intersection, the Obstacles in the entry road of the intersection, the exit road of the intersection, and the lane lines that cannot be crossed to obtain the hard boundary constraints of the lane topology curve; according to the intersection, the entry road of the intersection, and the exit road of the intersection can cross the lane line to obtain the soft boundary constraint of the lane topological curve; obtain K virtual boundary constraints of the lane topological curve according to the hard boundary constraint of the lane topological curve and the soft boundary constraint of the lane topological curve; according to the hard boundary constraint of the lane topological curve
  • the constraints, the lane topological curve soft boundary constraints and the K lane topological curve virtual boundary constraints generate the M lane topological curves.
  • This method obtains soft and hard boundary constraints and virtual boundary constraints based on high-precision maps and real-time perception of road obstacles and lane lines, which ensures the human-like nature of virtual lane trajectories and the universality of vehicle types, and improves the ability to generate lane topology curves. reliability.
  • the K lane topological curve virtual boundary constraints correspond to the K lane topologies
  • any lane topology curve virtual boundary constraint A in the K lane topological curve virtual boundary constraints is obtained by placing the leftmost lane of the intersection
  • the hard boundary constraint and/or soft boundary constraint on the left side of the topology is shifted to the right by a first preset distance
  • the hard boundary constraint and/or soft boundary constraint on the rightmost lane topology right side of the intersection is shifted to the left by a first distance
  • the preset distance is obtained, wherein the first preset distance is determined according to the lane order of the lane topology A', or the first preset distance is determined according to the vehicle preset passing width, lane width determined by at least one item of the lane topology A' and the lane position order of the lane topology A', the lane topology curve virtual boundary constraint A corresponds to the lane topology A'
  • the K lane topologies include the leftmost side of the intersection lane topology and the
  • the scheme considers the interference effects of other traffic trajectories, the indirect constraints of soft and hard boundaries on the gradual weakening of lanes in the same direction, etc., and generates virtual boundary constraints to ensure the human-like nature of virtual lane trajectories and the universality of vehicle types.
  • the generating the M lane topological curves according to the lane topological curve hard boundary constraints, the lane topological curve soft boundary constraints and the K lane topological curve virtual boundary constraints includes : Angle sampling is performed on the end of each incoming lane in the incoming road to obtain at least one starting point pose vector in the incoming road, and the angle of each outgoing lane in the outgoing road is respectively Performing angle sampling at the starting point to obtain at least one end point pose vector in the exiting road; performing curve sampling on the at least one starting point pose vector and the at least one end point pose vector to obtain the entering road and A plurality of curves between the outgoing roads; performing screening processing on the plurality of curves according to the hard boundary constraints of the lane topological curves, the soft boundary constraints of the lane topological curves and the virtual boundary constraints of the K lane topological curves , to obtain the topological curves of the M lanes.
  • the screening process may be, for example, to first screen curves satisfying the above constraints, and then to screen until there is at most one optimal curve between each incoming lane and each outgoing lane.
  • each curve is adaptively adjusted, and then the M lane topological curves are obtained.
  • This processing manner is only an example, and it may also be other manners, which are not specifically limited in this solution.
  • This scheme performs angle sampling based on the end of the entering lane and the starting point of exiting the lane, and then generates multiple lane topological curves. Based on the soft and hard boundary constraints and virtual boundary constraints obtained above, the multiple curves are screened, and then Get M lane topological curves. By adopting this method, the human-likeness, flexibility and vehicle type universality of the trajectory of the virtual lane are guaranteed, and the traffic conflict with other lanes is reduced.
  • the performing curve sampling on the at least one start point pose vector and the at least one end point pose vector to obtain a plurality of curves between the entry road and the exit road includes: A plurality of control points are generated between the end of each incoming lane of the incoming road and the starting point of each outgoing lane of the outgoing road; according to the at least one starting point pose vector, the at least A terminal pose vector and the plurality of control points generate the plurality of smooth curves.
  • this scheme not only performs angle sampling based on the end of the entering lane and the starting point of exiting the lane, but also based on control point sampling, which makes the number of generated curves more and improves the flexibility of curve generation.
  • the pose vector of at least one starting point in the entering road is obtained by extending the end of each entering lane by a second preset distance and performing sampling.
  • the sampling points of the beginning and end attitudes of the virtual intersection side trajectory are reasonably extended to the outside, so as to improve the quality of the generated trajectory, enhance the human-like nature of the trajectory, and avoid the trajectory caused by high-precision map drawing It is unreasonable, and it also avoids wrong screening due to unreasonable trajectory generation in topology screening, and ensures the completeness of road topology analysis.
  • performing rationality detection processing on the M lane topological curves to obtain K' lane topological curves includes: according to the direction vector of the entering road, the direction vector of the exiting road
  • the direction vector obtains the projection line between the entering road and the exiting road
  • the projection line is the straight line where the bisector of the angle obtained by the intersection of the direction vector of the entering road and the direction vector of the exiting road is located , or, the projection line is a straight line perpendicular to the direction vector of the exiting road and passing through the starting point of the exiting road
  • the alignment coefficient between each exit lane wherein, the alignment coefficient between each entry lane and each exit lane is the ratio between the first parameter and the second parameter, and the first
  • the parameter is the overlapping length between two line segments obtained by extending the lane edge of each entering lane and the lane edge of each exiting lane respectively to the projection line
  • the second parameter is the The lane sideline of each entering lane and the lane sideline of each exiting lane are respectively
  • the K' lane topological curves include the end of the leftmost inbound lane of the inbound road and the starting point of the leftmost outbound lane of the outbound road.
  • the curve whose starting point is the end point also includes the curve whose end points are respectively the end of the rightmost inbound lane of the inbound road and the starting point of the rightmost outbound lane of the outbound road.
  • the K' lane topological curves include the end of the incoming lane X in the incoming road and the left lane of the outgoing lane Y in the outgoing road.
  • the lane topological curve with the starting point as the end point, and the lane topological curve with the end point of the entering lane X and the starting point of the right lane of the exiting lane Y as the end point also includes the The end point and the starting point of the exit lane Y are lane topological curves, wherein there are lanes on the left and right sides of the exit lane Y; or, the K' lane topological curves include the The end point of the entry lane X in the entry road and the starting point of the left lane or the right lane of the exit lane Y in the exit road are the lane topological curves with the endpoints, and also include the The end of the entering lane X and the starting point of the exiting lane Y are the lane topological curves, wherein the exiting
  • the maximum curvature of each lane topological curve in the K' lane topological curves is not greater than a second preset threshold, and each lane topological curve is identical to the lane topological curve
  • the distance between the soft boundary constraint and the hard boundary constraint of the lane topological curve is not less than a third preset distance, and the distance between any two lane topological curves is not less than a fourth preset distance.
  • the intersection includes at least one of an intersection, a roundabout, an intersection in an area to be turned, a small S-curve, an elevated entrance, a multi-lane road section without lane markings, a continuous turning intersection, and a narrow road U-turn intersection.
  • the present application provides a device for guiding vehicles, including: a curve generation module, configured to generate curves according to obstacles in the intersection, the entry road of the intersection, the exit road of the intersection, and lane lines M lane topological curves, the lane topological curves are curves whose endpoints are the end of the entering lane of the entering road and the starting point of the exiting lane of the exiting road; the detection processing module is used to The M lane topological curves are subjected to rationality detection processing to obtain K' lane topological curves, wherein K' is not greater than M; the determination module is used to determine from the Determine the target path in the topological curves of K' lanes.
  • a curve generation module configured to generate curves according to obstacles in the intersection, the entry road of the intersection, the exit road of the intersection, and lane lines M lane topological curves, the lane topological curves are curves whose endpoints are the end of the entering lane of the entering road and the starting point of the exiting lane of
  • the curve generating module is configured to: obtain the hard boundary constraint of the lane topological curve according to the intersection, the entry road of the intersection, the obstacle in the exit road of the intersection, and the non-crossing lane line; , the entry road of the intersection, the crossable lane line in the exit road of the intersection to obtain the soft boundary constraint of the lane topology curve; obtain K according to the hard boundary constraint of the lane topology curve and the soft boundary constraint of the lane topology curve virtual boundary constraints of lane topological curves; generating the M lane topological curves according to the lane topological curve hard boundary constraints, the lane topological curve soft boundary constraints and the K lane topological curve virtual boundary constraints.
  • the K lane topological curve virtual boundary constraints correspond to the K lane topologies
  • any lane topology curve virtual boundary constraint A in the K lane topological curve virtual boundary constraints is obtained by placing the leftmost The hard boundary constraint and/or soft boundary constraint on the left side of the side lane topology is translated to the right by a first preset distance, and the hard boundary constraint and/or soft boundary constraint on the right side of the rightmost lane topology of the intersection is translated to the left
  • the first preset distance is obtained, wherein the first preset distance is determined according to the lane sequence of the lane topology A', or the first preset distance is determined according to the vehicle preset passing width, lane determined by at least one of the width and the lane order of the lane topology A', the lane topology curve virtual boundary constraint A corresponds to the lane topology A', and the K lane topologies include the most The left lane topology and the rightmost lane topology.
  • the curve generating module is further configured to: separately perform angle sampling on the end of each lane in the entering road to obtain at least one starting point pose vector in the entering road, and respectively Angle sampling is performed on the starting point of each exiting lane in the exiting road to obtain at least one end point pose vector in the exiting road; the at least one starting point pose vector and the at least one end point pose vector Vector curve sampling to obtain multiple curves between the incoming road and the outgoing road; according to the hard boundary constraints of the lane topological curves, the soft boundary constraints of the lane topological curves, and the K lane topological curves The virtual boundary constraints screen the multiple curves to obtain the M lane topological curves.
  • the curve generating module is further configured to: generate a plurality of control points between the end of each incoming lane of the incoming road and the starting point of each outgoing lane of the outgoing road ; Generate the plurality of smooth curves according to the at least one start point pose vector, the at least one end point pose vector, and the plurality of control points.
  • At least one starting point pose vector of the entering road is obtained by extending the end of each entering lane by a second preset distance and performing sampling.
  • the detection processing module is configured to: obtain a projection line between the entry road and the exit road according to the direction vector of the entry road and the direction vector of the exit road, and the projection line is the The straight line where the bisector of the angle obtained by the intersection of the direction vector of the entering road and the direction vector of the exiting road is located, or the projection line is perpendicular to the direction vector of the exiting road and passes through the exiting road A straight line at the starting point of ; calculate the alignment coefficient between each entry lane of the entry road and each exit lane of the exit road, wherein each entry lane and each The alignment coefficient between the exiting lanes is the ratio between the first parameter and the second parameter, and the first parameter is the extension of the lane edge of each entering lane and the lane edge of each exiting lane respectively.
  • the overlapping length between the two line segments obtained from the projected line, the second parameter is that the lane sideline of each entering lane and the lane sideline of each exiting lane are respectively extended to the projected
  • K' lane topology is obtained Curves, wherein the K' lane topology curves include curves whose endpoints are the end of the incoming lane and the starting point of the outgoing lane whose alignment coefficient is greater than the first preset threshold.
  • the K' lane topology curves include curves whose endpoints are respectively the end of the leftmost inbound lane of the inbound road and the starting point of the leftmost outbound lane of the outbound road, It also includes curves whose endpoints are respectively the end of the rightmost inbound lane of the inbound road and the starting point of the rightmost outbound lane of the outbound road.
  • the K' lane topological curves include lane topological curves whose endpoints are the end of the incoming lane X in the incoming road and the starting point of the left lane of the outgoing lane Y in the outgoing road.
  • the lane topology curve with the end point of the entry lane X and the starting point of the right lane of the exit lane Y as the endpoint and also includes the end point of the entry lane X and the exit lane Y
  • the lane topological curve whose starting point is the end point, wherein, there are lanes on the left and right sides of the exit lane Y; or, the K' lane topological curves include the entry lane in the entry road
  • the end point of X and the starting point of the left lane or the right lane of the exit lane Y in the exit road are the lane topological curves, which also include the end of the entry lane X in the exit road,
  • the start point of the exit lane Y is a lane top
  • the maximum curvature of each lane topological curve in the K' lane topological curves is not greater than a second preset threshold, and each lane topological curve is bounded by the soft boundary of the lane topological curve, the The distance between the hard boundary constraints of the lane topological curves is not less than a third preset distance, and the distance between any two lane topological curves is not less than a fourth preset distance.
  • the device further includes an evaluation module, configured to: calculate the evaluation value of each lane topological curve in the K' lane topological curves, and the evaluation value is related to the curvature, curvature change rate, and oblique crossing of the lane topological curves.
  • the number of lanes is related to at least one of lane intersection information, traffic rule information, estimated value of traffic flow, and drivable distance of the lane corresponding to the lane topology curve;
  • the determination module is configured to: when the vehicle is located at the When entering the road, the target path is determined according to the evaluation value of each lane topological curve in the K' lane topological curves.
  • the intersection includes at least one of an intersection, a roundabout, an intersection in a waiting area, a small S-curve, an elevated entrance and exit, a multi-lane road section without lane markings, a continuous turning intersection, and a narrow road U-turn intersection.
  • the present application provides an intersection-based map generation device, including: a curve generation module, configured to generate information according to the intersection, the entry road of the intersection, the obstacles in the exit road of the intersection, and the lane line Generate M lane topological curves, the lane topological curves are curves whose endpoints are the end of the entry lane of the entry road and the starting point of the exit lane of the exit road; a detection processing module for The M lane topological curves are subjected to rationality detection processing to obtain K' lane topological curves, wherein K' is not greater than M; the map generation module is used to generate the K' lane topological curves according to the intersection. map of intersections.
  • the curve generating module is configured to: obtain the hard boundary constraint of the lane topological curve according to the intersection, the entry road of the intersection, the obstacle in the exit road of the intersection, and the non-crossing lane line; , the entry road of the intersection, the crossable lane line in the exit road of the intersection to obtain the soft boundary constraint of the lane topology curve; obtain K according to the hard boundary constraint of the lane topology curve and the soft boundary constraint of the lane topology curve virtual boundary constraints of lane topological curves; generating the M lane topological curves according to the lane topological curve hard boundary constraints, the lane topological curve soft boundary constraints and the K lane topological curve virtual boundary constraints.
  • the K lane topological curve virtual boundary constraints correspond to the K lane topologies
  • any lane topology curve virtual boundary constraint A in the K lane topological curve virtual boundary constraints is obtained by placing the leftmost The hard boundary constraint and/or soft boundary constraint on the left side of the side lane topology is translated to the right by a first preset distance, and the hard boundary constraint and/or soft boundary constraint on the right side of the rightmost lane topology of the intersection is translated to the left
  • the first preset distance is obtained, wherein the first preset distance is determined according to the lane sequence of the lane topology A', or the first preset distance is determined according to the vehicle preset passing width, lane determined by at least one of the width and the lane order of the lane topology A', the lane topology curve virtual boundary constraint A corresponds to the lane topology A', and the K lane topologies include the most The left lane topology and the rightmost lane topology.
  • the curve generating module is further configured to: separately perform angle sampling on the end of each lane in the entering road to obtain at least one starting point pose vector in the entering road, and respectively Angle sampling is performed on the starting point of each exiting lane in the exiting road to obtain at least one end point pose vector in the exiting road; the at least one starting point pose vector and the at least one end point pose vector Vector curve sampling to obtain multiple curves between the incoming road and the outgoing road; according to the hard boundary constraints of the lane topological curves, the soft boundary constraints of the lane topological curves, and the K lane topological curves The virtual boundary constraints screen the multiple curves to obtain the M lane topological curves.
  • the curve generating module is further configured to: generate a plurality of control points between the end of each incoming lane of the incoming road and the starting point of each outgoing lane of the outgoing road ; Generate the plurality of smooth curves according to the at least one start point pose vector, the at least one end point pose vector, and the plurality of control points.
  • At least one starting point pose vector of the entering road is obtained by extending the end of each entering lane by a second preset distance and performing sampling.
  • the detection processing module is configured to: obtain a projection line between the entry road and the exit road according to the direction vector of the entry road and the direction vector of the exit road, and the projection line is the The straight line where the bisector of the angle obtained by the intersection of the direction vector of the entering road and the direction vector of the exiting road is located, or the projection line is perpendicular to the direction vector of the exiting road and passes through the exiting road A straight line at the starting point of ; calculate the alignment coefficient between each entry lane of the entry road and each exit lane of the exit road, wherein each entry lane and each The alignment coefficient between the exiting lanes is the ratio between the first parameter and the second parameter, and the first parameter is the extension of the lane edge of each entering lane and the lane edge of each exiting lane respectively.
  • the overlapping length between the two line segments obtained from the projected line, the second parameter is that the lane sideline of each entering lane and the lane sideline of each exiting lane are respectively extended to the projected
  • K' lane topology is obtained Curves, wherein the K' lane topology curves include curves whose endpoints are the end of the incoming lane and the starting point of the outgoing lane whose alignment coefficient is greater than the first preset threshold.
  • the K' lane topology curves include curves whose endpoints are respectively the end of the leftmost inbound lane of the inbound road and the starting point of the leftmost outbound lane of the outbound road, It also includes curves whose endpoints are respectively the end of the rightmost inbound lane of the inbound road and the starting point of the rightmost outbound lane of the outbound road.
  • the K' lane topological curves include lane topological curves whose endpoints are the end of the incoming lane X in the incoming road and the starting point of the left lane of the outgoing lane Y in the outgoing road.
  • the lane topology curve with the end point of the entry lane X and the starting point of the right lane of the exit lane Y as the endpoint and also includes the end point of the entry lane X and the exit lane Y
  • the lane topological curve whose starting point is the end point, wherein, there are lanes on the left and right sides of the exit lane Y; or, the K' lane topological curves include the entry lane in the entry road
  • the end point of X and the starting point of the left lane or the right lane of the exit lane Y in the exit road are the lane topological curves, which also include the end of the entry lane X in the exit road,
  • the start point of the exit lane Y is a lane top
  • the maximum curvature of each of the K' lane topological curves is not greater than a second preset threshold, and each of the lane topological curves is constrained by the soft boundary of the lane topological curve, the The distance between the hard boundary constraints of the lane topological curves is not less than a third preset distance, and the distance between any two lane topological curves is not less than a fourth preset distance.
  • the intersection includes at least one of an intersection, a roundabout, an intersection in a waiting area, a small S-curve, an elevated entrance and exit, a multi-lane road section without lane markings, a continuous turning intersection, and a narrow road U-turn intersection.
  • the present application provides a device for guiding a vehicle, including a processor and a memory; wherein the memory is used to store program codes, and the processor is used to call the program codes to execute the first
  • the memory is used to store program codes
  • the processor is used to call the program codes to execute the first
  • the present application provides an intersection-based map generation device, including a processor and a memory; wherein the memory is used to store program codes, and the processor is used to call the program codes to execute the first
  • the memory is used to store program codes
  • the processor is used to call the program codes to execute the first
  • the present application provides a computer-readable storage medium, the computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement any method and/or provided in the first aspect Or any method provided in the second aspect.
  • the present application provides a computer program product that, when the computer program product is run on a computer, causes the computer to execute any of the methods provided in the first aspect and/or any of the methods provided in the second aspect. Methods.
  • the present application provides a chip system, the chip system is applied to electronic equipment; the chip system includes one or more interface circuits, and one or more processors; the interface circuit and the processing The devices are interconnected by wires; the interface circuit is used to receive signals from the memory of the electronic device and send the signals to the processor, the signals include computer instructions stored in the memory; when the processor When executing the computer instructions, the electronic device executes any method provided in the first aspect and/or any method provided in the second aspect.
  • the present application provides an intelligent driving vehicle, which is characterized in that it includes a travel system, a sensing system, a control system and a computer system, wherein the computer system is used to execute any of the functions provided in the first aspect. method and/or any method provided in the second aspect.
  • the device described in the third aspect, the device described in the fourth aspect, the device described in the fifth aspect, the device described in the sixth aspect, the computer storage medium described in the seventh aspect, or the The computer program product described in the eighth aspect, the chip system described in the ninth aspect, and the intelligent driving vehicle described in the tenth aspect are all used to execute any of the methods provided in the first aspect and any of the methods provided in the second aspect. Methods. Therefore, the beneficial effects that it can achieve can refer to the beneficial effects in the corresponding method, and will not be repeated here.
  • FIG. 1 is a schematic diagram of a system architecture for guiding vehicles provided by an embodiment of the present application
  • FIG. 2 is a schematic flowchart of a method for guiding a vehicle to travel provided by an embodiment of the present application
  • FIG. 3 is a schematic diagram of a lane topology provided by an embodiment of the present application.
  • FIG. 4 is a schematic diagram of a boundary constraint provided by an embodiment of the present application.
  • Fig. 5 is a schematic diagram of a curve generation method provided by an embodiment of the present application.
  • FIG. 6 is a schematic diagram of a method for solving an alignment coefficient provided in an embodiment of the present application.
  • Fig. 7 is a schematic diagram of a lane topology curve screening provided by an embodiment of the present application.
  • Fig. 8a is a schematic diagram of a scene of a special-shaped many-to-many straight intersection in the waiting area provided by the embodiment of the present application;
  • Fig. 8b is a schematic diagram of boundary constraints provided by the embodiment of the present application.
  • Fig. 8c is a schematic diagram of curve screening provided by the embodiment of the present application.
  • Figure 8d is a schematic diagram of the first overlapping length provided by the embodiment of the present application.
  • Fig. 8e is a schematic diagram of the second overlapping length provided by the embodiment of the present application.
  • Fig. 8f is a schematic diagram of the first application scenario provided by the embodiment of the present application.
  • Fig. 8g is a schematic diagram of the second application scenario provided by the embodiment of the present application.
  • Fig. 8h is a schematic diagram of a third application scenario provided by the embodiment of the present application.
  • Fig. 9a is a schematic diagram of a roundabout scene provided by an embodiment of the present application.
  • Fig. 9b is a schematic diagram of boundary constraints provided by the embodiment of the present application.
  • Fig. 9c is a schematic diagram of curve generation provided by the embodiment of the present application.
  • Fig. 9d is a schematic diagram of curve screening provided by the embodiment of the present application.
  • Fig. 10a is a schematic diagram of a scene of a left-turn intersection with waiting areas and traffic lights provided by the embodiment of the present application;
  • Fig. 10b is a schematic diagram of another left-turn intersection scene with waiting area and traffic lights provided by the embodiment of the present application;
  • Fig. 11a is a schematic diagram of a small S-curve scene provided by an embodiment of the present application.
  • Fig. 11b is a schematic diagram of another small S-curve scene provided by the embodiment of the present application.
  • FIG. 12 is a schematic diagram of a ramp scene provided by an embodiment of the present application.
  • FIG. 13 is a schematic diagram of a multi-lane road section without lane markings provided by an embodiment of the present application.
  • Fig. 14 is a schematic diagram of a continuous turning scene provided by an embodiment of the present application.
  • Fig. 15 is a schematic diagram of a U-turn scene in a narrow road provided by an embodiment of the present application.
  • Fig. 16 is a schematic structural diagram of a device for guiding a vehicle provided in an embodiment of the present application.
  • Fig. 17 is a schematic structural diagram of another device for guiding a vehicle provided in an embodiment of the present application.
  • FIG. 1 it is a schematic diagram of a system architecture for guiding a vehicle according to an embodiment of the present application.
  • the system may include: a lane topology generation module, a boundary constraint generation module, a topology curve generation module, a topology rationality screening module and a real-time decision module.
  • the lane topology generation module is used to generate lane-level fully connected topology in combination with the road structure of the current scene, laying the foundation for building a complete road topology space.
  • the boundary constraint generation module is used to generate soft boundary constraints, hard boundary constraints and virtual boundary constraints for each lane topology in the lane-level fully connected topology based on high-precision maps and sensor perception data.
  • the topology curve generating module is used to generate a smooth human-like lane topology curve satisfying the above-mentioned boundary constraints for the above-mentioned lane topology.
  • the topology rationality screening module is used to screen the rationality of the lane-level fully connected topology and its curves based on various human-like screening principles, delete unreasonable lane topologies, and obtain a complete and reasonable set of lane topology curves to complete the current Road topology analysis of the scene.
  • the real-time decision-making module is used to select the optimal target path at the current moment based on the above-mentioned road topology analysis, dynamic and static traffic environment, and the real-time status of the vehicle.
  • the above system may also include a navigation recommendation evaluation module.
  • the navigation recommendation evaluation module introduces a lane topology recommendation function to evaluate the navigation priority of each lane topology curve in the above lane topology curve set.
  • the real-time decision-making module can also select the optimal target route based on the navigation priority evaluation.
  • the above system is just an example, and the system may only include a boundary constraint generation module, a topology curve generation module, a topology rationality screening module, and a real-time decision module, etc., which are not specifically limited in this solution.
  • This solution can be applied to self-driving vehicles driving on open roads.
  • the driving range includes road scenes where there are no actual lane lines or multiple reasonable driving trajectories, it is necessary to perform reasonable road topology analysis and provide corresponding topology navigation guidance information.
  • vehicle intent prediction trajectory (curve) prediction, lane decision-making and motion planning.
  • trajectory (curve) prediction lane decision-making and motion planning.
  • the above-mentioned road scenarios include but are not limited to intersections, roundabouts, intersections in waiting areas, small S-curves, elevated entrances and exits, multi-lane road sections without lane markings, continuous turning intersections, narrow road U-turn intersections, etc. Of course, it may also be other scenarios, which are not specifically limited in this solution.
  • This embodiment can be executed by a vehicle-mounted device (such as a car machine), and it can also be executed by a terminal device such as a mobile phone or a computer. This plan does not specifically limit this.
  • the method for guiding a vehicle can be executed locally or by the cloud.
  • the cloud may be implemented by a server, and the server may be a virtual server, a physical server, etc., or other devices, which are not specifically limited in this solution.
  • FIG. 2 it is a schematic flowchart of a method for guiding a vehicle to travel provided by an embodiment of the present application. As shown in Figure 2, the method includes steps 201-203, specifically as follows:
  • the M lane topological curves may be generated according to the high-precision map and the environment perception information acquired by the sensor, the intersection, the entry road, the obstacles in the exit road, and lane lines.
  • the lane topological curve hard boundary constraint, the lane topological curve soft boundary constraint and the lane topological curve virtual boundary constraint are obtained based on the obstacles and lane lines, and then the M lane topological curves are generated based on the above constraints.
  • step 201 may include steps 2011-2014:
  • the hard boundary constraint of the lane topology curve can be understood as the boundary where vehicles cannot drive.
  • the static obstacles near the intersection scene and the lane lines that cannot be crossed are obtained, that is, the hard boundary constraints of the lane topology curve are obtained.
  • the above-mentioned static obstacles include curbs, safety islands, green belts, etc. on the road.
  • Lane lines that cannot be crossed such as solid lane lines, diversion lines, etc.
  • the soft boundary constraint of the lane topology curve can be understood as a boundary constraint that the ego vehicle is allowed to cross, but the ego vehicle is best not to cross in the traffic rules.
  • the soft boundary constraints are obtained by obtaining the actual marking lines on the ground in the intersection scene.
  • the actual markings on the ground are, for example, lane lines in the waiting area, lane lines that can be crossed, and the like.
  • the K lane topological curve virtual boundary constraints correspond to the K lane topologies.
  • K lane topologies of the intersection are obtained according to the topological structure of each lane in the intersection, the entry road of the intersection, and the exit road of the intersection.
  • the road topology in the current scene specifically, it includes intersections, entry roads and entry lanes in entry roads, exit roads and exit lanes in exit roads, and roads without markings Elements and their topological relationships; then based on the traffic rules related to the current scene, all possible K lane-level fully connected topologies in this scene can be generated according to all passable K1 entry lanes and K2 exit lanes, where, K ⁇ K1*K2.
  • FIG. 3 it is a schematic diagram of a lane topology provided in the embodiment of the present application.
  • the scene includes 3 entry lanes entering the road and 4 exit lanes exiting the road.
  • 12 lane-level fully connected topologies that is, 12 virtual lanes.
  • the line segments of the 12 lane-level fully connected topologies shown in the figure are only used to represent the preceding and following relationships between lanes, and do not represent the final curve (trajectory shape) of the virtual lane.
  • traffic rules may have an impact on the lane topology full connection relationship and its quantity in the current scenario.
  • traffic rules may have an impact on the lane topology full connection relationship and its quantity in the current scenario.
  • special lanes such as bus lanes and tidal lanes
  • K1 entering lanes and K2 The number K of topologically fully connected lanes may be less than K1*K2.
  • any lane topology curve virtual boundary constraint A among the above-mentioned K lane topological curve virtual boundary constraints is obtained by shifting the leftmost hard boundary constraint and/or soft boundary constraint on the left side of the leftmost lane topology of the intersection to the right.
  • a preset distance obtained by shifting the hard boundary constraint and/or soft boundary constraint on the right side of the rightmost lane topology of the intersection to the left by a first preset distance.
  • the leftmost lane topology of the intersection is the left-incoming lane versus the left-outbound lane.
  • the rightmost lane topology of the intersection is the right-incoming lane versus the right-exiting lane.
  • the above-mentioned first preset distance may be determined according to the lane sequence of the lane topology A'.
  • the above-mentioned first preset distance may be determined according to at least one of the preset passing width of the vehicle, the lane width, and the lane order of the lane topology A'.
  • the preset passing width of the vehicle can be understood as the width required for the vehicle to pass safely.
  • the lane order above can be understood as the order in which the lanes are located.
  • the lane order can be set based on preset rules, for example, take the minimum value of the entering lane number and the exiting lane number, and then subtract 1 from the minimum value to obtain the above lane order.
  • the lane order of the first left-incoming lane to the left-first exiting lane is 0;
  • the order of the lane from the right of the entering lane to the third exiting lane on the right is 1.
  • the aforementioned preset distance may be determined according to the lane order of each lane topology. For example, the greater the order of the lane, the greater the corresponding preset distance.
  • the aforementioned preset distance may also be determined according to at least one of the safe passage width of the vehicle, the lane width, and the lane sequence of each lane topology.
  • n lane represents the lane sequence of the current lane topology
  • W lane represents the width of the lane
  • W vehicle represents the preset (safe) passing width of the vehicle
  • the translation direction corresponding to the above translation distance is the opposite direction of the collision depth between the hard and soft boundary and the reference topological curve.
  • the above-mentioned reference topological trajectory includes a smooth curve S1 corresponding to the first left entering lane to the first left exiting lane, a smooth curve S2 corresponding to the second left entering lane to the second left exiting lane, and the like.
  • the shaded area in Figure 4 indicates curbs, flower beds, safety islands, etc., and the shaded area shows the hard boundary constraints of each lane.
  • the lane line L1 and the lane line L2 of the lane to be turned are the soft boundary constraints of each lane.
  • the direction pointed by the arrow t corresponding to the reference topology curve S1 is the direction of the collision depth
  • the dotted line areas U1, U2, and U3 and the dotted line L3 are the hard boundary constraints (that is, the shadow area) and the soft boundary on the left side of the leftmost lane topology at the intersection Constraint (L2) is the virtual boundary constraint of the next adjacent lane obtained by translating a certain distance along the opposite direction of the collision depth.
  • step 2014 may include steps 20141-20143, specifically as follows:
  • a sampling algorithm based on a Bezier curve an optimization algorithm based on a Spiral curve, etc. may be used to generate the above lane topology curve.
  • an optimization algorithm based on a spiral curve is taken as an example for illustration.
  • the angle sampling based on the end of the entering lane and the starting point of exiting the lane can obtain the vectors of different attitude angles starting from the end of entering the lane and the starting point of exiting the lane respectively.
  • the above starting point pose vector can be understood as a vector located at the end of the driving lane and having different pose angles.
  • vectors in multiple different directions at the end A of the entering lane That is, the above starting point pose vector.
  • the above-mentioned terminal pose vector can be understood as a vector located at the starting point of the exit lane and having different pose angles.
  • vectors in multiple different directions at the starting point B of the exit lane That is, the above-mentioned terminal pose vector.
  • any start point pose vector and any end point pose vector are combined, and multiple smooth curves can be obtained based on the above multiple combinations.
  • distance sampling is performed on the line connecting the end of the entering lane and the starting point of the exiting lane to generate multiple sets of intermediate control points, as shown by point P i in FIG. 5 .
  • the distance sampling may be to generate several control points at intervals of preset distances, and of course other methods may also be used, which is not specifically limited in this solution.
  • the curve evaluation function is constructed by considering factors such as the curvature of the generated curve, the rate of curvature change, the collision cost to the soft and hard boundary, the passage space, and the length of the curve (traffic efficiency).
  • the optimal curve that satisfies the safety boundary constraints and vehicle performance constraints is selected from the above smooth curves as the optimal curve from the end of the entering lane to the starting point of the exiting lane.
  • the curve S3 between the end of the entering lane and the starting point of leaving the lane is an optimal curve. Among them, the shorter the curve between the same starting and ending points, the shorter the traveling distance of the traffic traveling along the corresponding curve, and the higher the traffic efficiency.
  • the collision control point is added according to the collision position and collision depth to locally adjust the collision position The shape of the nearby trajectory, so as to obtain a safe and non-collision lane topology curve S4 with smooth curvature.
  • the principle of topological projection screening can be understood as, by obtaining the alignment coefficient (overlap coefficient) between the entering lane and the exiting lane, if the alignment coefficient is not less than the preset threshold, it means that the distance between the entering lane and the exiting lane is With the alignment feature, the lane topology and its curves are retained; if the alignment coefficient is less than the preset threshold, it means that there is no alignment feature between the entering lane and the exiting lane, and the lane topology and its curves are deleted.
  • the direction vector of the entering lane and the direction vector of exiting the lane are extended back and forth. If the intersection point is between the end of the entering lane and the starting point of the exiting lane at the intersection, the angle passing through the intersection is bisected
  • the straight line where the line is located is used as a projection line, as shown by projection line L3 in FIG. 6 .
  • the direction vector of the entering lane may be understood as a direction vector of entering the road, that is, a vector parallel to the entering lane and consistent with a driving direction corresponding to the entering lane.
  • intersection point is not located between the end of the entry lane and the starting point of the exit lane at the intersection, a straight line perpendicular to the direction vector of the exit road and passing through the starting point of the exit road is used as the projection line, that is, in Fig. 6 Shown by straight line L4.
  • the lane lines of the entering lane and the exiting lane are both extended to the above projection line L3
  • the overlapping lengths of the left and right lane edges of the entering lane and the left and right lane edges of the exiting lane on the projection line are calculated, and the entering lane
  • the left and right lane sidelines and the left and right lane sidelines of the exit lane are respectively extended to the length of the shortest line segment among the two line segments obtained by the projection line.
  • the two lane sidelines of the straight driving into the left one of the lane are extended to the projection line L3, and the line segment C1C2 is obtained; Extend the sidelines of the two lanes from the second left of the lane to the projection line L3 to obtain the line segment D1D2; obtain the overlapping length L between the line segment C1C2 and the line segment D1D2, and the length L min of the shortest line segment between the line segment C1C2 and the line segment D1D2, then the above overlap
  • the ratio between the length L and the length of the shortest line segment L min is the alignment coefficient between the first lane on the left and the second lane on the left.
  • the alignment coefficient between other lane topologies can be calculated similarly.
  • the lane topology is further screened.
  • Lane topology that complies with traffic flow rules and removes lanes that do not conform to the current direction of traffic from all lanes entering and exiting the road. For example, in a straight intersection, delete the left turn, U-turn, and right turn into the lane before the intersection, and only keep the lane that can go straight.
  • the collision control point is added according to the collision location and the collision depth to locally adjust the trajectory shape near the collision location and perform local curve adjustment.
  • all lane topological curves can pass at the same time. If the distance between any two lane topological curves is too close, for example, less than the preset adjacent lane interference distance, Then the state of the topological curve of the lane is that there is traffic interference, and the interference distance of the adjacent lane is generally smaller than the width of the lane and slightly larger than the width of the vehicle body.
  • the lane curve spacing is adjusted locally, so that the spacing distribution between adjacent lane curves is more reasonable, and there is no traffic flow interference phenomenon.
  • the topological curve of the straight lane and the topological curve of the left-turning lane in this scene are two lane topological curves that can pass simultaneously between different road directions and different lanes, but these two If the topological curve of the lane is too close near the exit of the turning area, the distance between the straight-going vehicle and the left-turning traffic flow will be too close, interfering with each other and causing lateral extrusion.
  • the distance between the two lane topological curves needs to be adjusted locally , specifically, the topology of the through lane and the topology of the same-going lane on the right can be properly shifted to the right, so as to ensure that there is no traffic interference with the topology of the left-turning lane.
  • Left Align Supplementary Principle Checks if the topology between the left lane of the entering road and the left lane of the exiting road is removed by the above filtering principle. If the topology between the left lane of the entering road and the left lane of the exiting road is deleted by the above filtering principle, then the topology is added back, and correspondingly, the corresponding curve is also added back.
  • Right Align Supplementary Principle Check if the topology between the right lane of the incoming road and the right lane of the outgoing road is removed by the above filtering principle. If the topology between the right lane of the entering road and the right lane of the exiting road is deleted by the above filtering principle, then the topology will be supplemented, and correspondingly, the corresponding curve will also be supplemented.
  • exit topology when exit road topology exists, it should be ensured that all exit lanes in the exit road have lane topology curves; if a lane does not have one, a new lane topology should be added nearby to ensure the completeness of exit topology sex.
  • Adjacent lane supplementary principle If there is no lane topological curve between an outgoing lane and incoming lane x, but there are lane topological curves between the left and right lanes of the outgoing lane and the incoming lane x, Then a new lane topology is added to the exit lane nearby to ensure the rationality of the adjacent lane topology. Wherein, if the exit lane is the left lane of the exit road, only the right lane thereof is considered; if the exit lane is the right exit lane of the exit road, only the left lane is considered.
  • the above-mentioned left and right lanes may be the lanes adjacent to the lane, or the lanes at intervals, which is not limited in this scheme.
  • the lane topological curve is determined from the lane topological curves corresponding to the first entry lane, which is the target path.
  • the topological curve of the lane can be arbitrarily selected as the target path from the topological curves of the K' lanes, and the optimal lane topological curve can also be selected as the target path based on real-time traffic conditions.
  • the optimal curve can be further determined from the K' lane topology curves based on traffic conditions and the like. For example, when entering an intersection from the optimal lane topology curve S, other optimal curves are determined in real time due to being occupied by other vehicles.
  • the method may also include:
  • the evaluation value is related to the curvature of the lane topological curve, the rate of curvature change, the number of diagonally crossing lanes, and the lane of the lane corresponding to the lane topological curve It is related to at least one of intersection information, traffic rule information, estimated value of traffic flow, and travelable distance.
  • the lane topology recommendation evaluation function is designed to evaluate the navigation priority of each lane topological curve in the above-mentioned lane topological curve set to indicate that there is no other dynamic
  • the navigation recommendation priority of each curve in the same cluster lane topology (multiple lane topological curves entering the scene from the same incoming lane) in the traffic scene of traffic flow interference is used to select the comprehensive optimal lane topology curve.
  • the lane topology recommendation evaluation function can be expressed as:
  • C1, C2, C3, C4, C5 and C6 denote navigation cost, topology cost, smoothness cost, traffic intersection cost, traffic rule cost and traffic efficiency cost respectively, and w1, w2, w3, w4, w5 and w6 are coefficient.
  • the above navigation cost is based on the global navigation information to evaluate the target accessibility of the lane topology in the current scene. If the lane-level path planning where the lane topology is located is longer in the direction of reaching the specified end point of the automatic driving task, the lower the navigation cost C1 of the lane topology in the same cluster of lane topologies, w1 is the navigation cost in the total cost account for weight.
  • the above topological cost is to evaluate the human-likeness of the lane topology in the current scene according to the physical positional relationship between the front and rear entering lanes and exiting lanes connected by the lane topology. If the front and rear in-lanes and out-of-lanes of the topological connection of this lane cross multiple lanes to the left or right for oblique crossing, the topological trajectory of the lane will not be human-like, and the risk of grabbing the lane with other vehicles will be increased. Therefore, this The fewer the number of inclined lanes in the lane topology, the lower the topological cost C2, where w2 is the weight of the topological cost in the total cost.
  • the above smoothness cost is that the topological curve of the lane is composed of the trajectory of the entering lane, the trajectory of the virtual lane, and the trajectory of the exiting lane in the current scene.
  • the curvature and curvature change rate of the lane topological curve are evaluated. The smaller the curvature and the curvature change rate , indicating that the smoother the lane topology curve is, the lower the smoothness cost C3 is, and w3 is the weight of the smoothness cost in the total cost.
  • the above-mentioned traffic flow intersection cost is based on the topological connection relationship of all lanes of the outgoing lane connected by this lane topology, and evaluates the attribute of intersection of this lane topology with other traffic flows in the current scene. If the exit lane of this lane topology also belongs to other lane topologies at the same time, for example: the exit lane of this through lane topology is also the exit lane of other left-turn, U-turn, and right-turn lane topologies, it means that this lane topology will be consistent with Vehicles from other directions will merge and have a higher risk of lane grabbing or lateral extrusion, so the vehicle interaction relationship will be more complex, the traffic efficiency will be lower, and the lane intersection cost C4 will be higher, where w4 is the share of the traffic flow intersection cost in the total cost Weights.
  • the traffic rule cost is the cost that represents the preference priority of the traffic rule to the lane topology in the current scene.
  • the lane topology with a higher traffic rule preference priority has a lower traffic rule cost C5, and w5 is the traffic rule cost in the total cost accounted for weight.
  • the lane topology with a higher proportion of traffic flow in the macroscopic traffic flow means that the closer to the choice of human drivers, the higher the traffic efficiency, and the lower the traffic efficiency cost C6, where w6 is the proportion of the traffic efficiency cost in the total cost Weights.
  • the total cost evaluation of each lane curve in the above-mentioned lane topological curve set can represent the navigation recommendation priority of each curve in the same cluster of lane topology related to the same incoming lane, and is used to select the comprehensive optimal Lane topology.
  • the lane topology with straighter topology, smoother trajectory, no lane intersection, and higher traffic efficiency is the optimal recommended lane topology curve for navigation.
  • the lane topological curve with the best evaluation value is selected as the target path.
  • the vehicle can select the optimal lane topology curve for navigation recommendation by relying on the global vision of the navigation recommendation evaluation, so that Change lanes in advance to avoid the high-risk lane topology of traffic conflicts such as lane crossing, lateral extrusion of other vehicles, and multi-lane merging.
  • the optimal recommended topological alignment is a lane topology curve that is straighter and has a longer drivable distance.
  • the topological curve of the lane after entering the waiting area is optimally recommended.
  • the lane topology curve along the lane line is optimally recommended.
  • the optimal recommendation is to satisfy kinematics and the innermost lane topology curve.
  • the navigation recommends the optimal lane topology curve, and the vehicle can synthesize each lane Topology navigation recommendation evaluation and real-time dynamic risk, select the optimal lane topology in the current scene, such as the suboptimal lane topology curve that may be the navigation recommendation, etc.
  • the optimal recommendation is the lane topology curve with less lateral traffic interference and longer driving distance.
  • the optimal recommendation is the topological curve of the lane that cuts the curve.
  • the optimal recommendation is the topological curve of the inscribed lane that does not enter the waiting area.
  • the optimal recommendation is to satisfy kinematics and unoccupied inner lane topology curves.
  • M lane topological curves are generated according to the intersection, the entry road of the intersection, the obstacles in the exit road of the intersection, and the lane lines, and the rationality detection and processing of the M lane topological curves are performed to obtain K' lane topological curves, and then determine a target path from the K' lane topological curves to guide the vehicle to travel.
  • this method by generating a complete and reasonable lane topology curve and providing driving guidance information for the vehicle, the trajectory of the vehicle passing through the intersection is effectively improved and the traffic efficiency is improved.
  • FIG. 8a it is a scene of a special-shaped many-to-many straight intersection in the waiting area provided by the embodiment of the present application.
  • the scene includes waiting areas, bus lanes, parking lanes, unaligned entry and exit lanes, and special-shaped obtuse-angle intersections.
  • the inbound road shown in Figure 8a includes 1 left-turn lane, 1 right-turn lane, 1 bus through lane, and 2 ordinary through lanes; the outbound road includes 1 parking lane and 3 ordinary lanes. Drive straight out of the driveway.
  • the lane-level fully connected topology includes all alternative 6 lane topologies: L i1 ⁇ L o1 , L i1 ⁇ L o2 , L i1 ⁇ L o3 , L i2 ⁇ L o1 , L i2 ⁇ L o2 , L i2 ⁇ L o3 .
  • Environment perception obstacle information and lane line information obtained based on high-precision maps and sensors such as obstacles U4, U5, U6, U7, U8, U9, and lane markings R1 and R3 shown in Figure 8a.
  • the hard boundary constraints include the curbs, safety islands, green belts, etc. shown in the above obstacles U4, U5, U6, U7, U8, and U9, as well as the boundary lines of intersections, etc.; the soft boundary constraints are defined by the two left turns shown in Figure 8a
  • the lane markings R1 and R3 in the waiting area are formed.
  • the above obstacle areas are distributed on the left and right sides of the 6 lane topologies generated above, and for each lane topology, the distribution of the left and right obstacle areas is exactly the same, therefore, the hard boundary constraints of each lane topology are the same. Likewise, the distribution of soft boundary constraints is exactly the same for each lane topology, so the soft boundary constraints for each lane topology are the same.
  • the virtual boundary constraints consider the interference of traffic trajectories in adjacent lanes, and reflect the indirect constraints of soft and hard boundaries on the same direction lanes. Therefore, the virtual boundary constraints of each lane topology are different. For example, taking the lane topology L i2 ⁇ L o2 as an example, since the traffic trajectory in the lane topology L i1 ⁇ L o1 is affected by its soft and hard boundaries, the soft and hard boundaries of the adjacent lanes will also affect the lane
  • the topology L i2 ⁇ L o2 produces virtual boundary constraints.
  • the reference topological curve of the left nearest neighbor lane L i1 ⁇ L o1 is shown as curve S5 in Fig. 8b, and its collision depth vector is shown.
  • the left virtual boundary constraint of the lane topology L i2 ⁇ L o2 is shown in the areas U40, U50, U60 and dashed lines S6 and S7 in Figure 8b, along the collision depth vector
  • the translation distance is
  • the left-aligned lane order lane sequence And d 1 >W vehicle , and d 1 ⁇ W lane that is, the translation distance is larger than the width of the vehicle body and smaller than the width of the lane.
  • the lane topology L i1 ⁇ L o2 is affected by the interference of the traffic flow trajectory of the nearest neighbor lane on the right, resulting in virtual boundary constraints.
  • the reference topological curve of the right nearest neighbor lane L i2 ⁇ L o3 is shown as the curve S8 in Fig. 8b, and its collision depth vector is shown.
  • Right-aligned lane order along the collision depth vector The reverse translation distance is d 2 , and d 2 >W vehicle , and d 2 ⁇ W lane , so the right virtual boundary constraint of the lane topology L i1 ⁇ L o2 is shown in area U70.
  • an optimization algorithm based on Spiral curves is used to generate smooth and human-like virtual lane trajectories for the six lane topologies in the lane-level fully connected topology.
  • the six lane topologies in the lane-level fully connected topology generate several candidate curves.
  • the optimal curve b of the lane topology L i1 ⁇ L o1 and the optimal curve e of the lane topology L i2 ⁇ L o2 both satisfy the soft and hard boundary constraints and the virtual boundary constraints, and have no collision with the boundary obstacles in the scene, so there is no Local curve shape adjustment is required.
  • the lane sidelines of the entering lane L i1 , L i2 and the exiting lane L o1 , L o2 , L o3 to the projection line and calculate the alignment coefficient of the left and right lane sidelines on the projection line (ie, the projection overlap coefficient); if the alignment coefficient is greater than the preset threshold, the lane topology is retained; if the alignment coefficient is not greater than the preset threshold, the lane topology is deleted from the lane-level fully connected topology. If all alignment coefficients of an incoming lane are less than the set threshold, optionally, the lane topology with the largest alignment coefficient may be retained.
  • the alignment factor 0.5 is greater than 1/3.
  • the alignment factor 0.5 is greater than 1/3.
  • the alignment factor 0.4 is greater than 1/3.
  • the alignment factor 0.6 is greater than 1/3.
  • the overlap length of the lane topology L i1 ⁇ L o2 lane width alignment factor is greater than 1/3.
  • the overlapping length w same (L in-1 ,L out-3 ) of the lane topology L i1 ⁇ L o3 0, the lane width
  • the alignment factor is 0.
  • the overlap length of the lane topology L i2 ⁇ L o3 lane width alignment factor is greater than 1/3.
  • the alignment coefficients of the lane topologies L i1 ⁇ L o2 and L i2 ⁇ L o3 are greater than the preset threshold, the corresponding lane topologies are retained, but the lane topologies L i1 ⁇ L o1 , L i1 ⁇ L o3 and L i2 ⁇ L o2 , the alignment coefficient of L i2 ⁇ L o1 is less than the preset threshold, the corresponding lane topology and its curve should be deleted from the lane-level fully connected topology.
  • the set of lane topological curves retained after the above two examples are filtered ⁇ L i1 ⁇ L o1 , L i1 ⁇ L o2 , L i2 ⁇ L o2 , L i2 ⁇ L o3 ⁇ or ⁇ L i1 ⁇ L o2 , L i2 ⁇ L o3 ⁇ , the curvature of each lane topology is smooth, which meets the requirements of the turning radius of the vehicle, so it meets the kinematics screening principle; because the traffic rules of bus lanes, parking lanes, left-turning lanes, and right-turning lanes have been considered in the generation of lane topology , so the generated lane-level fully connected topology also meets the screening conditions of traffic rules; and the topological curves of each lane do not collide with the obstacle areas in the scene, and there is no interference between the trajectories and the waiting area, and There is enough space for passage, so it also meets the collision detection screening principle and the traffic interference screening
  • the set of lane topological curves retained after screening does not need to be topologically supplemented.
  • the set of lane topological curves retained after screening is ⁇ L i1 ⁇ L o2 , L i2 ⁇ L o3 ⁇ , and the exit lane L o1 has no reasonable lane topology, so it should be aligned according to the left / Right-aligned supplementary principles, exit topology supplementary principles, and adjacent lane supplementary principles are used for topological complementation.
  • the lane topology L i1 ⁇ L o1 should be supplemented for the exit lane L o1 ; because according to the right-aligned lane correspondence, the exit lane L o1 has no corresponding entry lane, so there is no need to supplement the right alignment Lane topology, the set of lane topology curves retained so far is ⁇ L i1 ⁇ L o1 , L i1 ⁇ L o2 , L i2 ⁇ L o3 ⁇ , so it does not meet the conditions of exit topology supplementary principle and adjacent lane supplementary principle.
  • a complete and reasonable lane topology curve set ⁇ L i1 ⁇ L o1 , L i1 ⁇ L o2 , L i2 ⁇ L o2 can be obtained , L i2 ⁇ L o3 ⁇ or ⁇ L i1 ⁇ L o1 , L i1 ⁇ L o2 , L i2 ⁇ L o3 ⁇ , so as to complete the road topology analysis of the current scene.
  • This example takes the lane topology recommendation evaluation function including navigation cost, topology cost, smoothness cost, traffic intersection cost, traffic rule cost and traffic efficiency cost as an example, and calculates the topological curves of multiple lanes entering the scene from the same incoming lane
  • the navigation recommendation priority is used to select the comprehensive and optimal lane topology.
  • the global navigation planning route turns left after the intersection to reach the end point, so the distance between each exit lane and the end point is:
  • Topological cost C 2 according to the left-aligned lane correspondence, the lane topology L i1 ⁇ L o1 is left-aligned, and the number of crossing lanes is 0, but the lane topology L i1 ⁇ L o2 crosses to the right according to the principle of left alignment The number of lanes is 1, so
  • Traffic intersection cost C 4 Since there is only one lane topology L i1 ⁇ L o1 in the outgoing lane L o1 , but there are two lane topologies L i1 ⁇ L o2 and L i2 ⁇ L o2 in the outgoing lane L o2 , There is a risk of merging traffic, so
  • Traffic rule cost C 5 the scene of this embodiment is a straight intersection, and there is no priority difference for each exit lane, so
  • Traffic efficiency cost C 6 According to human driving experience and macro traffic flow distribution, in the scenario of this embodiment, the probability of selecting the lane topology L i1 ⁇ L o1 is greater than that of the lane topology L i1 ⁇ L o2 , so
  • the priority of navigation recommendation is: (L i1 ⁇ L o1 )>(L i1 ⁇ L o2 ).
  • the priority of navigation recommendation is: (L i2 ⁇ L o2 )>(L i2 ⁇ L o3 ).
  • the self-vehicle can choose the lane topology L i1 ⁇ L o2 in real time to continue driving, thereby improving comfort and traffic efficiency, as shown in Figure 8g ;
  • the own vehicle can choose the lane topology L i2 ⁇ L o3 in real time, so that Reduce the risk of conflict with other vehicles, improve safety and traffic efficiency, as shown in Figure 8h.
  • This embodiment considers the traffic rules and traffic characteristics of special lanes such as bus lanes, parking lanes, left-turn waiting lanes, and right-turn lanes, and generates a lane-level fully connected topological space on this basis, including all The possible lane topology and the exclusion of unreasonable lane topology that violates traffic rules lay the foundation for the construction of a complete and reasonable road topology space.
  • special lanes such as bus lanes, parking lanes, left-turn waiting lanes, and right-turn lanes
  • the hard boundary constraints of each lane topology in this scheme not only consider high-precision maps, but also consider real-time changes in the physical world perceived by sensors, so that this method can be used for both offline map generation and online generation of lane topology trajectories.
  • the soft and hard boundaries such as the waiting area are considered for the nearest neighbor traffic trajectory L i1 ⁇ L o1 or L i2 ⁇ L o3 , and the second nearest neighbor traffic trajectory L i2 ⁇ L o2 or L
  • the interference effects of i1 ⁇ L o1 , L i1 ⁇ L o2 and other lanes in the same direction are gradually weakened, which ensures the passing space and safety between topological trajectories of different lanes, and is more in line with the habits of human drivers and the actual road traffic laws.
  • the road topology analysis results are highly human-like: for the case of many-to-many and large projected overlapping areas in Figure 8d, the lane topology is preserved Trajectory set ⁇ L i1 ⁇ L o1 , L i1 ⁇ L o2 , L i2 ⁇ L o2 , L i2 ⁇ L o3 ⁇ ; For the case of many-to-many but small overlapping area in Figure 8e, filter out the invalid lane topology, and Add side road topology for topology completion, and get the lane topological trajectory set ⁇ L i1 ⁇ L o1 , L i1 ⁇ L o2 , L i2 ⁇ L o3 ⁇ , improve the richness of the lane topological trajectory set, and ensure the completeness of road topology analysis And rationality, fundamentally guarantee the freedom and flexibility of real-time lane decision-making of vehicles
  • This embodiment also introduces global navigation information into the navigation recommendation and evaluation of the lane topological trajectories in the lane topological trajectory set ⁇ L i1 ⁇ L o1 , L i1 ⁇ L o2 , L i2 ⁇ L o2 , L i2 ⁇ L o3 ⁇ , lane topology and intersection, traffic rule priority, macro-traffic flow information, etc.
  • it can also combine dynamic and static traffic environments in real-time navigation to assist vehicles in choosing the optimal target path in real time, so as to have a global vision and avoid lane crossings in advance.
  • the topology of high-risk lanes with traffic conflicts such as lateral extrusion of other vehicles and multi-lane merging can reduce complex interactions with other vehicles, and effectively improve traffic efficiency and comfort in situations of vehicle intersection and dense traffic.
  • the lane-level fully connected topological space generation of this scheme takes into account the traffic rules and traffic characteristics of special lanes such as bus lanes, parking lanes, left-turn waiting lanes, and right-turn lanes, including all Possible lane topology, and exclude unreasonable lane topology that violates traffic rules, laying the foundation for building a complete and reasonable road topology space;
  • the boundary constraint generation of the lane topology considers the real-time changes of the physical world perceived by the sensor, so that this method can be used for the online generation of the lane topology trajectory; considering the soft and hard boundaries and the traffic trajectory gradually weakening the same direction lane Interference effects and indirect constraints ensure the passing space and safety between topological trajectories of different lanes, which is more in line with the habits of human drivers and actual road traffic laws; the straightness of curves (traffic efficiency) and The passage space between adjacent lanes ensures the smoothness and human-likeness of the virtual lane trajectory, and reduces the traffic conflict with other lanes;
  • this scheme deletes unreasonable lane topology through human-like projection overlap screening and side road topology completion, and constructs a complete and reasonable road topology space to ensure Ensure the completeness and rationality of road topology analysis in scenarios where there are multiple reasonable driving trajectories, and effectively improve the freedom and flexibility of real-time lane decision-making for vehicles in dense traffic scenarios;
  • this scheme considers the global navigation information, lane topology and intersection, traffic rule priority, macro traffic flow information, etc. in the navigation recommendation evaluation of the lane topological trajectory, so that the vehicle has a global vision and can combine dynamic and static In the traffic environment, avoid high-risk lanes in advance, reduce interaction with other vehicles, and effectively improve the comfort and traffic efficiency in the case of vehicle intersection and dense traffic.
  • the scene in this embodiment is a roundabout scene.
  • Vehicles need to enter the roundabout via the road below and leave the roundabout via the first exit on the lower right or the second exit below.
  • the road entering the roundabout may include one or more driving lanes, and the road cruising along the roundabout may also have one or more lanes along the roundabout. According to specific driving tasks, it is possible to leave the roundabout after passing through several roundabouts, or to leave the roundabout at the next roundabout.
  • entering the roundabout is a road junction A
  • leaving the roundabout is a road junction B or C.
  • the topology of entering the roundabout and driving out of the roundabout intersection is shown in Figure 9a .
  • the generated boundary constraints are introduced by taking the lane topologies 1-4 and 2-5 entering the roundabout as examples.
  • Hard boundary constraints include road boundaries, curbs, green belts, and lane lines that cannot be crossed, as shown in the thick line and the route boundary around the island in Figure 9b;
  • the soft boundary constraint is a lane line that can be crossed, such as a dashed lane boundary;
  • the virtual boundary constraint is the shape constraint obtained after translation of the soft boundary and hard boundary constraints, as shown in the curve pointed by the arrow in Fig. 9b.
  • lane topology 2-5 may collide with the right hard boundary constraint, so the hard boundary will affect the traffic trajectory of lane topology 2-5; indirectly, the hard boundary will affect the next adjacent lane topology 1- 4, so the hard constraint is moved to the corresponding position (the curve pointed by the arrow in Figure 9b) through the principle of translation, and the virtual boundary constraints of lane topology 1-4 are obtained.
  • the end entering the roundabout belongs to the real intersection (boundary L1), and the end along the roundabout (boundary L2) belongs to the virtual intersection boundary. Therefore, for roundabout intersections, where the boundary line is at the boundary of the virtual intersection in the roundabout, the start-stop pose sampling points generated by the lane topology curve can extend to the outside of the virtual intersection, as shown in point a, point b, and point c in Figure 9c, thereby generating reasonable, Flat human-like lane topology curves without being overly constrained by virtual intersection boundaries in high-resolution maps.
  • the method for generating the lane curve is similar to that of the first embodiment, and will not be repeated here. Among them, the virtual lane curves generated by lane topologies 1-4 are shown by the dotted curves in Fig. 9c, and the optimal trajectory obtained by the comprehensive evaluation is shown by the solid line curves.
  • topologies of the reserved lanes entering the intersection are shown as a, b, and c in Figure 9d, and 1- 5, 2-3, 2-4 three alternative lane topologies; similarly, the exit lane topology of the intersection is shown in Figure 9d as d, e, f three, and the three lanes 3-6, 4-6, 5-7 are screened out Alternative lane topologies.
  • the costs that have a greater impact in the evaluation function of the navigation recommendation include:
  • Topological cost topological cost of c > topological cost of b > topological cost of a, topological cost of d > topological cost of e > topological cost of f.
  • Traffic flow intersection cost For the case where the percentage of the roundabout route is relatively high (the second one leaves the roundabout intersection), the traffic flow intersection cost of lanes 3, 4, and 5 in the roundabout is: the traffic flow intersection of lane 5 Cost >> intersection cost of lane 4 >> intersection cost of lane 3;
  • Traffic regulation cost Traffic regulation cost of c>traffic regulation cost of b>traffic regulation cost of a, traffic regulation cost of d>traffic regulation cost of e>traffic regulation cost of f.
  • Traffic efficiency cost traffic efficiency cost of a > traffic efficiency cost of b > traffic efficiency cost of c, traffic efficiency cost of 5 > traffic efficiency cost of 4 > traffic efficiency cost of 3, traffic efficiency cost of f > e The traffic efficiency cost of > the traffic efficiency cost of d.
  • the optimal path 2-a-5-4-5-f-6 can be realized by changing lanes; similarly, considering that lane 4 is better than lane 3, Therefore, the optimal path 1-b-4-5-f-6 can be achieved by changing lanes.
  • the optimal routes obtained from the navigation recommendation evaluation in this embodiment are: 1-b-4-5-f-6 and 2-a-5-4-5-f-6.
  • this embodiment reasonably extends the sampling points of the beginning and end attitudes of the side trajectory of the virtual intersection to the outside, improves the quality of the generated trajectory, improves the human-like nature of the trajectory, and avoids unreasonable trajectory caused by high-precision map drawing. It also avoids wrong screening due to unreasonable trajectory generation in topology screening, ensuring the completeness of road topology analysis.
  • the indirect constraint effect of the virtual boundary under the topology of multiple parallel lanes is considered, which improves the human-likeness and passability of the generated trajectory, and avoids collisions with adjacent roads.
  • the interference effect of lane topology improves the safety of entering and exiting intersections around the island;
  • This embodiment builds a complete lane topological space by retaining multiple reasonable lane topologies to ensure that traffic efficiency, safety, and comfort can be effectively improved in various traffic environments: when the traffic flow in the roundabout is dense, enter the roundabout as soon as possible The outermost circle drives along the roundabout; when too many vehicles enter the outer circle of the roundabout, directly enter the middle lane and drive along the roundabout to avoid long waiting times caused by traffic congestion and improve traffic efficiency; when the traffic flow in the roundabout is smooth, drive directly into the The innermost lane runs along the roundabout, which can effectively reduce the intersection of traffic entering and exiting the roundabout, reduce the risk of self-cars, short driving distance, and high traffic efficiency.
  • FIG. 10a and FIG. 10b it is a scene of a left-turn intersection with waiting areas and traffic lights provided by the embodiment of the present application.
  • the position of the stop line considering the traffic rules (traffic light status and the information of the area to be turned at the stop line), and other boundary constraints, generate: (1) When the left-turn light is red, the traffic at the stop line in the area to be turned The topological curve of lane-level fully connected lanes is shown as the dotted line at the intersection in Figure 10a; (2) When the left turn light is green, the topological trajectory of lane-level fully connected lanes at the stop line at the intersection is shown as the dotted line at the intersection in Figure 10b.
  • the foregoing embodiments which will not be repeated here.
  • the vehicle drives to the left-turn intersection shown in the figure, if the sensor senses in real time that the current left-turn light is red and the straight-going light is green, the vehicle first enters the waiting area and waits for the left-turn light to turn green.
  • the optimal recommended lane topology curve is shown as the solid line curve in Fig. 10a. If the sensor senses that the current left-turn light is green in real time, the curvature and traffic efficiency costs of the optimal lane topology recommended by the navigation in Figure 10a are too high, so the optimal recommendation is the innermost inscribed lane curve that does not consider the area to be turned, as shown in Shown by the solid line curve in Figure 10b.
  • different lane-level full-connection topology curves are generated to ensure the completeness of road topology analysis by considering different traffic times (red lights, green lights) and different positions of the vehicle. , can recommend the best lane topology curve that is most suitable for the current moment under any traffic light state, avoiding the lane topology curve being too rigid, unable to adapt to real-time traffic state changes, and improving the human-like trajectory.
  • FIG. 11a and FIG. 11b it is a small S-curve scene provided by the embodiment of the present application.
  • human drivers often drive across lane lines in pursuit of a straighter driving path. If the self-driving vehicle can only drive along the lane topological curve determined by the lane line, when encountering a scene where a human driver crosses the lane line, it will trigger an emergency avoidance behavior, which will reduce the comfort level, the experience is not good, and even There is a risk of collision, as shown in Figure 11a.
  • the self-driving vehicle can select the virtual lane topology curve 3 according to the behavior of other car 1 to realize cross-lane driving through this small S-curve scene, avoiding the possibility of collision with other car 1, and improving the safety of the self-driving car ; but when there is no influence of other cars, since the total cost of navigation recommendation evaluation for the virtual lane topological curve crossing the lane line is higher, the virtual lane topological curve that does not cross the lane line will be recommended for the self-vehicle first, as shown in Figure 11b. Curves 1, 2, 4, and 6 are shown.
  • FIG. 12 it is an urban elevated transition section-ramp scene provided by the embodiment of the present application.
  • the only way to enter the ramp provided by the general high-precision map is the lane topology curve 1 as shown in Figure 12.
  • the own car may be suppressed by the right vehicle (or obstacle) in the second right lane, and cannot change lanes for a long time. The rightmost lane, thus missing the opportunity to enter the ramp.
  • the three lane topological curves 1, 2, and 3 shown in FIG. 12 can be generated by adopting the method described in the foregoing embodiments, and a complete road topology analysis can be obtained.
  • the vehicle is suppressed by the right vehicle (or obstacle) in the second right lane, it can choose the virtual lane topology curve 2 to enter the ramp; if the traffic is smooth or there are no obstacles, Since the navigation recommendation evaluation of virtual lane topological curve 1 is the best, virtual lane topological curve 2 is next, and virtual lane topological curve 3 is again.
  • the ego vehicle will first enter the rightmost lane by changing lanes, and then choose virtual lane topological curve 1 to enter On the ramp, it is more in line with the driving habits of people.
  • FIG. 13 it is a multi-lane scene without lane markings provided by the embodiment of the present application.
  • multi-lane unmarked road sections usually appear in traffic scenarios such as changes in the number of lanes and non-intersection pedestrian crossings. Since there is no road intersecting with the traffic flow, this kind of road section scene is not a conventional intersection, but this scene is also applicable to the intersection scene described in this scheme.
  • a reasonable and complete set of virtual lane topological curves can be generated by adopting the method described in the preceding embodiments and an appropriate navigation recommendation evaluation can be given, as shown in the lane topological curve (solid line curve in the figure) in Figure 13 .
  • FIG. 14 it is a continuous turning scene provided by the embodiment of the present application.
  • This scenario needs to complete the driving task of turning right immediately after turning left.
  • High-precision maps generally only provide the virtual lane topology curve 1 of entering the left lane of the road and leaving the left lane of the road at left-turn intersections.
  • the vehicle will face the problem of changing lanes three times in a short distance. problem, it is difficult to complete in actual scenarios.
  • this solution can generate a reasonable and complete set of virtual lane topological curves by using the method described in the previous embodiment, and perform navigation recommendation evaluation, which obtains virtual lane topological curves 1, 2 and 3. Therefore, the vehicle can Select the virtual lane topology curve 3 to turn left at the first intersection, and then only need to change lanes once to turn right at the second intersection, which ensures the freedom of lane selection and greatly improves the success rate and traffic efficiency.
  • FIG. 15 it is a narrow road U-turn scene provided by the embodiment of the present application.
  • a complete lane-level fully-connected topology can be generated by using the method described in the foregoing embodiments, and virtual lane topology curves 1 and 2 can be obtained.
  • the virtual lane topological curve 1 cannot satisfy the kinematic screening principle (small-radius U-turn curve)
  • the virtual lane topological curve 1 is deleted, and only the virtual lane topological curve 2 is retained; if the virtual lane topological curve 1 , and 2 do not satisfy the kinematic screening principle, a lane topological curve with a larger curvature radius is retained to ensure the connectivity of the lane topology, thereby forming a complete and reasonable set of lane topological curves and ensuring the human-like nature of the lane topological trajectory.
  • this solution also provides an intersection-based map generation method, including: generating M Lane topological curves, the lane topological curves are curves whose endpoints are the end of the entry lane of the entry road and the starting point of the exit lane of the exit road; Rationality detection processing to obtain K' lane topological curves, wherein K' is not greater than M; generating a map of the intersection according to the K' lane topological curves in the intersection.
  • FIG. 16 it is a schematic diagram of a device for guiding a vehicle according to an embodiment of the present application. As shown in Figure 16, it includes a curve generation module 1601, a detection processing module 1602 and a determination module 1603, wherein:
  • the curve generation module 1601 is configured to generate M lane topological curves according to the intersection, the entry road of the intersection, the obstacles in the exit road of the intersection, and the lane lines, and the lane topological curves are based on the entry a curve terminating at the end of the entry lane of the road and the beginning of the exit lane of said exit road;
  • the detection processing module 1602 is configured to perform rationality detection processing on the M lane topological curves to obtain K' lane topological curves, wherein K' is not greater than M;
  • a determination module 1603, configured to determine a target path from the K' lane topology curves when the vehicle is on the entry road.
  • the curve generation module 1601 is used for:
  • the entry road of the intersection, the obstacle in the exit road of the intersection and the lane line that cannot be crossed, the hard boundary constraint of the lane topology curve is obtained;
  • the M lane topological curves are generated according to the lane topological curve hard boundary constraints, the lane topological curve soft boundary constraints, and the K lane topological curve virtual boundary constraints.
  • the K lane topological curve virtual boundary constraints correspond to the K lane topologies
  • any lane topology curve virtual boundary constraint A in the K lane topological curve virtual boundary constraints is obtained by placing the leftmost The hard boundary constraint and/or soft boundary constraint on the left side of the side lane topology is translated to the right by a first preset distance, and the hard boundary constraint and/or soft boundary constraint on the right side of the rightmost lane topology of the intersection is translated to the left
  • the first preset distance is obtained, wherein the first preset distance is determined according to the lane sequence of the lane topology A', or the first preset distance is determined according to the vehicle preset passing width, lane determined by at least one of the width and the lane order of the lane topology A', the lane topology curve virtual boundary constraint A corresponds to the lane topology A', and the K lane topologies include the most The left lane topology and the rightmost lane topology.
  • curve generating module 1601 is also used for:
  • the plurality of curves are screened according to the lane topological curve hard boundary constraints, the lane topological curve soft boundary constraints and the K lane topological curve virtual boundary constraints to obtain the M lane topological curves.
  • the curve generation module 1601 is also used for:
  • the plurality of smooth curves are generated according to the at least one start point pose vector, the at least one end point pose vector, and the plurality of control points.
  • the pose vector of at least one starting point in the incoming road is obtained by extending the end of each incoming lane for a second preset distance and performing sampling.
  • the detection processing module 1602 is used for:
  • the projection line between the entering road and the exiting road is obtained, and the projection line is the direction vector of the entering road, the direction vector of the exiting road
  • the alignment coefficient is the ratio between the first parameter and the second parameter, and the first parameter is the lane edge line of each entering lane and the lane edge line of each exiting lane extended to the projection line respectively.
  • the overlapping length between the obtained two line segments, the second parameter is the two obtained by extending the lane sideline of each entering lane and the lane sideline of each exiting lane respectively to the projection line the length of the shortest line segment among the line segments;
  • K' lane topological curves are obtained, wherein the K' lane topological curves include A curve whose end point is the end of the entry lane and the starting point of the exit lane is the end point of the entry lane whose coefficient is greater than the first preset threshold.
  • the K' lane topological curves include curves whose endpoints are respectively the end of the leftmost inbound lane of the inbound road and the starting point of the leftmost outbound lane of the outbound road, and Curves whose endpoints are respectively the end of the rightmost inbound lane of the inbound road and the starting point of the rightmost outbound lane of the outbound road.
  • the K' lane topology curves include lane topologies whose endpoints are the end of the entry lane X in the entry road and the starting point of the left lane of the exit lane Y in the exit road curve, and the lane topology curve with the end point of the entry lane X and the starting point of the right lane of the exit lane Y as the endpoint, and also includes the end point of the entry lane X and the exit lane
  • the starting point of Y is the lane topological curve of the end point, wherein there are lanes on the left and right sides of the exit lane Y; or,
  • the K' lane topology curves include lanes whose endpoints are the end of the entry lane X in the entry road and the starting point of the left or right lane of the exit lane Y in the exit road
  • the topological curve further includes a lane topological curve whose endpoints are the end of the incoming lane X in the incoming road and the starting point of the outgoing lane Y, wherein the outgoing lane Y in the outgoing road is only There is a lane on the left or right.
  • the maximum curvature of each lane topological curve in the K' lane topological curves is not greater than a second preset threshold, and each lane topological curve is bounded by the soft boundary of the lane topological curve, the The distance between the hard boundary constraints of the lane topological curves is not less than a third preset distance, and the distance between any two lane topological curves is not less than a fourth preset distance.
  • the device also includes an evaluation module for:
  • the evaluation value is related to the curvature of the lane topological curve, the rate of curvature change, the number of diagonally crossing lanes, and the lane of the lane corresponding to the lane topological curve at least one of intersection information, traffic rule information, estimated traffic flow, and travelable distance;
  • the determining module 1603 is configured to:
  • the target path is determined according to the evaluation value of each lane topological curve in the K' lane topological curves.
  • the intersection includes at least one of an intersection, a roundabout, an intersection in a waiting area, a small S-curve, an elevated entrance and exit, a multi-lane road section without lane markings, a continuous turning intersection, and a narrow road U-turn intersection.
  • the guiding vehicle traveling device is presented in the form of modules.
  • a “module” here may refer to an application-specific integrated circuit (ASIC), a processor and memory executing one or more software or firmware programs, an integrated logic circuit, and/or other devices that can provide the above functions .
  • ASIC application-specific integrated circuit
  • processor may execute one or more software or firmware programs, an integrated logic circuit, and/or other devices that can provide the above functions .
  • the above curve generation module 1601 , detection processing module 1602 and determination module 1603 can be realized by the processor 1702 of the guiding vehicle driving device shown in FIG. 17 .
  • the present solution also provides an intersection-based map generation device, including: a curve generation module, configured to obtain information according to the intersection, the entry road of the intersection, the obstacles in the exit road of the intersection, and the lane line Generate M lane topological curves, the lane topological curves are curves whose endpoints are the end of the entry lane of the entry road and the starting point of the exit lane of the exit road; a detection processing module for The M lane topological curves are subjected to rationality detection processing to obtain K' lane topological curves, wherein K' is not greater than M; the map generation module is used to generate the K' lane topological curves according to the intersection. map of intersections.
  • a curve generation module configured to obtain information according to the intersection, the entry road of the intersection, the obstacles in the exit road of the intersection, and the lane line Generate M lane topological curves, the lane topological curves are curves whose endpoints are the end of the entry lane of the entry road and the starting point of
  • the device may also include the above-mentioned modules, which are not specifically limited in this solution.
  • Fig. 17 is a schematic diagram of the hardware structure of the device for guiding vehicles provided by the embodiment of the present application.
  • the guiding vehicle driving device 1700 shown in FIG. 17 includes a memory 1701 , a processor 1702 , a communication interface 1703 and a bus 1704 .
  • the memory 1701 , the processor 1702 , and the communication interface 1703 are connected to each other through a bus 1704 .
  • the memory 1701 may be a read-only memory (Read Only Memory, ROM), a static storage device, a dynamic storage device or a random access memory (Random Access Memory, RAM).
  • ROM Read Only Memory
  • RAM Random Access Memory
  • the memory 1701 may store programs, and when the programs stored in the memory 1701 are executed by the processor 1702, the processor 1702 and the communication interface 1703 are used to execute various steps of the method for guiding a vehicle in the embodiment of the present application.
  • the processor 1702 may be a general-purpose central processing unit (Central Processing Unit, CPU), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), a graphics processing unit (graphics processing unit, GPU) or one or more
  • the integrated circuit is used to execute related programs to realize the functions required by the units in the device for guiding the vehicle in the embodiment of the present application, or to execute the method for guiding the vehicle in the method embodiment of the present application.
  • the processor 1702 may also be an integrated circuit chip with signal processing capability. During implementation, each step of the method for guiding a vehicle in the present application may be completed by an integrated logic circuit of hardware in the processor 1702 or instructions in the form of software.
  • the above-mentioned processor 1702 can also be a general-purpose processor, a digital signal processor (Digital Signal Processing, DSP), an application-specific integrated circuit (ASIC), a ready-made programmable gate array (Field Programmable Gate Array, FPGA) or other programmable logic devices , discrete gate or transistor logic devices, discrete hardware components.
  • DSP Digital Signal Processing
  • ASIC application-specific integrated circuit
  • FPGA Field Programmable Gate Array
  • Various methods, steps, and logic block diagrams disclosed in the embodiments of the present application may be implemented or executed.
  • a general-purpose processor may be a microprocessor, or the processor may be any conventional processor, and the like.
  • the steps of the method disclosed in connection with the embodiments of the present application may be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor.
  • the software module can be located in a mature storage medium in the field such as random access memory, flash memory, read-only memory, programmable read-only memory or electrically erasable programmable memory, register.
  • the storage medium is located in the memory 1701, and the processor 1702 reads the information in the memory 1701, and combines its hardware to complete the functions required by the units included in the guiding vehicle driving device of the embodiment of the application, or execute the guidance of the method embodiment of the application Vehicle driving method.
  • the communication interface 1703 implements communication between the apparatus 1700 and other devices or communication networks by using a transceiver device such as but not limited to a transceiver. For example, data can be acquired through the communication interface 1703 .
  • the bus 1704 may include pathways for transferring information between various components of the device 1700 (eg, memory 1701 , processor 1702 , communication interface 1703 ).
  • the device 1700 shown in FIG. 17 only shows a memory, a processor, and a communication interface, in the specific implementation process, those skilled in the art should understand that the device 1700 also includes other devices necessary for normal operation. . Meanwhile, according to specific needs, those skilled in the art should understand that the apparatus 1700 may also include hardware devices for implementing other additional functions. In addition, those skilled in the art should understand that the device 1700 may also only include the devices necessary to realize the embodiment of the present application, and does not necessarily include all the devices shown in FIG. 17 .
  • the present application also provides an intelligent driving vehicle, including a traveling system, a sensor system, a control system and a computer system, wherein the computer system is used to execute one or more steps in any one of the above methods.
  • the embodiment of the present application also provides a computer-readable storage medium, the computer-readable storage medium stores instructions, and when it is run on a computer or a processor, the computer or the processor executes one of the above-mentioned methods or multiple steps.
  • the embodiment of the present application also provides a computer program product including instructions.
  • the computer program product is run on the computer or the processor, the computer or the processor is made to perform one or more steps in any one of the above methods.
  • words such as “first” and “second” are used to distinguish the same or similar items with basically the same function and effect.
  • words such as “first” and “second” do not limit the quantity and execution order, and words such as “first” and “second” do not necessarily limit the difference.
  • words such as “exemplary” or “for example” are used as examples, illustrations or illustrations. Any embodiment or design scheme described as “exemplary” or “for example” in the embodiments of the present application shall not be interpreted as being more preferred or more advantageous than other embodiments or design schemes.
  • the use of words such as “exemplary” or “such as” is intended to present related concepts in a concrete manner for easy understanding.
  • the disclosed systems, devices and methods may be implemented in other ways.
  • the division of this unit is only a logical function division, and there may be other division methods in actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored, or not implement.
  • the mutual coupling, or direct coupling, or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection of devices or units may be in electrical, mechanical or other forms.
  • a unit described as a separate component may or may not be physically separated, and a component displayed as a unit may or may not be a physical unit, that is, it may be located in one place, or may be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.
  • all or part of them may be implemented by software, hardware, firmware or any combination thereof.
  • software When implemented using software, it may be implemented in whole or in part in the form of a computer program product.
  • the computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on the computer, the processes or functions according to the embodiments of the present application will be generated in whole or in part.
  • the computer can be a general purpose computer, a special purpose computer, a computer network, or other programmable device.
  • the computer instructions may be stored in or transmitted over a computer-readable storage medium.
  • the computer instructions can be sent from one website site, computer, server, or data center to another by wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.)
  • wired such as coaxial cable, optical fiber, digital subscriber line (DSL)
  • wireless such as infrared, wireless, microwave, etc.
  • the computer-readable storage medium may be any available medium that can be accessed by a computer, or a data storage device such as a server or a data center integrated with one or more available media.
  • the usable medium can be read-only memory (read-only memory, ROM), or random access memory (random access memory, RAM), or magnetic medium, for example, floppy disk, hard disk, magnetic tape, magnetic disk, or optical medium, such as , a digital versatile disc (digital versatile disc, DVD), or a semiconductor medium, for example, a solid state disk (solid state disk, SSD) and the like.
  • read-only memory read-only memory
  • RAM random access memory
  • magnetic medium for example, floppy disk, hard disk, magnetic tape, magnetic disk, or optical medium, such as , a digital versatile disc (digital versatile disc, DVD), or a semiconductor medium, for example, a solid state disk (solid state disk, SSD) and the like.

Landscapes

  • Engineering & Computer Science (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Automation & Control Theory (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Traffic Control Systems (AREA)
  • Navigation (AREA)

Abstract

一种引导车辆行驶的方法、路口的地图生成方法及相关系统、存储介质。引导车辆行驶的方法包括：根据路口、路口的驶入道路、路口的驶出道路中的障碍物以及车道线生成M条车道拓扑曲线，车道拓扑曲线为以驶入道路的驶入车道的末端和驶出道路的驶出车道的起始点为端点的曲线（201）；对M条车道拓扑曲线进行合理性检测处理，以得到K'条车道拓扑曲线，其中，K'不大于M（202）；当车辆位于驶入道路中时，从K'条车道拓扑曲线中确定目标路径（203）。由此生成的完备合理的车道拓扑曲线更加符合人类驾驶习惯，可以实现在自动驾驶时为车辆提供相应的导航引导信息，有效提升车辆通过路口的轨迹类人性和通行效率。

Description

引导车辆行驶的方法、地图生成方法及相关系统
本申请要求于2022年1月12日提交中国专利局、申请号为202210046464.0、申请名称为“引导车辆行驶的方法、地图生成方法及相关系统”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及车辆技术领域，尤其涉及一种引导车辆行驶的方法、路口的地图生成方法及相关系统、存储介质。
背景技术
在汽车自动驾驶领域，对于不存在实际车道线的交叉路口或存在多种合理驾驶轨迹的S弯、环岛出入口、高架入口等场景中，自动驾驶系统需要进行合理的道路拓扑分析，并提供相应的拓扑导航引导信息，以供车辆进行意图预测、轨迹预测、车道决策和运动规划等。而高质量的道路拓扑分析，不仅要求车道轨迹符合人类驾驶习惯，而且可供合理通行的车道拓扑数量及其导航引导信息也应与人类驾驶经验相符，才能提高自动驾驶车辆行驶轨迹的类人性和智能性，用于他车意图和轨迹预测也更准确。
现有的道路拓扑分析方法大致可以分为两类：基于高精地图制图的道路拓扑生成或基于车辆轨迹聚类的道路拓扑生成。基于高精地图制图的道路拓扑生成方法，大多依赖制图方式获取驶入道路和驶出道路的端点或方向向量，并计算方向向量夹角，再对端点或方向向量进行曲线拟合，因此生成的线型单一且泛化性不足，不能覆盖现实场景中的差异化场景，需要人工干预；而基于车辆轨迹聚类的道路拓扑生成方法，依赖大数据技术收集人驾数据或众包轨迹，再对轨迹进行筛选和聚类，生成车道拓扑曲线，因此必然将带来大量的数据预处理工作，且道路拓扑生成质量与采集的数据质量密切相关，无法得到保证。此外，上述两种方法均无法保证生成的道路拓扑的完备性，也不能区分多条道路拓扑对于当前场景下驾驶行为的优劣，难以辅助意图预测和驾驶决策，且生产过程耗时费力，更新周期长，只能用于离线制图。
因此，如何解决上述问题，实现复杂场景下具有良好泛化性类人性的道路拓扑分析和导航引导是当前需要解决的问题。
目前，通过从两条交汇道路边界上位于交叉路口的结点中，选取路口车道的参考线端点并确定其边界线端点，基于路口驶入道路和驶出道路的端点或方向向量，计算方向向量夹角，最后对端点或方向向量进行曲线拟合，进而实现自动生成路口内虚拟车道参考线及边界。然而，该方案生成的参考线线型单一，对不同路口需要人工调节参数来保证输出质量；且对于非正对的直行路口、有花坛或路牙、围栏等障碍物的场景，无法自动生成可用的参考曲线或者生成的参考曲线轨迹不类人；且对于复杂的多对多路口，生成的道路拓扑不完备，同时也可能存在不合理的道路拓扑，需要人工干预进行合理的拓扑连接关系选择。
发明内容
本申请公开了一种引导车辆行驶的方法、路口的地图生成方法及相关系统、存储介质，可以实现在自动驾驶时为车辆提供相应的导航引导信息，有效提升车辆通过该路口的轨迹类人性和通行效率。
第一方面，本申请实施例提供一种引导车辆行驶的方法，包括：根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及车道线生成M条车道拓扑曲线，所述车道拓扑曲线为以所述驶入道路的驶入车道的末端和所述驶出道路的驶出车道的起始点为端点的曲线；对所述M条车道拓扑曲线进行合理性检测处理，以得到K’条车道拓扑曲线，其中，K’不大于M；当所述车辆位于所述驶入道路中的第一驶入车道时，从所述K’条车道拓扑曲线中确定目标路径，所述目标路径包括所述K’条车道拓扑曲线中端点为所述第一驶入车道的末端的车道拓扑曲线。
通过本申请实施例，基于路口、路口的驶入道路、路口的驶出道路中的障碍物以及车道线生成车道拓扑曲线，进而通过合理性检测处理得到K’条车道拓扑曲线，当车辆位于驶入道路中的第一驶入车道时，从所述K’条车道拓扑曲线中确定目标路径。本方案基于实际场景生成的完备合理的车道拓扑曲线更加符合人类驾驶习惯，更加合理，无需人工干预，具有良好泛化性类人性，可以实现在自动驾驶时为车辆提供相应的导航引导信息，有效提升车辆通过该路口的轨迹类人性和通行效率。
作为一种可选的实现方式，所述根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及车道线生成M条车道拓扑曲线，包括：根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及不可跨越车道线得到车道拓扑曲线硬边界约束；根据所述路口、所述路口的驶入道路、所述路口的驶出道路中的可跨越车道线得到车道拓扑曲线软边界约束；根据所述车道拓扑曲线硬边界约束和所述车道拓扑曲线软边界约束得到K个车道拓扑曲线虚拟边界约束；根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束生成所述M条车道拓扑曲线。
该手段基于高精度地图和实时感知的道路中的障碍物以及车道线得到软、硬边界约束以及虚拟边界约束，保证了虚拟车道轨迹的类人性以及车型普适性，提高了生成车道拓扑曲线的可靠性。
其中，所述K个车道拓扑曲线虚拟边界约束与K个车道拓扑对应，所述K个车道拓扑曲线虚拟边界约束中任一车道拓扑曲线虚拟边界约束A是通过将所述路口的最左侧车道拓扑左侧的硬边界约束和/或软边界约束向右平移第一预设距离，以及将所述路口的最右侧车道拓扑右侧的硬边界约束和/或软边界约束向左平移第一预设距离得到的，其中，所述第一预设距离是根据车道拓扑A’的车道位序确定的，或者，所述第一预设距离是根据所述车辆预设通过宽度、车道宽度中的至少一项以及所述车道拓扑A’的车道位序确定的，所述车道拓扑曲线虚拟边界约束A与所述车道拓扑A’对应，所述K个车道拓扑包括所述路口的最左侧车道拓扑和所述最右侧车道拓扑。
该方案考虑其他车流轨迹的干涉影响、软硬边界对同向车道逐渐减弱的间接约束等，生成虚拟边界约束，保证了虚拟车道轨迹的类人性以及车型普适性。
作为一种可选的实现方式，所述根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束生成所述M条车道拓扑曲线，包括：分别对所述驶入道路中每条驶入车道的末端进行角度采样得到所述驶入道路中的至少一个起始点位姿向量，并分别对所述驶出道路中每条驶出车道的起始点进行角度采样得到所述驶出道路中的至少一个终点位姿向量；对所述至少一个起始点位姿向量和所述至少一个终点位姿向量进行曲线采样，得到所述驶入道路和所述驶出道路之间的多条曲线；根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束对所述多条曲线进行筛选处理，得到所述M条车道拓扑曲线。
该筛选处理例如可以是先筛选满足上述约束的曲线，然后再筛选达到每条驶入车道和每条驶出车道之间至多有一条最优曲线。通过对该最优曲线进行碰撞检测处理，以便适应性调整各曲线，进而得到所述M条车道拓扑曲线。该处理方式仅为一种示例，其还可以是其他方式，本方案对此不做具体限定。
该方案基于驶入车道的末端、驶出车道的起始点进行角度采样，进而生成多条车道拓扑曲线，基于上述得到的软、硬边界约束以及虚拟边界约束对该多条曲线进行筛选处理，进而得到M条车道拓扑曲线。采用该手段，保证了虚拟车道轨迹的类人性、灵活性以及车型普适性，减少与其他车道的车流冲突。
进一步地，所述对所述至少一个起始点位姿向量和所述至少一个终点位姿向量进行曲线采样，得到所述驶入道路和所述驶出道路之间的多条曲线，包括：在所述驶入道路的每条驶入车道的末端和所述驶出道路的每条驶出车道的起始点之间生成多个控制点；根据所述至少一个起始点位姿向量、所述至少一个终点位姿向量以及所述多个控制点，生成所述多条平滑曲线。
该方案在曲线生成时，不仅基于驶入车道的末端、驶出车道的起始点进行角度采样，而且还基于控制点采样，使得生成的曲线数量更多，提高了曲线生成的灵活性。
作为一种可选的实现方式，所述驶入道路中的至少一个起始点位姿向量是通过将所述每条驶入车道的末端延伸第二预设距离并进行采样得到的。
采用该手段，针对驶入、驶出环岛路口特点等场景，对虚拟路口侧轨迹始末姿态采样点向外侧进行合理延伸，提高生成轨迹的质量，提升轨迹类人性，避免高精度地图制图引起的轨迹不合理，也避免了拓扑筛选因轨迹生成不合理而错误筛除，保障道路拓扑分析的完备性。
作为一种可选的实现方式，所述对所述M条车道拓扑曲线进行合理性检测处理，以得到K’条车道拓扑曲线，包括：根据所述驶入道路的方向向量、驶出道路的方向向量得到所述驶入道路、驶出道路之间的投影线，所述投影线为所述驶入道路的方向向量、驶出道路的方向向量相交所得到的夹角的平分线所在的直线，或者，所述投影线为垂直于所述驶出道路的方向向量且通过所述驶出道路的起始点的直线；计算所述驶入道路的每条驶入车道和所述驶出道路的每条驶出车道之间的对齐系数，其中，所述每条驶入车道和所述每条驶出车道之间的对齐系数为第一参数与第二参数之间的比值，所述第一参数为所述每条驶入车道的车道边线以及所述每条驶出车道的车道边线分别延长至所述投影线所得到的两条线段之间的重叠长度，所述第二参数为所述每条驶入车道的车道边线和所述每条驶出车道的车道边线分别延长至所述投影线所得到的两条线段中最短的线段的长度；根据所述驶入道路的每条驶入车道和所述驶出道路的每条驶出车道之间的对齐系数，得到K’条车道拓扑曲线，其中，所述K’条车道拓扑曲线包括以对齐系数大于第一预设阈值的驶入车道的末端、驶出车道的起始点为端点的曲线。
通过基于车道对齐性来筛选曲线，进而得到完备且合理的车道拓扑曲线。
作为又一种可选的实现方式，所述K’条车道拓扑曲线包括分别以所述驶入道路的最左侧驶入车道的末端、所述驶出道路的最左侧驶出车道的起始点为端点的曲线，还包括分别以所述驶入道路的最右侧驶入车道的末端、所述驶出道路的最右侧驶出车道的起始点为端点的曲线。
通过基于拓扑补齐原则来对曲线进行补齐处理，进而得到完备且合理的车道拓扑曲线，保证了虚拟车道曲线的类人性。
作为又一种可选的实现方式，所述K’条车道拓扑曲线包括以所述驶入道路中的驶入车道 X的末端以及所述驶出道路中的驶出车道Y的左侧车道的起始点为端点的车道拓扑曲线，以及以所述驶入车道X的末端以及所述驶出车道Y的右侧车道的起始点为端点的车道拓扑曲线，还包括以所述驶入车道X的末端以及所述驶出车道Y的起始点为端点的车道拓扑曲线，其中，所述驶出车道Y的左侧和右侧均存在车道；或者，所述K’条车道拓扑曲线包括以所述驶入道路中的驶入车道X的末端以及所述驶出道路中的驶出车道Y的左侧车道或者右侧车道的起始点为端点的车道拓扑曲线，还包括以所述驶入道路中的驶入车道X的末端、所述驶出车道Y的起始点为端点的车道拓扑曲线，其中，所述驶出道路中的驶出车道Y仅左侧或者右侧存在车道。
通过基于拓扑补齐原则来对曲线进行补齐处理，进而得到完备且合理的车道拓扑曲线，保证了虚拟车道曲线的合理性、完备性。
作为又一种可选的实现方式，所述K’条车道拓扑曲线中每条车道拓扑曲线的最大曲率不大于第二预设阈值，且，所述每条车道拓扑曲线与所述车道拓扑曲线软边界约束、所述车道拓扑曲线硬边界约束之间的距离不小于第三预设距离，且，任意两条车道拓扑曲线之间的距离不小于第四预设距离。
基于车辆运动学、碰撞检测、交通规则、车流干扰检测等筛选原则进行车道拓扑合理性筛选，进而得到完备且合理的车道拓扑曲线。
作为一种可选的实现方式，所述方法还包括：计算所述K’条车道拓扑曲线中每条车道拓扑曲线的评价值，所述评价值与所述车道拓扑曲线曲率、曲率变化率以及斜穿车道数量、以及所述车道拓扑曲线对应的车道的车道交汇信息、交通规则信息、车流量估计值、可行驶距离中的至少一项有关；所述当所述车辆位于所述驶入道路中的第一驶入车道时，从所述K’条车道拓扑曲线中确定目标路径，包括：当所述车辆位于所述驶入道路中的第一驶入车道时，根据所述K’条车道拓扑曲线中每条车道拓扑曲线的评价值确定所述目标路径，所述目标路径包括所述K’条车道拓扑曲线中端点为所述第一驶入车道的末端的车道拓扑曲线中评价值最高的车道拓扑曲线。
采用该手段，结合全局导航信息、宏观交通流等信息，提供全局视野，对车道级拓扑进行导航推荐评价，为车辆在行驶过程中提供全局视野，提前规避高风险车道拓扑，提升自车通行效率，降低自车风险。
其中，所述路口包括交叉路口、环岛、待转区路口、小S弯、高架出入口、多车道且无车道标线路段、连续转弯路口、窄道掉头路口中的至少一种。
第二方面，本申请提供了一种基于路口的地图生成方法，包括：根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及车道线生成M条车道拓扑曲线，所述车道拓扑曲线为以所述驶入道路的驶入车道的末端和所述驶出道路的驶出车道的起始点为端点的曲线；对所述M条车道拓扑曲线进行合理性检测处理，以得到K’条车道拓扑曲线，其中，K’不大于M；根据所述路口内的K’条车道拓扑曲线生成所述路口的地图。
通过本申请实施例，基于路口、路口的驶入道路、路口的驶出道路中的障碍物以及车道线生成车道拓扑曲线，进而通过合理性检测处理得到K’条车道拓扑曲线，进而生成该路口的K’条车道拓扑曲线。本方案的地图生成，基于实际场景生成的完备合理的车道拓扑曲线更加符合人类驾驶习惯，更加合理，无需人工干预，具有良好泛化性类人性，可以实现在自动驾驶时为车辆提供相应的导航引导信息，有效提升车辆通过该路口的轨迹类人性和通行效率。
作为一种可选的实现方式，所述根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及车道线生成M条车道拓扑曲线，包括：根据路口、所述路口的驶入道路、所 述路口的驶出道路中的障碍物以及不可跨越车道线得到车道拓扑曲线硬边界约束；根据所述路口、所述路口的驶入道路、所述路口的驶出道路中的可跨越车道线得到车道拓扑曲线软边界约束；根据所述车道拓扑曲线硬边界约束和所述车道拓扑曲线软边界约束得到K个车道拓扑曲线虚拟边界约束；根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束生成所述M条车道拓扑曲线。
该手段基于高精度地图和实时感知的道路中的障碍物以及车道线得到软、硬边界约束以及虚拟边界约束，保证了虚拟车道轨迹的类人性以及车型普适性，提高了生成车道拓扑曲线的可靠性。
其中，所述K个车道拓扑曲线虚拟边界约束与K个车道拓扑对应，所述K个车道拓扑曲线虚拟边界约束中任一车道拓扑曲线虚拟边界约束A是通过将所述路口的最左侧车道拓扑左侧的硬边界约束和/或软边界约束向右平移第一预设距离，以及将所述路口的最右侧车道拓扑右侧的硬边界约束和/或软边界约束向左平移第一预设距离得到的，其中，所述第一预设距离是根据车道拓扑A’的车道位序确定的，或者，所述第一预设距离是根据所述车辆预设通过宽度、车道宽度中的至少一项以及所述车道拓扑A’的车道位序确定的，所述车道拓扑曲线虚拟边界约束A与所述车道拓扑A’对应，所述K个车道拓扑包括所述路口的最左侧车道拓扑和所述最右侧车道拓扑。
该方案考虑其他车流轨迹的干涉影响、软硬边界对同向车道逐渐减弱的间接约束等，生成虚拟边界约束，保证了虚拟车道轨迹的类人性以及车型普适性。
作为一种可选的实现方式，所述根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束生成所述M条车道拓扑曲线，包括：分别对所述驶入道路中每条驶入车道的末端进行角度采样得到所述驶入道路中的至少一个起始点位姿向量，并分别对所述驶出道路中每条驶出车道的起始点进行角度采样得到所述驶出道路中的至少一个终点位姿向量；对所述至少一个起始点位姿向量和所述至少一个终点位姿向量进行曲线采样，得到所述驶入道路和所述驶出道路之间的多条曲线；根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束对所述多条曲线进行筛选处理，得到所述M条车道拓扑曲线。
该筛选处理例如可以是先筛选满足上述约束的曲线，然后再筛选达到每条驶入车道和每条驶出车道之间至多有一条最优曲线。通过对该最优曲线进行碰撞检测处理，以便适应性调整各曲线，进而得到所述M条车道拓扑曲线。该处理方式仅为一种示例，其还可以是其他方式，本方案对此不做具体限定。
该方案基于驶入车道的末端、驶出车道的起始点进行角度采样，进而生成多条车道拓扑曲线，基于上述得到的软、硬边界约束以及虚拟边界约束对该多条曲线进行筛选处理，进而得到M条车道拓扑曲线。采用该手段，保证了虚拟车道轨迹的类人性、灵活性以及车型普适性，减少与其他车道的车流冲突。
进一步地，所述对所述至少一个起始点位姿向量和所述至少一个终点位姿向量进行曲线采样，得到所述驶入道路和所述驶出道路之间的多条曲线，包括：在所述驶入道路的每条驶入车道的末端和所述驶出道路的每条驶出车道的起始点之间生成多个控制点；根据所述至少一个起始点位姿向量、所述至少一个终点位姿向量以及所述多个控制点，生成所述多条平滑曲线。
该方案在曲线生成时，不仅基于驶入车道的末端、驶出车道的起始点进行角度采样，而且还基于控制点采样，使得生成的曲线数量更多，提高了曲线生成的灵活性。
作为一种可选的实现方式，所述驶入道路中的至少一个起始点位姿向量是通过将所述每条驶入车道的末端延伸第二预设距离并进行采样得到的。
采用该手段，针对驶入、驶出环岛路口特点等场景，对虚拟路口侧轨迹始末姿态采样点向外侧进行合理延伸，提高生成轨迹的质量，提升轨迹类人性，避免高精度地图制图引起的轨迹不合理，也避免了拓扑筛选因轨迹生成不合理而错误筛除，保障道路拓扑分析的完备性。
作为一种可选的实现方式，所述对所述M条车道拓扑曲线进行合理性检测处理，以得到K’条车道拓扑曲线，包括：根据所述驶入道路的方向向量、驶出道路的方向向量得到所述驶入道路、驶出道路之间的投影线，所述投影线为所述驶入道路的方向向量、驶出道路的方向向量相交所得到的夹角的平分线所在的直线，或者，所述投影线为垂直于所述驶出道路的方向向量且通过所述驶出道路的起始点的直线；计算所述驶入道路的每条驶入车道和所述驶出道路的每条驶出车道之间的对齐系数，其中，所述每条驶入车道和所述每条驶出车道之间的对齐系数为第一参数与第二参数之间的比值，所述第一参数为所述每条驶入车道的车道边线以及所述每条驶出车道的车道边线分别延长至所述投影线所得到的两条线段之间的重叠长度，所述第二参数为所述每条驶入车道的车道边线和所述每条驶出车道的车道边线分别延长至所述投影线所得到的两条线段中最短的线段的长度；根据所述驶入道路的每条驶入车道和所述驶出道路的每条驶出车道之间的对齐系数，得到K’条车道拓扑曲线，其中，所述K’条车道拓扑曲线包括以对齐系数大于第一预设阈值的驶入车道的末端、驶出车道的起始点为端点的曲线。
通过基于车道对齐性来筛选曲线，进而得到完备且合理的车道拓扑曲线。
作为又一种可选的实现方式，所述K’条车道拓扑曲线包括分别以所述驶入道路的最左侧驶入车道的末端、所述驶出道路的最左侧驶出车道的起始点为端点的曲线，还包括分别以所述驶入道路的最右侧驶入车道的末端、所述驶出道路的最右侧驶出车道的起始点为端点的曲线。
通过基于拓扑补齐原则来对曲线进行补齐处理，进而得到完备且合理的车道拓扑曲线。
作为又一种可选的实现方式，所述K’条车道拓扑曲线包括以所述驶入道路中的驶入车道X的末端以及所述驶出道路中的驶出车道Y的左侧车道的起始点为端点的车道拓扑曲线，以及以所述驶入车道X的末端以及所述驶出车道Y的右侧车道的起始点为端点的车道拓扑曲线，还包括以所述驶入车道X的末端以及所述驶出车道Y的起始点为端点的车道拓扑曲线，其中，所述驶出车道Y的左侧和右侧均存在车道；或者，所述K’条车道拓扑曲线包括以所述驶入道路中的驶入车道X的末端以及所述驶出道路中的驶出车道Y的左侧车道或者右侧车道的起始点为端点的车道拓扑曲线，还包括以所述驶入道路中的驶入车道X的末端、所述驶出车道Y的起始点为端点的车道拓扑曲线，其中，所述驶出道路中的驶出车道Y仅左侧或者右侧存在车道。
通过基于拓扑补齐原则来对曲线进行补齐处理，进而得到完备且合理的车道拓扑曲线。
作为又一种可选的实现方式，所述K’条车道拓扑曲线中每条车道拓扑曲线的最大曲率不大于第二预设阈值，且，所述每条车道拓扑曲线与所述车道拓扑曲线软边界约束、所述车道拓扑曲线硬边界约束之间的距离不小于第三预设距离，且，任意两条车道拓扑曲线之间的距离不小于第四预设距离。
基于车辆运动学、碰撞检测、交通规则、车流干扰检测等筛选原则进行车道拓扑合理性筛选，进而得到完备且合理的车道拓扑曲线。
其中，所述路口包括交叉路口、环岛、待转区路口、小S弯、高架出入口、多车道且无 车道标线路段、连续转弯路口、窄道掉头路口中的至少一种。
第三方面，本申请提供了一种引导车辆行驶的装置，包括：曲线生成模块，用于根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及车道线生成M条车道拓扑曲线，所述车道拓扑曲线为以所述驶入道路的驶入车道的末端和所述驶出道路的驶出车道的起始点为端点的曲线；检测处理模块，用于对所述M条车道拓扑曲线进行合理性检测处理，以得到K’条车道拓扑曲线，其中，K’不大于M；确定模块，用于当所述车辆位于所述驶入道路中时，从所述K’条车道拓扑曲线中确定目标路径。
其中，所述曲线生成模块，用于：根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及不可跨越车道线得到车道拓扑曲线硬边界约束；根据所述路口、所述路口的驶入道路、所述路口的驶出道路中的可跨越车道线得到车道拓扑曲线软边界约束；根据所述车道拓扑曲线硬边界约束和所述车道拓扑曲线软边界约束得到K个车道拓扑曲线虚拟边界约束；根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束生成所述M条车道拓扑曲线。
可选的，所述K个车道拓扑曲线虚拟边界约束与K个车道拓扑对应，所述K个车道拓扑曲线虚拟边界约束中任一车道拓扑曲线虚拟边界约束A是通过将所述路口的最左侧车道拓扑左侧的硬边界约束和/或软边界约束向右平移第一预设距离，以及将所述路口的最右侧车道拓扑右侧的硬边界约束和/或软边界约束向左平移第一预设距离得到的，其中，所述第一预设距离是根据车道拓扑A’的车道位序确定的，或者，所述第一预设距离是根据所述车辆预设通过宽度、车道宽度中的至少一项以及所述车道拓扑A’的车道位序确定的，所述车道拓扑曲线虚拟边界约束A与所述车道拓扑A’对应，所述K个车道拓扑包括所述路口的最左侧车道拓扑和所述最右侧车道拓扑。
进一步地，所述曲线生成模块，还用于：分别对所述驶入道路中每条驶入车道的末端进行角度采样得到所述驶入道路中的至少一个起始点位姿向量，并分别对所述驶出道路中每条驶出车道的起始点进行角度采样得到所述驶出道路中的至少一个终点位姿向量；对所述至少一个起始点位姿向量和所述至少一个终点位姿向量进行曲线采样，得到所述驶入道路和所述驶出道路之间的多条曲线；根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束对所述多条曲线进行筛选处理，得到所述M条车道拓扑曲线。
可选的，所述曲线生成模块，还用于：在所述驶入道路的每条驶入车道的末端和所述驶出道路的每条驶出车道的起始点之间生成多个控制点；根据所述至少一个起始点位姿向量、所述至少一个终点位姿向量以及所述多个控制点，生成所述多条平滑曲线。
作为一种实现方式，所述驶入道路中的至少一个起始点位姿向量是通过将所述每条驶入车道的末端延伸第二预设距离并进行采样得到的。
其中，所述检测处理模块，用于：根据所述驶入道路的方向向量、驶出道路的方向向量得到所述驶入道路、驶出道路之间的投影线，所述投影线为所述驶入道路的方向向量、驶出道路的方向向量相交所得到的夹角的平分线所在的直线，或者，所述投影线为垂直于所述驶出道路的方向向量且通过所述驶出道路的起始点的直线；计算所述驶入道路的每条驶入车道和所述驶出道路的每条驶出车道之间的对齐系数，其中，所述每条驶入车道和所述每条驶出车道之间的对齐系数为第一参数与第二参数之间的比值，所述第一参数为所述每条驶入车道的车道边线以及所述每条驶出车道的车道边线分别延长至所述投影线所得到的两条线段之间的重叠长度，所述第二参数为所述每条驶入车道的车道边线和所述每条驶出车道的车道边线 分别延长至所述投影线所得到的两条线段中最短的线段的长度；根据所述驶入道路的每条驶入车道和所述驶出道路的每条驶出车道之间的对齐系数，得到K’条车道拓扑曲线，其中，所述K’条车道拓扑曲线包括以对齐系数大于第一预设阈值的驶入车道的末端、驶出车道的起始点为端点的曲线。
可选的，所述K’条车道拓扑曲线包括分别以所述驶入道路的最左侧驶入车道的末端、所述驶出道路的最左侧驶出车道的起始点为端点的曲线，还包括分别以所述驶入道路的最右侧驶入车道的末端、所述驶出道路的最右侧驶出车道的起始点为端点的曲线。
其中，所述K’条车道拓扑曲线包括以所述驶入道路中的驶入车道X的末端以及所述驶出道路中的驶出车道Y的左侧车道的起始点为端点的车道拓扑曲线，以及以所述驶入车道X的末端以及所述驶出车道Y的右侧车道的起始点为端点的车道拓扑曲线，还包括以所述驶入车道X的末端以及所述驶出车道Y的起始点为端点的车道拓扑曲线，其中，所述驶出车道Y的左侧和右侧均存在车道；或者，所述K’条车道拓扑曲线包括以所述驶入道路中的驶入车道X的末端以及所述驶出道路中的驶出车道Y的左侧车道或者右侧车道的起始点为端点的车道拓扑曲线，还包括以所述驶入道路中的驶入车道X的末端、所述驶出车道Y的起始点为端点的车道拓扑曲线，其中，所述驶出道路中的驶出车道Y仅左侧或者右侧存在车道。
可选的，所述K’条车道拓扑曲线中每条车道拓扑曲线的最大曲率不大于第二预设阈值，且，所述每条车道拓扑曲线与所述车道拓扑曲线软边界约束、所述车道拓扑曲线硬边界约束之间的距离不小于第三预设距离，且，任意两条车道拓扑曲线之间的距离不小于第四预设距离。
进一步地，所述装置还包括评价模块，用于：计算所述K’条车道拓扑曲线中每条车道拓扑曲线的评价值，所述评价值与所述车道拓扑曲线曲率、曲率变化率以及斜穿车道数量、以及所述车道拓扑曲线对应的车道的车道交汇信息、交通规则信息、车流量估计值、可行驶距离中的至少一项有关；所述确定模块，用于：当所述车辆位于所述驶入道路中时，根据所述K’条车道拓扑曲线中每条车道拓扑曲线的评价值确定所述目标路径。
其中，所述路口包括交叉路口、环岛、待转区路口、小S弯、高架出入口、多车道且无车道标线路段、连续转弯路口、窄道掉头路口中的至少一种。
第四方面，本申请提供了一种基于路口的地图生成装置，包括：曲线生成模块，用于根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及车道线生成M条车道拓扑曲线，所述车道拓扑曲线为以所述驶入道路的驶入车道的末端和所述驶出道路的驶出车道的起始点为端点的曲线；检测处理模块，用于对所述M条车道拓扑曲线进行合理性检测处理，以得到K’条车道拓扑曲线，其中，K’不大于M；地图生成模块，用于根据所述路口内的K’条车道拓扑曲线生成所述路口的地图。
其中，所述曲线生成模块，用于：根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及不可跨越车道线得到车道拓扑曲线硬边界约束；根据所述路口、所述路口的驶入道路、所述路口的驶出道路中的可跨越车道线得到车道拓扑曲线软边界约束；根据所述车道拓扑曲线硬边界约束和所述车道拓扑曲线软边界约束得到K个车道拓扑曲线虚拟边界约束；根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束生成所述M条车道拓扑曲线。
可选的，所述K个车道拓扑曲线虚拟边界约束与K个车道拓扑对应，所述K个车道拓扑曲线虚拟边界约束中任一车道拓扑曲线虚拟边界约束A是通过将所述路口的最左侧车道拓扑左侧的硬边界约束和/或软边界约束向右平移第一预设距离，以及将所述路口的最右侧车道 拓扑右侧的硬边界约束和/或软边界约束向左平移第一预设距离得到的，其中，所述第一预设距离是根据车道拓扑A’的车道位序确定的，或者，所述第一预设距离是根据所述车辆预设通过宽度、车道宽度中的至少一项以及所述车道拓扑A’的车道位序确定的，所述车道拓扑曲线虚拟边界约束A与所述车道拓扑A’对应，所述K个车道拓扑包括所述路口的最左侧车道拓扑和所述最右侧车道拓扑。
进一步地，所述曲线生成模块，还用于：分别对所述驶入道路中每条驶入车道的末端进行角度采样得到所述驶入道路中的至少一个起始点位姿向量，并分别对所述驶出道路中每条驶出车道的起始点进行角度采样得到所述驶出道路中的至少一个终点位姿向量；对所述至少一个起始点位姿向量和所述至少一个终点位姿向量进行曲线采样，得到所述驶入道路和所述驶出道路之间的多条曲线；根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束对所述多条曲线进行筛选处理，得到所述M条车道拓扑曲线。
可选的，所述曲线生成模块，还用于：在所述驶入道路的每条驶入车道的末端和所述驶出道路的每条驶出车道的起始点之间生成多个控制点；根据所述至少一个起始点位姿向量、所述至少一个终点位姿向量以及所述多个控制点，生成所述多条平滑曲线。
作为一种实现方式，所述驶入道路中的至少一个起始点位姿向量是通过将所述每条驶入车道的末端延伸第二预设距离并进行采样得到的。
其中，所述检测处理模块，用于：根据所述驶入道路的方向向量、驶出道路的方向向量得到所述驶入道路、驶出道路之间的投影线，所述投影线为所述驶入道路的方向向量、驶出道路的方向向量相交所得到的夹角的平分线所在的直线，或者，所述投影线为垂直于所述驶出道路的方向向量且通过所述驶出道路的起始点的直线；计算所述驶入道路的每条驶入车道和所述驶出道路的每条驶出车道之间的对齐系数，其中，所述每条驶入车道和所述每条驶出车道之间的对齐系数为第一参数与第二参数之间的比值，所述第一参数为所述每条驶入车道的车道边线以及所述每条驶出车道的车道边线分别延长至所述投影线所得到的两条线段之间的重叠长度，所述第二参数为所述每条驶入车道的车道边线和所述每条驶出车道的车道边线分别延长至所述投影线所得到的两条线段中最短的线段的长度；根据所述驶入道路的每条驶入车道和所述驶出道路的每条驶出车道之间的对齐系数，得到K’条车道拓扑曲线，其中，所述K’条车道拓扑曲线包括以对齐系数大于第一预设阈值的驶入车道的末端、驶出车道的起始点为端点的曲线。
可选的，所述K’条车道拓扑曲线包括分别以所述驶入道路的最左侧驶入车道的末端、所述驶出道路的最左侧驶出车道的起始点为端点的曲线，还包括分别以所述驶入道路的最右侧驶入车道的末端、所述驶出道路的最右侧驶出车道的起始点为端点的曲线。
其中，所述K’条车道拓扑曲线包括以所述驶入道路中的驶入车道X的末端以及所述驶出道路中的驶出车道Y的左侧车道的起始点为端点的车道拓扑曲线，以及以所述驶入车道X的末端以及所述驶出车道Y的右侧车道的起始点为端点的车道拓扑曲线，还包括以所述驶入车道X的末端以及所述驶出车道Y的起始点为端点的车道拓扑曲线，其中，所述驶出车道Y的左侧和右侧均存在车道；或者，所述K’条车道拓扑曲线包括以所述驶入道路中的驶入车道X的末端以及所述驶出道路中的驶出车道Y的左侧车道或者右侧车道的起始点为端点的车道拓扑曲线，还包括以所述驶入道路中的驶入车道X的末端、所述驶出车道Y的起始点为端点的车道拓扑曲线，其中，所述驶出道路中的驶出车道Y仅左侧或者右侧存在车道。
可选的，所述K’条车道拓扑曲线中每条车道拓扑曲线的最大曲率不大于第二预设阈值， 且，所述每条车道拓扑曲线与所述车道拓扑曲线软边界约束、所述车道拓扑曲线硬边界约束之间的距离不小于第三预设距离，且，任意两条车道拓扑曲线之间的距离不小于第四预设距离。
其中，所述路口包括交叉路口、环岛、待转区路口、小S弯、高架出入口、多车道且无车道标线路段、连续转弯路口、窄道掉头路口中的至少一种。
第五方面，本申请提供了一种引导车辆行驶的装置，包括处理器和存储器；其中，所述存储器用于存储程序代码，所述处理器用于调用所述程序代码，以执行所述第一方面任一种可能的实施方式提供的方法。
第六方面，本申请提供了一种基于路口的地图生成装置，包括处理器和存储器；其中，所述存储器用于存储程序代码，所述处理器用于调用所述程序代码，以执行所述第二方面任一种可能的实施方式提供的方法。
第七方面，本申请提供了一种计算机可读存储介质，所述计算机可读存储介质存储有计算机程序，所述计算机程序被处理器执行以实现第一方面中任一所提供的方法和/或第二方面中任一所提供的方法。
第八方面，本申请提供了一种计算机程序产品，当计算机程序产品在计算机上运行时，使得所述计算机执行第一方面中任一所提供的方法和/或第二方面中任一所提供的方法。
第九方面，本申请提供了一种芯片系统，所述芯片系统应用于电子设备；所述芯片系统包括一个或多个接口电路，以及一个或多个处理器；所述接口电路和所述处理器通过线路互联；所述接口电路用于从所述电子设备的存储器接收信号，并向所述处理器发送所述信号，所述信号包括所述存储器中存储的计算机指令；当所述处理器执行所述计算机指令时，所述电子设备执行第一方面中任一所提供的方法和/或第二方面中任一所提供的方法。
第十方面，本申请提供了一种智能驾驶车辆，其特征在于，包括行进系统、传感系统、控制系统和计算机系统，其中，所述计算机系统用于执行第一方面中任一所提供的方法和/或第二方面中任一所提供的方法。
可以理解地，上述提供的第三方面所述的装置、第四方面所述的装置、第五方面所述的装置、第六方面所述的装置、第七方面所述的计算机存储介质或者第八方面所述的计算机程序产品、第九方面所述的芯片系统以及第十方面所述的智能驾驶车辆均用于执行第一方面中任一所提供的方法以及第二方面中任一所提供的方法。因此，其所能达到的有益效果可参考对应方法中的有益效果，此处不再赘述。
附图说明
下面对本申请实施例用到的附图进行介绍。
图1是本申请实施例提供的一种引导车辆行驶的系统架构示意图；
图2是本申请实施例提供的一种引导车辆行驶的方法的流程示意图；
图3是本申请实施例提供的一种车道拓扑示意图；
图4是本申请实施例提供的一种边界约束示意图；
图5是本申请实施例提供的一种曲线生成方法示意图；
图6是本申请实施例提供的一种对齐系数求解方法示意图；
图7是本申请实施例提供的一种车道拓扑曲线筛选示意图；
图8a是本申请实施例提供的一种待转区的异形多对多直行路口场景示意图；
图8b是本申请实施例提供的一种边界约束示意图；
图8c是本申请实施例提供的一种曲线筛选示意图；
图8d是本申请实施例提供的第一种重叠长度示意图；
图8e是本申请实施例提供的第二种重叠长度示意图；
图8f是本申请实施例提供的第一种应用场景示意图；
图8g是本申请实施例提供的第二种应用场景示意图；
图8h是本申请实施例提供的第三种应用场景示意图；
图9a是本申请实施例提供的一种环岛场景示意图；
图9b是本申请实施例提供的一种边界约束示意图；
图9c是本申请实施例提供的一种曲线生成示意图；
图9d是本申请实施例提供的一种曲线筛选示意图；
图10a是本申请实施例提供的一种有待转区、红绿灯的左转路口场景示意图；
图10b是本申请实施例提供的又一种有待转区、红绿灯的左转路口场景示意图；
图11a是本申请实施例提供的一种小S弯场景示意图；
图11b是本申请实施例提供的又一种小S弯场景示意图；
图12是本申请实施例提供的一种匝道场景示意图；
图13是本申请实施例提供的一种多车道无车道标线路段场景示意图；
图14是本申请实施例提供的一种连续转弯场景示意图；
图15是本申请实施例提供的一种窄道掉头场景示意图；
图16是本申请实施例提供的一种引导车辆行驶的装置的结构示意图；
图17是本申请实施例提供的另一种引导车辆行驶的装置的结构示意图。
具体实施方式
下面结合本申请实施例中的附图对本申请实施例进行描述。本申请实施例的实施方式部分使用的术语仅用于对本申请的具体实施例进行解释，而非旨在限定本申请。
参照图1所示，是本申请实施例提供的一种引导车辆行驶的系统架构示意图。该系统可包括：车道拓扑生成模块、边界约束生成模块、拓扑曲线生成模块、拓扑合理性筛选模块以及实时决策模块。其中，车道拓扑生成模块用于结合当前场景的道路结构生成车道级全连接拓扑，为构建完备的道路拓扑空间奠定基础。边界约束生成模块用于基于高精度地图和传感器感知数据为车道级全连接拓扑中的每条车道拓扑生成软边界约束、硬边界约束和虚拟边界约束。拓扑曲线生成模块用于为上述车道拓扑生成满足上述边界约束且平滑类人的车道拓扑曲线。拓扑合理性筛选模块用于基于各种类人性筛选原则，对车道级全连接拓扑及其曲线进行合理性筛选，删除不合理的车道拓扑，从而得到完备且合理的车道拓扑曲线集合，以完成当前场景的道路拓扑分析。实时决策模块用于基于上述道路拓扑分析、动静态交通环境和自车实时状态等，选择当前时刻的最优目标路径。
进一步地，上述系统还可以包括导航推荐评价模块。导航推荐评价模块引入车道拓扑推荐函数，对上述车道拓扑曲线集合中的各车道拓扑曲线进行导航优先级评价。相应地，实时决策模块还可以基于该导航优先级评价来选择最优目标路径。
上述系统仅为一种示例，其中，该系统还可以仅包含边界约束生成模块、拓扑曲线生成模块、拓扑合理性筛选模块以及实时决策模块等，本方案对此不做具体限定。
本方案可应用于自动驾驶车辆在开放道路上行驶，行驶范围内包含不存在实际车道线或 存在多种合理驾驶轨迹的道路场景时，需要进行合理的道路拓扑分析并提供相应的拓扑导航引导信息，以供车辆进行意图预测、轨迹(曲线)预测、车道决策和运动规划。上述道路场景包括但不限于交叉路口、环岛、待转区路口、小S弯、高架出入口、多车道且无车道标线路段、连续转弯路口、窄道掉头路口等。当然，其还可以是其他场景，本方案对此不做具体限定。
上述仅以本申请实施例应用于自动驾驶场景为例进行说明。其中，本申请提供的引导车辆行驶的方法，还可以应用于辅助驾驶场景，本方案对此不做具体限定。
本实施例可以由车载装置(如车机)来执行，其还可以由手机、电脑等终端设备来执行。本方案对此不做具体限定。
需要说明的是，本申请提供的引导车辆行驶的方法，可以在本地执行，也可以由云端执行。其中，云端可以由服务器来实现，该服务器可以是虚拟服务器、实体服务器等，其还可以是其他装置，本方案对此不做具体限定。
参照图2所示，为本申请实施例提供的一种引导车辆行驶的方法的流程示意图。如图2所示，该方法包括步骤201-203，具体如下：
201、根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及车道线生成M条车道拓扑曲线，所述车道拓扑曲线为以所述驶入道路的驶入车道的末端和所述驶出道路的驶出车道的起始点为端点的曲线；
其中，可根据高精度地图和传感器获取的环境感知信息中所述路口、所述驶入道路、所述驶出道路中的障碍物以及车道线来生成M条车道拓扑曲线。
例如，基于上述障碍物以及车道线得到车道拓扑曲线硬边界约束、车道拓扑曲线软边界约束和车道拓扑曲线虚拟边界约束，进而基于上述各约束来生成所述M条车道拓扑曲线。
具体地，上述步骤201可包括步骤2011-2014：
2011、根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及不可跨越车道线得到车道拓扑曲线硬边界约束；
该车道拓扑曲线硬边界约束可以理解为，车辆不能行驶的边界。
具体地，获取路口场景附近的静态障碍物以及不可跨越的车道线等，即得到车道拓扑曲线硬边界约束。
上述静态障碍物包括道路上的马路牙子、安全岛、绿化带等。不可跨越的车道线例如车道实线、导流线等。
2012、根据所述路口、所述路口的驶入道路、所述路口的驶出道路中的可跨越车道线得到车道拓扑曲线软边界约束；
该车道拓扑曲线软边界约束可以理解为，交通规则中允许自车跨越，而自车最好不跨越的边界约束。
具体地，通过获取路口场景中的地面实际标线即得到软边界约束。该地面实际标线例如待转区车道线、可跨越的车道线等。
2013、根据所述车道拓扑曲线硬边界约束和所述车道拓扑曲线软边界约束得到K个车道拓扑曲线虚拟边界约束；
对于多车道平行车流场景，不仅最近邻车道的车流轨迹会受到上述软边界、硬边界的直接影响，次邻车道的车流轨迹也会由于最近邻车流轨迹干涉而受到上述软边界、硬边界的间接影响。因此，考虑邻近车道中车流轨迹的干涉影响，以及软硬边界对同向车道逐渐减弱的 间接约束作用，生成虚拟边界约束。
其中，上述K个车道拓扑曲线虚拟边界约束与K个车道拓扑对应。
具体地，根据路口、所述路口的驶入道路、所述路口的驶出道路中的各车道的拓扑结构得到所述路口的K个车道拓扑。
通过获取当前场景中的道路拓扑结构，具体地，其包括路口、驶入道路以及驶入道路中的驶入车道、驶出道路以及驶出道路中的驶出车道，以及无标线区域等道路元素及其拓扑关系；然后可基于当前场景相关的交通规则，根据所有可通行的K1条驶入车道和K2条驶出车道，生成该场景下所有可能的K个车道级全连接拓扑，其中，K≤K1*K2。
如图3所示，为本申请实施例提供的一种车道拓扑示意图。该场景中包括驶入道路的3条驶入车道和驶出道路的4条驶出车道。基于该3条驶入车道和4条驶出车道生成了12个车道级全连接拓扑，即12条虚拟车道。需要说明的是，图中所示12个车道级全连接拓扑的线段仅用于表征车道之间的前后续接关系，并不表示虚拟车道的最终曲线(轨迹形态)。
其中，交通规则有可能会对当前场景下车道拓扑全连接关系及其数量产生影响。例如：由于公交车道、潮汐车道等特殊车道的分时通行特性，在其可通行时段只有符合特殊通行规则的车道拓扑是合理的，可生成相应的车道拓扑，因此K1条驶入车道和K2条驶出车道的车道拓扑全连接数量K可能小于K1*K2。
具体地，上述K个车道拓扑曲线虚拟边界约束中任一车道拓扑曲线虚拟边界约束A是通过将所述路口的最左侧车道拓扑左侧的硬边界约束和/或软边界约束向右平移第一预设距离，以及将所述路口的最右侧车道拓扑右侧的硬边界约束和/或软边界约束向左平移第一预设距离得到的。
该路口的最左侧车道拓扑即为左一驶入车道对左一驶出车道。该路口的最右侧车道拓扑即为右一驶入车道对右一驶出车道。
其中，上述第一预设距离可以是根据车道拓扑A’的车道位序确定的。
可替代的，上述第一预设距离可以是根据所述车辆预设通过宽度、车道宽度中的至少一项以及所述车道拓扑A’的车道位序确定的。
该车辆预设通过宽度可以理解为车辆安全通过时所需的宽度。
上述车道位序可以理解为车道所在的顺序。该车道位序可以基于预设规则进行设置，例如取驶入车道序号和驶出车道序号中的最小值，然后将该最小值再减1，即得到上述车道位序。
具体地，从左开始计起，左一驶入车道对左一驶出车道的车道位序为0；左一驶入车道对左二驶出车道的右起车道位序为0；右二驶入车道对右三驶出车道的右起车道位序为1。
上述仅为一种示例，其还可以是其他确定方式，本方案对此不做具体限定。
其中，上述预设距离可根据所述每条车道拓扑的车道位序确定。例如，车道位序越大，其对应的预设距离越大。
可替代的，上述预设距离还可以根据所述车辆安全通过宽度、车道宽度中的至少一项以及所述每条车道拓扑的车道位序确定。
具体地，上述平移距离
Figure PCTCN2022138461-appb-000001
可基于如下方式得到：
Figure PCTCN2022138461-appb-000002
其中，n lane表示当前车道拓扑的车道位序；W lane表示车道宽度；W vehicle表示车辆预设(安全)通过宽度；
Figure PCTCN2022138461-appb-000003
表示碰撞深度，该碰撞深度为车道拓扑曲线与障碍物边界曲线之间最小距离位置、指向靠近障碍物方向的向量。
上述平移距离对应的平移方向为软硬边界与参考拓扑曲线的碰撞深度的反方向。如图4所示，上述参考拓扑轨迹包括左一驶入车道对左一驶出车道所对应的平滑曲线S1、左二驶入车道对左二驶出车道所对应的平滑曲线S2等。其中，图4中阴影区域表示马路牙子、花坛、安全岛等，该阴影区域所示即为各车道的硬边界约束。待转区车道的车道线L1和车道线L2即为各车道的软边界约束。参考拓扑曲线S1对应的箭头t所指方向即为碰撞深度方向，虚线区域U1、U2和U3以及虚线L3即为路口的最左侧车道拓扑左侧的硬边界约束(即阴影区域)和软边界约束(L2)沿碰撞深度反方向平移一定距离得到的次临近车道的虚拟边界约束。
2014、根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束生成所述M条车道拓扑曲线。
基于上述硬边界约束、软边界约束和虚拟边界约束、车辆物理属性、车辆运动性能等因素，为车道级全连接拓扑中的每条车道拓扑生成曲率平滑且类人的车道拓扑曲线。
可选的，步骤2014可包括步骤20141-20143，具体如下：
20141、分别对所述驶入道路中每条驶入车道的末端进行角度采样得到所述驶入道路中的至少一个起始点位姿向量，并分别对所述驶出道路中每条驶出车道的起始点进行角度采样得到所述驶出道路中的至少一个终点位姿向量；
其中，可以采用基于贝塞尔曲线的采样算法、基于Spiral曲线的优化算法等生成上述车道拓扑曲线。
该实施例以基于Spiral曲线的优化算法为例进行说明。如图5所示，基于驶入车道的末端、驶出车道的起始点进行角度采样可得到分别以驶入车道的末端、驶出车道的起始点为起点的不同姿态角度的向量。也就是说，上述起始点位姿向量可以理解为，位于驶入车道的末端且具有不同姿态角度的向量。如图5所示，驶入车道的末端A处的多个不同方向的向量
Figure PCTCN2022138461-appb-000004
即为上述起始点位姿向量。相应地，上述终点位姿向量可以理解为，位于驶出车道的起点且具有不同姿态角度的向量。如图5所示，驶出车道的起始点B处的多个不同方向的向量
Figure PCTCN2022138461-appb-000005
即为上述终点位姿向量。
20142、对所述至少一个起始点位姿向量和所述至少一个终点位姿向量进行曲线采样，得到所述驶入道路和所述驶出道路之间的多条曲线；
作为第一种实现方式，任意一个起始点位姿向量和任意一个终点位姿向量进行组合，进而基于上述多个组合可得到多条平滑曲线。
作为另一种实现方式，在驶入车道的末端和驶出车道的起点的连线上进行距离采样，生成多组中间控制点，如图5中点P i所示。该距离采样，可以是间隔预设距离生成若干控制点，当然还可以是采用其他方式，本方案对此不做具体限定。通过将起始点位姿向量、终点位姿向量以及控制点进行组合，进而生成多条曲率平滑的Spiral曲线，如图5中虚线所示的驶入车道的末端和驶出车道的起始点之间的各曲线。
当然，还可以是基于其他方式来得到多条平滑的曲线，本方案对此不做具体限定。
20143、根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束对所述多条曲线进行筛选处理，得到所述M条车道拓扑曲线。
作为第一种实现方式，直接删除不满足上述车道拓扑曲线硬边界约束、软边界约束和虚拟边界约束的曲线，即得到上述M条车道拓扑的车道拓扑曲线。
作为第二种实现方式，通过考虑上述生成的曲线的曲率、曲率变化率、到软硬边界的碰撞代价、通行空间、曲线长度(通行效率)等因素，构建曲线评价函数。基于该曲线评价函数，从上述多条平滑曲线中选取满足安全性边界约束和车辆性能约束的最优曲线作为从驶入 车道末端到驶出车道起始点的最优曲线。如图5中驶入车道的末端和驶出车道的起始点之间的曲线S3，即为一条最优曲线。其中，相同起止点之间的曲线越短，沿着相应曲线行驶的车流行驶距离越短，通行效率越高。通过选取相同起止点之间较短的曲线，并通过对最优曲线进行碰撞检测，若与场景中的边界障碍物发生碰撞，则根据碰撞位置和碰撞深度增加碰撞控制点，以局部调整碰撞位置附近的轨迹形态，从而得到安全无碰且曲率平滑的车道拓扑曲线S4。
上述仅为一种示例，其还可以是其他处理方式，本方案对此不做具体限定。
202、对所述M条车道拓扑曲线进行合理性检测处理，以得到K’条车道拓扑曲线，其中，K’不大于M；
考虑车道拓扑轨迹的几何形态、车辆运动学、交通规则、碰撞检测、车流干扰等因素对车道拓扑合理性和曲线形态类人性的影响，可建立多项车道拓扑曲线合理性筛选原则。通过对上述车道级全连接拓扑及其曲线进行合理性筛选，删除不合理的车道拓扑，从而得到完备且合理的车道拓扑曲线集合，即完成了对当前场景的道路拓扑分析。
本方案以如下筛选原则为例进行介绍：
(1)拓扑投影筛选原则
拓扑投影筛选原则可以理解为，通过获取到驶入车道和驶出车道之间的对齐系数(重叠系数)，若该对齐系数不小于预设阈值，说明该驶入车道和该驶出车道之间具备对齐特性，保留该车道拓扑及其曲线；若该对齐系数小于预设阈值，说明该驶入车道和该驶出车道之间不具备对齐特性，删除该车道拓扑及其曲线。
下面介绍本实施例提供的获取对齐系数的方式。
首先获取驶入车道和驶出车道之间的投影线。如图6所示，将驶入车道的方向向量和驶出车道的方向向量进行前后延伸，若其交点位于路口的驶入车道末端和驶出车道起点之间，则以通过该交点的角平分线所在的直线作为投影线，如图6中投影线L3所示。该驶入车道的方向向量可以理解为驶入道路的方向向量，即与驶入车道平行且与驶入车道对应的行驶方向一致的向量。
其中，若上述交点不位于路口的驶入车道末端和驶出车道起点之间，则以垂直于驶出道路的方向向量且通过所述驶出道路的起点的直线作为投影线，即图6中直线L4所示。
然后，将驶入车道和驶出车道的车道线均延伸至上述投影线L3，计算驶入车道左右车道边线和驶出车道左右车道边线在投影线上的重叠长度，并获取所述驶入车道左右车道边线和驶出车道左右车道边线分别延长至所述投影线所得到的两条线段中最短的线段的长度。通过计算该重叠长度与该最短的线段长度之间的比值即为对齐系数。
如图6所示，若计算直行驶入车道左一和驶出车道左二之间的对齐系数，则将直行驶入车道左一的两条车道边线延长至投影线L3，得到线段C1C2；将驶出车道左二的两条车道边线延长至投影线L3，得到线段D1D2；获取线段C1C2和线段D1D2之间的重叠长度L，以及线段C1C2和线段D1D2中最短的线段长度L min，则上述重叠长度L与最短的线段长度L min之间的比值即为直行驶入车道左一和驶出车道左二之间的对齐系数。
基于上述方法，同样地可以计算其他车道拓扑之间的对齐系数。
通过将上述求得的对齐系数与预设阈值进行比较，进而对车道拓扑进行筛选。
(2)运动学筛选原则
通过计算车道级全连接拓扑中每条车道拓扑曲线的最大曲率，若该曲率对应的最小转弯半径小于根据车辆运动学模型得到的车辆最小转弯半径，则可认为车辆在行驶过程中很难实 现沿着该车道拓扑曲线行驶，因此应当从车道级全连接拓扑中筛除该不合理的车道拓扑曲线。
上述不合理的车道拓扑曲线，例如小半径的左/右转曲线，小半径的掉头曲线等。如图7中运动学筛选对应的曲线所示，该经过左转待转区的左转车道拓扑曲线存在接近90°的转弯角度，该处曲率过大，不满足车辆运动学约束，车辆无法按照该曲线行驶，因此删除该车道拓扑曲线。
(3)交通规则筛选原则
由于公交车道、潮汐车道等特殊车道的分时通行特性、待转区在红绿灯状态下的不同拓扑合理性特性、停靠车道特殊行驶逻辑等，需要根据交通规则从车道级全连接拓扑中删除不符合交通通行规则的车道拓扑，并从驶入道路和驶出道路的所有车道中删除不符合当前通行方向的车道。例如直行路口中，删除路口前的左转、掉头、右转驶入车道，仅保留可直行的车道。
(4)碰撞检测筛选原则
对于车道级全连接拓扑中每条车道拓扑曲线，遍历整条曲线上的所有点，若曲线上任一点与当前场景中的硬边界距离过近(小于预设碰撞安全距离)，则整条曲线的状态为存在碰撞。
可选的，若存在碰撞，则根据碰撞位置和碰撞深度增加碰撞控制点，以局部调整碰撞位置附近的轨迹形态，进行局部曲线调整。
若局部曲线调整失败，则可视为该车道拓扑曲线生成失败，无法得到安全无碰的曲线，因此应当从车道级全连接拓扑中筛除该不合理的车道拓扑曲线。
如图7中碰撞检测对应的曲线所示，最右侧的直行车道拓扑曲线与右侧障碍物(阴影区域)碰撞，需进行局部轨迹调整。
(5)车流干扰筛选原则
考虑整个场景中不同道路方向、不同车道之间的车道级全连接拓扑中可同时通行所有车道拓扑曲线，若任意两条车道拓扑曲线之间距离过近，例如小于预设相邻车道干扰距离，则该车道拓扑曲线的状态为存在车流干扰，其中该相邻车道干扰距离一般小于车道宽度，略大于车体宽度。
若存在车流干扰，则根据相互干扰的车道拓扑曲线及其相邻车道曲线之间的间距分布，对车道曲线间距进行局部调整，使得相邻各车道曲线之间的间距分布更合理，不存在车流干扰现象。
若车道拓扑曲线间距局部调整失败，则可视为该场景中无法同时存在多条无干扰的车道拓扑曲线，因此应当从车道级全连接拓扑中筛除一条或多条互相干扰的车道拓扑曲线。
如图7中车流干扰对应的曲线所示，该场景中直行车道拓扑曲线与左转待转车道拓扑曲线是不同道路方向、不同车道之间可同时通行的两条车道拓扑曲线，但这两条车道拓扑曲线在待转区出口附近距离过近，将导致直行车辆和左转待转车流距离过近，相互干涉并产生侧向挤压，因此需将这两条车道拓扑曲线的间距进行局部调整，具体可将直行车道拓扑及其右侧同向车道拓扑进行适当右移，从而保证与左转待转车道拓扑没有车流干涉的影响。
(6)拓扑补齐原则
左对齐补充原则：检查驶入道路左一车道和驶出道路左一车道之间的拓扑是否被上述筛选原则删除。若驶入道路左一车道和驶出道路左一车道之间的拓扑被上述筛选原则删除，则补回该拓扑，相应地，其对应的曲线也补回。
右对齐补充原则：检查驶入道路右一车道和驶出道路右一车道之间的拓扑是否被上述筛选原则删除。若驶入道路右一车道和驶出道路右一车道之间的拓扑被上述筛选原则删除，则 补回该拓扑，相应地，其对应的曲线也补回。
驶出拓扑补充原则：当驶出道路拓扑存在时，应保证驶出道路中的所有驶出车道均有车道拓扑曲线；若某车道没有，则应就近补充新车道拓扑，保证驶出拓扑的完备性。
相邻车道补充原则：若某条驶出车道与驶入车道x之间没有车道拓扑曲线，但该条驶出车道的左、右侧车道中均有车道与驶入车道x存在车道拓扑曲线，则为该驶出车道就近补充新车道拓扑，保证相邻车道拓扑的合理性。其中，若该驶出车道为驶出道路左一车道，则仅考虑其右侧车道即可；若该驶出车道为驶出道路右一车道，则仅考虑其左侧车道即可。上述左、右侧车道，可以是与该车道相邻的车道，也可以是间隔的车道，本方案对此不做限定。
通过上述筛选原则、补充原则等对车道级全连接拓扑进行曲线生成、拓扑合理性筛选和补充后，即可得到完备且合理的车道拓扑曲线集合，从而完成了对当前场景的道路拓扑分析。
需要说明的是，上述各原则仅为一种示例，具体实现时可任意选择或者组合。当然还可以是基于其他原则进行处理，本方案对此不做具体限定。
203、当所述车辆位于所述驶入道路中时，从所述K’条车道拓扑曲线中确定目标路径。
例如，若车辆从驶入道路中的第一驶入车道进入路口时，则从第一驶入车道对应的车道拓扑曲线中确定车道拓扑曲线，即为目标路径。
若车辆进入驶入道路时，则可从K’条车道拓扑曲线中任意选择车道拓扑曲线作为目标路径，还可以基于实时交通情况来选择最优车道拓扑曲线作为目标路径。
需要说明的是，当车辆位于路口中时，还可以基于交通情况等进一步从所述K’条车道拓扑曲线中再次确定最优曲线。例如，当从最优车道拓扑曲线S进入路口后，由于被其他车辆占道等，则实时确定其他最优曲线。
本方案对此不做具体限定。
其中，在步骤203之前，所述方法还可包括：
计算所述K’条车道拓扑曲线中每条车道拓扑曲线的评价值，所述评价值与所述车道拓扑曲线曲率、曲率变化率以及斜穿车道数量、以及所述车道拓扑曲线对应的车道的车道交汇信息、交通规则信息、车流量估计值、可行驶距离中的至少一项有关。
基于上述道路拓扑分析、全局导航信息、交通规则、人类驾驶经验等信息，设计车道拓扑推荐评价函数，对上述车道拓扑曲线集合中的各车道拓扑曲线进行导航优先级评价，以表示在无其他动态车流干扰的交通场景下同簇车道拓扑(由同一条驶入车道进入该场景的多条车道拓扑曲线)中各条曲线的导航推荐优先级，用于选取综合最优的车道拓扑曲线。
可选的，车道拓扑推荐评价函数可表示为：
C＝w1*C1+w2*C2+w3*C3+w4*C4+w5*C5+w6*C6；
其中，C1、C2、C3、C4、C5和C6分别表示导航代价、拓扑代价、平滑性代价、车流交汇代价、交通规则代价和通行效率代价，w1、w2、w3、w4、w5、w6均为系数。
上述导航代价是基于全局导航信息，对该车道拓扑在当前场景下的目标可达性进行评价。若该车道拓扑所在的车道级路径规划在到达自动驾驶任务指定终点方向的可行驶距离越长，则同簇车道拓扑中该车道拓扑的导航代价C1越低，w1是导航代价在总代价中所占权重。
上述拓扑代价是根据车道拓扑连接的前后驶入车道和驶出车道的物理位置关系，对该车道拓扑在当前场景下的类人性进行评价。若该车道拓扑连接的前后驶入车道和驶出车道向左或向右横跨多条车道进行斜穿，会导致该车道拓扑轨迹不类人，提升了与他车抢道的风险，所以该车道拓扑的斜穿车道数越少，其拓扑代价C2越低，w2是拓扑代价在总代价中所占权重。
上述平滑性代价是该车道拓扑曲线由当前场景下驶入车道轨迹、虚拟车道轨迹、驶出车道轨迹共同构成，对该车道拓扑曲线的曲率和曲率变化率进行评价，曲率和曲率变化率越小，说明该车道拓扑曲线越平滑，其平滑性代价C3越低，w3是平滑性代价在总代价中所占权重。
上述车流交汇代价是根据该车道拓扑连接的驶出车道的所有车道拓扑连接关系，对该车道拓扑在当前场景下与其他交通流交汇的属性进行评价。若该车道拓扑的驶出车道同时还属于其他方向车道拓扑，例如：该直行车道拓扑的驶出车道也是其他左转、掉头、右转车道拓扑的驶出车道，则意味着该车道拓扑将与其他方向车辆进行交汇并发生抢道或侧向挤压的风险更高，因此车辆交互关系将更复杂，通行效率更低，车道交汇代价C4越高，w4是车流交汇代价在总代价中所占权重。
在左转、右转、直行、掉头等不同交通场景中，交通规则对车道拓扑的选择有不同倾向性。例如，左转、右转、掉头场景中，按照交通规则倾向于选择内侧车道拓扑，而在直行场景中则更倾向于选择更平直的车道拓扑，以避免与其他方向车流发生复杂交汇。因此，交通规则代价是表征交通规则对当前场景下车道拓扑的倾向优先级的代价，交通规则倾向优先级越高的车道拓扑，其交通规则代价C5越低，w5是交通规则代价在总代价中所占权重。
考虑同一条驶入车道有多条驶出车道的同簇车道拓扑场景中，人类司机往往能根据驾驶经验选择通行效率更高的车道拓扑，从宏观角度看即表现为同簇车道场景中宏观交通流的车流分配比例。因此，宏观交通流中车流占比越高的车道拓扑，意味着越接近人类司机的选择，通行效率也越高，其通行效率代价C6应越低，w6是通行效率代价在总代价中所占权重。
由此，对上述车道拓扑曲线集合中的各车道曲线进行总代价评价，可表示对同一条驶入车道相关的同簇车道拓扑中各条曲线的导航推荐优先级，用于选取综合最优的车道拓扑。例如，拓扑更平直、轨迹更平滑、无车道交汇、通行效率更高的车道拓扑则是导航最优推荐车道拓扑曲线。
通过计算每条车道拓扑曲线的评价值，进而从中选取评价值最优的车道拓扑曲线作为目标路径。
示例性地，当车辆接近如多对多交叉路口、待转区路口、S弯、高架入口等场景前，依靠导航推荐评价具备的全局视野，车辆可选择导航推荐最优的车道拓扑曲线，从而提前换道，以规避车道斜穿、他车侧向挤压、多车道合并等车流冲突的高风险车道拓扑。具体地，对于多对多交叉路口，最优推荐拓扑对齐更平直、可行驶距离更长的车道拓扑曲线。对于待转区路口，最优推荐进入待转区后的车道拓扑曲线。对于S弯，最优推荐沿车道线的车道拓扑曲线。对于左/右转、掉头场景，最优推荐满足运动学的、最内侧的车道拓扑曲线。
当车辆处于多对多交叉路口、待转区路口、S弯、高架入口等场景中时，若传感器检测到实时交通流已占用或挤压导航推荐最优的车道拓扑曲线，车辆可综合各车道拓扑的导航推荐评价和实时动态风险，选择当前场景下最优车道拓扑，例如可能是导航推荐的次优车道拓扑曲线等。具体地，对于多对多交叉路口，最优推荐侧向车流干扰更少、可行驶距离更长的车道拓扑曲线。对于S弯，最优推荐切弯的车道拓扑曲线。对于待转区路口，最优推荐不进入待转区的内切的车道拓扑曲线。对于左/右转、掉头场景，最优推荐满足运动学的、无占用的内侧的车道拓扑曲线。
本申请实施例，通过根据路口、路口的驶入道路、路口的驶出道路中的障碍物以及车道线生成M条车道拓扑曲线，通过对该M条车道拓扑曲线进行合理性检测与处理，得到K’条车道拓扑曲线，进而从该K’条车道拓扑曲线中确定目标路径，以引导车辆行驶。采用该手段，通过生成完备合理的车道拓扑曲线，并为车辆提供驾驶引导信息，有效提升车辆通过该路口 的轨迹类人性和通行效率。
实施例一
参照图8a所示，为本申请实施例提供的一种待转区的异形多对多直行路口场景。如图8a所示，该场景包括有待转区、有公交车道、有停靠车道、驶入驶出车道不对齐、异形钝角路口。
图8a中所示驶入道路包含1条待转区的左转车道、1条右转车道、1条公交车直行车道以及2条普通直行车道；驶出道路包含1条停靠车道和3条普通直行驶出车道。
根据交通规则，在公交车道指定通行时段内其他车辆不得无故占用公交车道，而停靠车道内多为公交车站、故障临停区等特殊区域，因此该场景中驶入车道仅有2条直行车道(L i1，L i2)，驶出车道仅有3条直行车道(L o1，L o2，L o3)，则车道级全连接拓扑包括所有备选的6条车道拓扑：L i1→L o1，L i1→L o2，L i1→L o3，L i2→L o1，L i2→L o2，L i2→L o3。
基于高精度地图和传感器获取的环境感知障碍物信息以及车道线信息，如图8a中所示的障碍物U4、U5、U6、U7、U8、U9，以及车道标线R1和R3等。其中硬边界约束包括上述障碍物U4、U5、U6、U7、U8、U9所示的马路牙子、安全岛、绿化带等，以及路口边界线等；软边界约束由图8a中所示两个左转待转区的车道标线R1和R3构成。
上述障碍物区域分布在上述生成的6条车道拓扑的左右两侧，且对于每条车道拓扑来说，左右侧障碍物区域分布完全相同，因此，每条车道拓扑的硬边界约束相同。同样地，对于每条车道拓扑来说，软边界约束分布也完全相同，因此，每条车道拓扑的软边界约束相同。
虚拟边界约束考虑邻近车道中车流轨迹的干涉影响，反映软硬边界对同向车道逐渐减弱的间接约束作用，因此每条车道拓扑的虚拟边界约束各不相同。例如，以车道拓扑L i2→L o2为例，由于车道拓扑L i1→L o1中的车流轨迹受到其软硬边界影响，所以由于临近车道的车流轨迹干涉影响，其软硬边界也将对车道拓扑L i2→L o2产生虚拟边界约束。左侧最近邻车道L i1→L o1的参考拓扑曲线如图8b中曲线S5所示，其碰撞深度向量如
Figure PCTCN2022138461-appb-000006
所示。车道拓扑L i2→L o2的左侧虚拟边界约束如图8b中区域U40、U50、U60和虚线S6、S7所示，沿碰撞深度向量
Figure PCTCN2022138461-appb-000007
平移距离为
Figure PCTCN2022138461-appb-000008
其中，左对齐的车道位序
Figure PCTCN2022138461-appb-000009
车道位序
Figure PCTCN2022138461-appb-000010
且d 1＞W vehicle，以及d 1＜W lane，即平移距离大于车体宽度且小于车道宽度。
同样地，车道拓扑L i1→L o2受其右侧最近邻车道的车流轨迹干涉影响，产生虚拟边界约束。右侧最近邻车道L i2→L o3的参考拓扑曲线如图8b中曲线S8所示，其碰撞深度向量如
Figure PCTCN2022138461-appb-000011
所示。右对齐的车道位序
Figure PCTCN2022138461-appb-000012
沿碰撞深度向量
Figure PCTCN2022138461-appb-000013
反向平移距离为d 2，且d 2＞W vehicle，以及d 2＜W lane，因此车道拓扑L i1→L o2的右侧虚拟边界约束如区域U70所示。
车道级全连接拓扑中的其他车道拓扑，其软硬边界约束及虚拟边界约束生成方式可参阅上述介绍，在此不再赘述。
在上述软硬边界约束和虚拟边界约束下，采用基于Spiral曲线的优化算法为车道级全连接拓扑中的6条车道拓扑生成曲率平滑且类人的虚拟车道轨迹。基于起止位姿采样、控制点采样、曲线轨迹生成，车道级全连接拓扑中的6条车道拓扑都生成若干条备选曲线。
以车道拓扑L i1→L o1，L i2→L o2的三条备选曲线为例进行说明，如图8c所示。对于车道拓扑L i1→L o1的备选曲线a、b、c，曲线b因曲率及其变化率均更均匀，到软边界(待转区车道线)的距离满足通行空间需求，且曲线长度更短，因此曲线b为车道拓扑L i1→L o1的最优曲线；对于车道拓扑L i2→L o2，由于曲线f曲率变化过大，且曲线d与曲线e相比，虽然到软边界(待转区车道线)的距离满足通行空间需求，曲线长度仅略短，但曲线d与虚拟边界S6之间的距 离偏小，因此曲线e的虚拟车道曲线评价值更优，应选为车道拓扑L i2→L o2的最优曲线。最后，车道拓扑L i1→L o1的最优曲线b与车道拓扑L i2→L o2的最优曲线e均满足软硬边界约束和虚拟边界约束，与场景中的边界障碍物无碰撞，因此不需要进行局部曲线形态调整。
根据车道拓扑曲线合理性筛选原则，对上述车道级全连接拓扑中的6条车道拓扑及其曲线进行合理性筛选，删除不合理的车道拓扑，得到合理的车道拓扑曲线集合，从而完成了对当前场景的道路拓扑分析。
具体地，将该实施例场景驶入道路、驶出道路的方向向量进行前后延伸，其交点位于路口驶入道路和驶出道路之间，因此通过方向向量延长线交点作角平分线得到投影线。将驶入车道L i1，L i2和驶出车道L o1，L o2，L o3的车道边线延伸到投影线，计算左右车道边线在投影线上的对齐系数(即投影重叠系数)；若对齐系数大于预设阈值，则保留该车道拓扑；若对齐系数不大于预设阈值，则从车道级全连接拓扑中删除该车道拓扑。若某条驶入车道的所有对齐系数均小于该设定阈值，可选的，可保留对齐系数最大的车道拓扑。
下面以两种重叠结果为例进行说明。设定对齐系数预设阈值为1/3。
(1)如图8d所示，对于驶入车道L i1，车道拓扑L i1→L o1的重叠长度
Figure PCTCN2022138461-appb-000014
Figure PCTCN2022138461-appb-000015
车道宽度
Figure PCTCN2022138461-appb-000016
因此对齐系数
Figure PCTCN2022138461-appb-000017
Figure PCTCN2022138461-appb-000018
该对齐系数0.5大于1/3。
车道拓扑L i1→L o2的重叠长度
Figure PCTCN2022138461-appb-000019
车道宽度
Figure PCTCN2022138461-appb-000020
Figure PCTCN2022138461-appb-000021
因此对齐系数
Figure PCTCN2022138461-appb-000022
该对齐系数0.5大于1/3。
车道拓扑L i1→L o3的重叠长度w same(L in-1,L out-3)＝0，车道宽度
Figure PCTCN2022138461-appb-000023
Figure PCTCN2022138461-appb-000024
因此对齐系数为0。
同理可得，对于驶入车道L i2，车道拓扑L i2→L o1的重叠长度w same(L in-2,L out-1)＝0，车道宽度
Figure PCTCN2022138461-appb-000025
因此对齐系数为0。
车道拓扑L i2→L o2的重叠长度
Figure PCTCN2022138461-appb-000026
车道宽度
Figure PCTCN2022138461-appb-000027
Figure PCTCN2022138461-appb-000028
因此对齐系数
Figure PCTCN2022138461-appb-000029
该对齐系数0.4大于1/3。
车道拓扑L i2→L o3的重叠长度
Figure PCTCN2022138461-appb-000030
车道宽度
Figure PCTCN2022138461-appb-000031
Figure PCTCN2022138461-appb-000032
因此对齐系数
Figure PCTCN2022138461-appb-000033
该对齐系数0.6大于1/3。
综上，车道拓扑L i1→L o1，L i1→L o2和L i2→L o2，L i2→L o3的对齐系数均大于预设阈值，则保留相应的车道拓扑，而车道拓扑L i1→L o3和L i2→L o1的对齐系数小于预设阈值，则从车道级全连接拓扑中删除相应的车道拓扑及其曲线。
(2)如图8e所示，对于驶入车道L i1，车道拓扑L i1→L o1的重叠长度
Figure PCTCN2022138461-appb-000034
Figure PCTCN2022138461-appb-000035
车道宽度
Figure PCTCN2022138461-appb-000036
因此对齐系数
Figure PCTCN2022138461-appb-000037
Figure PCTCN2022138461-appb-000038
车道拓扑L i1→L o2的重叠长度
Figure PCTCN2022138461-appb-000039
车道宽度
Figure PCTCN2022138461-appb-000040
对齐系数
Figure PCTCN2022138461-appb-000041
该对齐系数0.85大于1/3。
车道拓扑L i1→L o3的重叠长度w same(L in-1,L out-3)＝0，车道宽度
Figure PCTCN2022138461-appb-000042
Figure PCTCN2022138461-appb-000043
对齐系数为0。
同理可得，对于驶入车道L i2，车道拓扑L i2→L o1的重叠长度w same(L in-2,L out-1)＝0，车道宽度
Figure PCTCN2022138461-appb-000044
因此对齐系数为0。
车道拓扑L i2→L o2的重叠长度
Figure PCTCN2022138461-appb-000045
车道宽度
Figure PCTCN2022138461-appb-000046
对齐系数
Figure PCTCN2022138461-appb-000047
车道拓扑L i2→L o3的重叠长度
Figure PCTCN2022138461-appb-000048
车道宽度
Figure PCTCN2022138461-appb-000049
对齐系数
Figure PCTCN2022138461-appb-000050
该对齐系数0.86大于1/3。
因此，车道拓扑L i1→L o2和L i2→L o3的对齐系数均大于预设阈值，则保留相应的车道拓扑，但车道拓扑L i1→L o1，L i1→L o3和L i2→L o2，L i2→L o1的对齐系数小于预设阈值，应从车道级全连接拓扑中删除相应的车道拓扑及其曲线。
上述两种示例筛选后保留的车道拓扑曲线集合{L i1→L o1，L i1→L o2，L i2→L o2，L i2→L o3}或{L i1→L o2，L i2→L o3}中，各车道拓扑均曲率平滑，满足车辆转弯半径要求，因此满足运动学筛选原则；由于在车道拓扑生成中已考虑公交车道、停靠车道、左转待转车道、右转车道的交通规则，因此生成的车道级全连接拓扑也符合交通规则筛选条件；且各车道拓扑曲线均未与场景中各障碍物区域发生碰撞，各轨迹之间、与待转区之间均无干涉影响，且有足够通行空间，因此也符合碰撞检测筛选原则和车流干扰筛选原则。
对于上述示例(1)的情况，筛选后保留的车道拓扑曲线集合不需要进行拓扑补充。但对于示例(2)的情况，筛选后保留的车道拓扑曲线集合为{L i1→L o2，L i2→L o3}，驶出车道L o1无合理保留的车道拓扑，因此应当根据左对齐/右对齐补充原则、驶出拓扑补充原则、相邻车道补充原则进行拓扑补齐。按照左对齐的车道对应关系，应当为驶出车道L o1补充车道拓扑L i1→L o1；由于按照右对齐的车道对应关系，驶出车道L o1无对应驶入车道，所以不需要补充右对齐车道拓扑，至此保留的车道拓扑曲线集合为{L i1→L o1，L i1→L o2，L i2→L o3}，因此也不满足驶出拓扑补充原则和相邻车道补充原则的条件。
经上述方法对车道级全连接拓扑进行曲线生成、拓扑合理性筛选和补充后，即可得到完备且合理的车道拓扑曲线集合{L i1→L o1，L i1→L o2，L i2→L o2，L i2→L o3}或{L i1→L o1，L i1→L o2，L i2→L o3}，从而完成对当前场景的道路拓扑分析。
下面仅以示例(1)产生的车道拓扑曲线集合{L i1→L o1，L i1→L o2，L i2→L o2，L i2→L o3}为例进行介绍。
该示例以车道拓扑推荐评价函数包含导航代价、拓扑代价、平滑性代价、车流交汇代价、交通规则代价和通行效率代价为例，对由同一条驶入车道进入该场景的多条车道拓扑曲线计算导航推荐优先级，用于选取综合最优的车道拓扑。
假设该实施例场景中，全局导航规划路线在该路口之后向左转弯到达终点，因此各驶出车道到终点方向的可行驶距离关系为：
Figure PCTCN2022138461-appb-000051
此外，设置各项代价的权重关系为：导航代价所占权重w 1＝5，拓扑代价所占权重w 2＝2，平滑性代价所占权重w 3＝0.5，车流交汇代价所占权重w 4＝1，交通规则代价所占权重w 5＝10，通行效率代价所 占权重w 6＝0.3。
对于由驶入车道L i1进入路口的车道拓扑L i1→L o1和L i1→L o2：
(1)导航代价C 1：由于
Figure PCTCN2022138461-appb-000052
因此
Figure PCTCN2022138461-appb-000053
(2)拓扑代价C 2：按照左对齐的车道对应关系，车道拓扑L i1→L o1是左对齐的，横向跨车道数为0，但车道拓扑L i1→L o2按照左对齐原则向右跨车道数为1，因此
Figure PCTCN2022138461-appb-000054
(3)平滑性代价C 3：车道拓扑L i1→L o1的曲率大于车道拓扑L i1→L o2的曲率，因此
Figure PCTCN2022138461-appb-000055
(4)车流交汇代价C 4：由于驶出车道L o1仅有一条车道拓扑L i1→L o1，但驶出车道L o2同时存在两条车道拓扑L i1→L o2和L i2→L o2，存在车流交汇风险，因此
Figure PCTCN2022138461-appb-000056
(5)交通规则代价C 5：该实施例场景为直行路口，各驶出车道无优先级区别，因此
Figure PCTCN2022138461-appb-000057
(6)通行效率代价C 6：根据人类驾驶经验和宏观交通流分配情况，该实施例场景中，选择车道拓扑L i1→L o1比车道拓扑L i1→L o2的概率更大，因此
Figure PCTCN2022138461-appb-000058
综上考虑，车道拓扑推荐评价函数总代价
Figure PCTCN2022138461-appb-000059
由驶入车道L i1进入路口的车道拓扑L i1→L o1和L i1→L o2，导航推荐的优先级为：(L i1→L o1)>(L i1→L o2)。
同理可得，车道拓扑推荐评价函数总代价
Figure PCTCN2022138461-appb-000060
由驶入车道L i2进入路口的车道拓扑L i2→L o2和L i2→L o3，导航推荐的优先级为：(L i2→L o2)>(L i2→L o3)。
当车辆在该实施例场景中行驶时，若该路口场景比较空旷，车辆通过该路口时无其他车辆产生侧向挤压，则将沿着车道拓扑L i1→L o1或L i2→L o2行驶，如图8f所示；
若车道拓扑L i1→L o1中存在前方车辆慢速行驶或时停时走，则自车可以实时选择车道拓扑L i1→L o2继续行驶，从而提升舒适性和通行效率，如图8g所示；
若车道拓扑L i2→L o2中存在其他车辆沿着车道拓扑L i1→L o2行驶，与自车抢道或产生侧向挤压，则自车可以实时选择车道拓扑L i2→L o3，从而减少与他车冲突的风险，提高安全性和通行效率，如图8h所示。
该实施例考虑了公交车道、停靠车道、左转待转车道、右转车道等特殊车道的交通规则通行特性，并在此基础上生成了车道级全连接拓扑空间，既包含该场景下所有可能的车道拓扑，又排除违背交通规则的不合理车道拓扑，为构建完备且合理的道路拓扑空间奠定基础。
本方案各条车道拓扑的硬边界约束中不仅考虑了高精度地图，还考虑了传感器实时感知的物理世界现实变更等，使得本方法既可以用于离线地图生成，又可用于车道拓扑轨迹在线生成；在软边界约束和虚拟边界约束生成中，考虑了待转区等软硬边界对其最近邻车流轨迹L i1→L o1或L i2→L o3、次近邻车流轨迹L i2→L o2或L i1→L o1、L i1→L o2等同向车道逐渐减弱的干涉影响，保证了不同车道拓扑轨迹之间的通行空间和安全性，更符合人类驾驶员习惯和实际道路交通规律。
其中，在车道拓扑L i1→L o1、L i2→L o2的类人性轨迹生成中，不仅考虑轨迹的曲率、曲率变化率，还考虑了车道轨迹之间的通行空间、通行效率等因素，从而从多条采样生成的备选轨迹中选择最优轨迹时，保证该待转区钝角路口中弯曲的外侧车道拓扑L i2→L o2其轨迹在曲率 尽量降低的同时保证与内侧车道拓扑L i1→L o1轨迹保持足够远距离，避免出现轨迹被过于拉直、导致虚拟车道轨迹之间间距过窄问题，保证了车道拓扑轨迹的安全性和平顺性。
同时，考虑了驶入、驶出车道的对齐程度以及边道拓扑补齐，使得道路拓扑分析结果类人程度高：对于图8d中多对多且投影重叠区域均较大的情况，保留车道拓扑轨迹集合{L i1→L o1，L i1→L o2，L i2→L o2，L i2→L o3}；对于图8e中多对多但重叠区域较小的情况，筛除无效车道拓扑，并增加边道拓扑进行拓扑补齐，得到车道拓扑轨迹集合{L i1→L o1，L i1→L o2，L i2→L o3}，提升车道拓扑轨迹集合的丰富性，保证道路拓扑分析的完备性和合理性，从根本上保证了密集车流场景中车辆实时车道决策的自由度和灵活性，以及用于他车意图和轨迹预测的准确性。
该实施例还对车道拓扑轨迹集合{L i1→L o1，L i1→L o2，L i2→L o2，L i2→L o3}中的车道拓扑轨迹的导航推荐评价中，引入了全局导航信息、车道拓扑与交汇、交通规则优先级、宏观交通流信息等，同时还能在实时导航中结合动静态交通环境，辅助车辆实时选择最优目标路径，从而具备全局视野，提前规避车道斜穿、他车侧向挤压、多车道合并等车流冲突的高风险车道拓扑，减少与他车的复杂交互，有效提升车辆交汇、车流密集情况下的通行效率和舒适性。
与现有技术相比，本方案的车道级全连接拓扑空间生成考虑了公交车道、停靠车道、左转待转车道、右转车道等特殊车道的交通规则通行特性，既包含该场景下所有可能的车道拓扑，又排除违背交通规则的不合理车道拓扑，为构建完备且合理的道路拓扑空间奠定基础；
与现有技术相比，车道拓扑的边界约束生成中考虑了传感器实时感知的物理世界现实变更，使本方法可用于车道拓扑轨迹在线生成；考虑了软硬边界和车流轨迹对同向车道逐渐减弱的干涉影响和间接约束，保证了不同车道拓扑轨迹之间的通行空间和安全性，更符合人类驾驶员习惯和实际道路交通规律；虚拟车道轨迹生成中考虑了曲线平直性(通行效率)以及相邻车道间的通行空间，保证了虚拟车道轨迹的平顺性和类人性，减少与其他车道的车流冲突；
与现有技术相比，本方案在车道级全连接拓扑的基础上，通过类人的投影重叠筛选、边道拓扑补齐，删除不合理的车道拓扑，构建完备且合理的道路拓扑空间，保证存在多种合理驾驶轨迹的场景中保证道路拓扑分析的完备性和合理性，有效提升密集车流场景中车辆实时车道决策的自由度和灵活性；
与现有技术相比，本方案在车道拓扑轨迹的导航推荐评价中，考虑全局导航信息、车道拓扑与交汇、交通规则优先级、宏观交通流信息等，使得车辆具备全局视野且能结合动静态交通环境，提前规避高风险车道，减少与他车交互，有效提升车辆交汇、车流密集情况下的舒适性和通行效率。
实施例二
如图9a所示，该实施例场景为环岛场景。车辆需经由下方道路进入环岛并经由右下第一个出口或下方第二个出口离开该环岛。其中，进入环岛的道路可能包含一条或多条驶入车道，沿环岛巡航的道路也可能具备一条或多条沿环岛车道。根据具体驾驶任务，可能途径若干个环岛路口后驶离环岛，也可能在下一个环岛路口驶离该环岛。
参照图9a所示，进入环岛为一个路口A，驶离环岛为一个路口B或C。首先，建立驶入环岛、驶出环岛路口的车道级全连接拓扑，以环岛任务A->B和环岛任务A->C为例，驶入环岛、驶出环岛路口的拓扑如图9a所示。
如图9b所示，以驶入环岛路口的车道拓扑1-4和2-5为例介绍生成的边界约束。
(1)硬边界约束包括道路边界、路沿、绿化带以及不可跨越的车道线等，如图9b中粗 线条以及环岛路线边界所示；
(2)软边界约束为可跨越的车道线，如虚线车道边界；
(3)虚拟边界约束为软边界、硬边界约束平移后得到的形状约束，如图9b中箭头所指向的曲线。
本实施例中，车道拓扑2-5可能与右侧硬边界约束发生碰撞，因此该硬边界将影响车道拓扑2-5的车流轨迹；间接地，该硬边界将影响到其次邻近车道拓扑1-4的车流轨迹，因此该硬约束通过平移原则移动至对应位置(图9b中箭头所指向的曲线)，得到车道拓扑1-4的虚拟边界约束。
本实施例的车道曲线生成中，对于该驶入环岛路口，如图9c所示，进入环岛端属于真实路口(边界L1)，而沿环岛端(边界L2)属于虚拟路口边界。因此对于环岛路口，边界线在环岛中的虚拟路口边界，车道拓扑曲线生成的起止位姿采样点可以向虚拟路口外侧延伸，如图9c中的点a、点b、点c，从而生成合理、平坦的类人性车道拓扑曲线，而不受高精度地图中虚拟路口边界的过分约束。车道曲线生成的方法与前述实施例一类似，在此不再赘述。其中，车道拓扑1-4生成的虚拟车道曲线如图9c中的虚线曲线所示，其中综合评价得到的最优轨迹如实线曲线所示。
经过拓扑投影筛选原则、运动学筛选原则、交通规则筛选、碰撞检测筛选和车流干扰筛选等原则后，保留的驶入路口车道拓扑如图9d中所示a、b、c三条，筛除了1-5，2-3、2-4三条备选车道拓扑；同理，驶出路口车道拓扑如图9d中所示d、e、f三条，筛除了3-6，4-6、5-7三条备选车道拓扑。
本实施例环岛场景中，导航推荐评价函数中影响较大的代价包括：
(1)导航代价：对于环岛绕行路径百分比较高(第二个驶出环岛路口离开环岛)的情况，环岛绕行路径百分比较高，可将第一个驶出环岛路口作为普通绕行处理，由于内侧车道总轨迹长度最短，环岛内的车道3、4、5导航代价大小为：车道5的导航代价>车道4的导航代价>车道3的导航代价；对于环岛绕行路径百分比较低(下一个路口就驶离环岛)的情况，导航代价为车道5的导航代价＝车道4的导航代价＝车道3的导航代价。
(2)拓扑代价：c的拓扑代价>b的拓扑代价>a的拓扑代价,d的拓扑代价>e的拓扑代价>f的拓扑代价。
(3)平滑性代价：曲线c的平滑性代价≈曲线b的平滑性代价≈曲线a的平滑性代价,曲线d的平滑性代价≈曲线e的平滑性代价≈曲线f的平滑性代价。
(4)车流交汇代价：对于环岛绕行路径百分比较高(第二个驶出环岛路口离开环岛)的情况，环岛内的车道3、4、5的车流交汇代价大小为：车道5的车流交汇代价>>车道4的车流交汇代价>>车道3的车流交汇代价；
对于环岛绕行路径百分比较低(下一个路口就驶离环岛)的情况，车流交汇代价为：车道5的车流交汇代价＝车道4的车流交汇代价＝车道3的车流交汇代价。
(5)交通规则代价：c的交通规则代价>b的交通规则代价>a的交通规则代价,d的交通规则代价>e的交通规则代价>f的交通规则代价。
(6)通行效率代价：a的通行效率代价>b的通行效率代价>c的通行效率代价,5的通行效率代价>4的通行效率代价>3的通行效率代价,f的通行效率代价>e的通行效率代价>d的通行效率代价。
基于上述代价，考虑到车道4比车道5较优，因此可通过变道实现最优路径2-a-5-4-5-f-6；同样地，考虑到车道4比车道3较优，因此可通过变道实现最优路径1-b-4-5-f-6。
综合以上考虑，本实施例导航推荐评价得到的最优路径为：1-b-4-5-f-6和2-a-5-4-5-f-6。
该实施例针对驶入、驶出环岛路口特点，对虚拟路口侧轨迹始末姿态采样点向外侧进行合理延伸，提高生成轨迹的质量，提升轨迹类人性，避免高精度地图制图引起的轨迹不合理，也避免了拓扑筛选因轨迹生成不合理而错误筛除，保障道路拓扑分析的完备性。
该实施例中，驶入、驶出环岛路口内车道拓扑轨迹生成时，考虑了多条平行的车道拓扑情况下的虚拟边界间接约束作用，提高生成轨迹的类人性和通过性，避免与相邻车道拓扑的干涉影响，提高环岛驶入驶出路口的安全性；
同时，考虑了环岛绕行路径百分比较高时，绕行过程中穿越环岛路口作为普通环岛绕行处理，降低环岛场景处理的复杂度；基于完备的车道拓扑空间，通过考虑车流交汇和通行效率代价，在绕行百分比高时推荐进入环岛内圈车道，极大提高通行效率，避免在环岛路口内车流汇聚的弯道区域中变道的问题，提升通行安全性和类人程度；
该实施例通过保留多条合理的车道拓扑，构建完备的车道拓扑空间，保障在各种交通环境下，均能够有效提升通行效率、安全性、舒适性：当环岛中车流密集时，尽快进入环岛最外圈进行沿环岛行驶；在环岛外圈汇入过多车辆时，直接进入中间车道沿环岛行驶，避免交通拥堵造成的长时间等待，提高通行效率；当环岛中车流通畅时，直接驶入最内侧车道沿环岛行驶，有效减少驶入驶出环岛路口的车流交汇，降低自车风险，行驶距离短，通行效率高。
实施例三
参照图10a、图10b所示，为本申请实施例提供的有待转区、红绿灯的左转路口场景。根据待转区域、停止线位置，考虑交通规则(红绿灯状态与停止线待转区信息)，以及其他边界约束，生成：(1)当左转灯为红灯时，待转区停止线处的车道级全连接车道拓扑曲线如图10a中路口虚线所示；(2)当左转灯为绿灯时，路口停止线处的车道级全连接车道拓扑轨迹如图10b中路口虚线所示。其中，具体生成方法可参阅前述实施例，在此不再赘述。
基于拓扑筛选、导航推荐评价得到导航推荐最优车道拓扑曲线，如图10a、图10b中实线曲线所示。
当车辆行驶到图中所示有待转区的左转路口时，若传感器实时感知当前左转灯为红灯、直行灯为绿灯时，车辆先进入待转区等待左转灯变为绿灯，因此最优推荐车道拓扑曲线如图10a中的实线曲线所示。若传感器实时感知当前左转灯为绿灯时，图10a中导航推荐最优车道拓扑的曲率、通行效率代价过高，因此最优推荐的是不考虑待转区的最内侧内切车道曲线，如图10b中的实线曲线所示。
该实施例对有待转区、红绿灯的左转路口，考虑不同交通时间(红灯、绿灯)时自车所处位置不同，分别生成不同的车道级全连接拓扑曲线，保证道路拓扑分析的完备性，能够在任何红绿灯状态下推荐最符合当前时刻的最佳车道拓扑曲线，避免车道拓扑曲线过于呆板，不能适应实时的交通状态变化，提高轨迹类人性。
实施例四
参照图11a、图11b所示，为本申请实施例提供的小S弯场景。在该场景中，人类驾驶员为了追求更平直的行驶路径，经常会跨车道线行驶。若自动驾驶车辆仅能沿着车道线所确定的车道拓扑曲线行驶，当遇到人类驾驶员跨车道线行驶的场景时，将会触发紧急避让行为，则会降低舒适度，体验不佳，甚至有碰撞风险，如图11a所示。
在该场景中，采用如前述实施例所述方法可生成如图11b所示的6条虚拟车道拓扑曲线， 得到完备的道路拓扑分析。在实时导航的过程中，自动驾驶车辆可以根据他车1的行为，选择虚拟车道拓扑曲线3实现跨车道线行驶通过此小S弯场景，避免和他车1的碰撞可能，提高自车安全性；但当没有他车影响时，由于跨越车道线的虚拟车道拓扑曲线的导航推荐评价总代价更高，会优先为自车推荐不跨越车道线的虚拟车道拓扑曲线，如图11b中的车道拓扑曲线1、2、4、6所示。
实施例五
参照图12所示，为本申请实施例提供的城区高架过渡段-匝道场景。目前一般高精度地图提供的进入匝道方式仅有如图12中车道拓扑曲线1。但在实际驾驶过程中，若自车遇到交通较为拥堵或右侧车道存在障碍物的情况时，自车可能被右侧车辆(或障碍物)压制在右二车道，长时间无法变道到最右侧车道，从而错失进入匝道的时机。
在该场景中，采用如前述实施例所述方法可生成如图12所示的3条车道拓扑曲线1、2、3，得到完备的道路拓扑分析。在交通较为拥堵或存在障碍物时，自车虽然被右侧车辆(或障碍物)压制在右二车道，但可以选择虚拟车道拓扑曲线2进入匝道中；若交通通畅或不存在障碍物时，由于虚拟车道拓扑曲线1的导航推荐评价最优，虚拟车道拓扑曲线2其次，虚拟车道拓扑曲线3再次，此时自车优先会通过换道进入最右侧车道，然后选择虚拟车道拓扑曲线1进入匝道中，更符合人驾习惯。
实施例六
参照图13所示，为本申请实施例提供的多车道无车道标线路段场景。在城区道路上，多车道无标线路段通常出现在车道数量变化、非路口人行横道处等交通场景中。由于不存在与之车流交汇的道路，这种路段场景不属于常规交叉路口，然而该场景同样适用本方案所述的路口场景。
针对该实施例场景，采用如前述实施例所述方法可生成合理且完备的虚拟车道拓扑曲线集合并赋予合适的导航推荐评价，如图13中的车道拓扑曲线(图中实线曲线)所示。
实施例七
参照图14所示，为本申请实施例提供的连续转弯场景。该场景需要完成左转之后立即右转的驾驶任务。高精地图一般在左转路口仅提供驶入道路左一车道对驶出道路左一车道的虚拟车道拓扑曲线1，但在左转后、右转前，车辆将面临短距离连续换道三次的问题，实际场景中很难完成。
针对该实施例场景，本方案采用如前述实施例所述方法可生成合理且完备的虚拟车道拓扑曲线集合，并进行导航推荐评价，其得到虚拟车道拓扑曲线1、2和3，因此，车辆可以选择虚拟车道拓扑曲线3左转过第一个路口，此后将只需要变道一次即可右转通过第二个路口，保证了车道选择的自由度，极大提升极端交通场景中的成功率和通行效率。
实施例八
参照图15所示，为本申请实施例提供的窄道掉头场景。在该实施例场景中，采用如前述实施例所述方法可生成完备的车道级全连接拓扑，得到虚拟车道拓扑曲线1、2。基于拓扑合理性筛选中，由于虚拟车道拓扑曲线1不能满足运动学筛选原则(小半径的掉头曲线)，从而将虚拟车道拓扑曲线1删除，仅保留虚拟车道拓扑曲线2；若虚拟车道拓扑曲线1、2均不满 足运动学筛选原则，则保留曲率半径更大的一条车道拓扑曲线，确保车道拓扑的连通性，从而构成完备且合理的车道拓扑曲线集合，保证车道拓扑轨迹的类人性。
上述各实施例介绍了不同场景的实现方式，当然，本方案还可以用于其他场景，本方案对此不做具体限定。
在上述实施例的基础上，本方案还提供一种基于路口的地图生成方法，包括：根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及车道线生成M条车道拓扑曲线，所述车道拓扑曲线为以所述驶入道路的驶入车道的末端和所述驶出道路的驶出车道的起始点为端点的曲线；对所述M条车道拓扑曲线进行合理性检测处理，以得到K’条车道拓扑曲线，其中，K’不大于M；根据所述路口内的K’条车道拓扑曲线生成所述路口的地图。
其中，针对该方法的具体实现可参阅前述实施例中的相关介绍，在此不再赘述。
参照图16所示，为本申请实施例提供的一种引导车辆行驶的装置示意图。如图16所示，其包括曲线生成模块1601、检测处理模块1602和确定模块1603，其中：
曲线生成模块1601，用于根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及车道线生成M条车道拓扑曲线，所述车道拓扑曲线为以所述驶入道路的驶入车道的末端和所述驶出道路的驶出车道的起始点为端点的曲线；
检测处理模块1602，用于对所述M条车道拓扑曲线进行合理性检测处理，以得到K’条车道拓扑曲线，其中，K’不大于M；
确定模块1603，用于当所述车辆位于所述驶入道路中时，从所述K’条车道拓扑曲线中确定目标路径。
其中，所述曲线生成模块1601，用于：
根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及不可跨越车道线得到车道拓扑曲线硬边界约束；
根据所述路口、所述路口的驶入道路、所述路口的驶出道路中的可跨越车道线得到车道拓扑曲线软边界约束；
根据所述车道拓扑曲线硬边界约束和所述车道拓扑曲线软边界约束得到K个车道拓扑曲线虚拟边界约束；
根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束生成所述M条车道拓扑曲线。
可选的，所述K个车道拓扑曲线虚拟边界约束与K个车道拓扑对应，所述K个车道拓扑曲线虚拟边界约束中任一车道拓扑曲线虚拟边界约束A是通过将所述路口的最左侧车道拓扑左侧的硬边界约束和/或软边界约束向右平移第一预设距离，以及将所述路口的最右侧车道拓扑右侧的硬边界约束和/或软边界约束向左平移第一预设距离得到的，其中，所述第一预设距离是根据车道拓扑A’的车道位序确定的，或者，所述第一预设距离是根据所述车辆预设通过宽度、车道宽度中的至少一项以及所述车道拓扑A’的车道位序确定的，所述车道拓扑曲线虚拟边界约束A与所述车道拓扑A’对应，所述K个车道拓扑包括所述路口的最左侧车道拓扑和所述最右侧车道拓扑。
进一步地，所述曲线生成模块1601，还用于：
分别对所述驶入道路中每条驶入车道的末端进行角度采样得到所述驶入道路中的至少一个起始点位姿向量，并分别对所述驶出道路中每条驶出车道的起始点进行角度采样得到所述驶出道路中的至少一个终点位姿向量；
对所述至少一个起始点位姿向量和所述至少一个终点位姿向量进行曲线采样，得到所述驶入道路和所述驶出道路之间的多条曲线；
根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束对所述多条曲线进行筛选处理，得到所述M条车道拓扑曲线。
其中，所述曲线生成模块1601，还用于：
在所述驶入道路的每条驶入车道的末端和所述驶出道路的每条驶出车道的起始点之间生成多个控制点；
根据所述至少一个起始点位姿向量、所述至少一个终点位姿向量以及所述多个控制点，生成所述多条平滑曲线。
可选的，所述驶入道路中的至少一个起始点位姿向量是通过将所述每条驶入车道的末端延伸第二预设距离并进行采样得到的。
其中，所述检测处理模块1602，用于：
根据所述驶入道路的方向向量、驶出道路的方向向量得到所述驶入道路、驶出道路之间的投影线，所述投影线为所述驶入道路的方向向量、驶出道路的方向向量相交所得到的夹角的平分线所在的直线，或者，所述投影线为垂直于所述驶出道路的方向向量且通过所述驶出道路的起始点的直线；
计算所述驶入道路的每条驶入车道和所述驶出道路的每条驶出车道之间的对齐系数，其中，所述每条驶入车道和所述每条驶出车道之间的对齐系数为第一参数与第二参数之间的比值，所述第一参数为所述每条驶入车道的车道边线以及所述每条驶出车道的车道边线分别延长至所述投影线所得到的两条线段之间的重叠长度，所述第二参数为所述每条驶入车道的车道边线和所述每条驶出车道的车道边线分别延长至所述投影线所得到的两条线段中最短的线段的长度；
根据所述驶入道路的每条驶入车道和所述驶出道路的每条驶出车道之间的对齐系数，得到K’条车道拓扑曲线，其中，所述K’条车道拓扑曲线包括以对齐系数大于第一预设阈值的驶入车道的末端、驶出车道的起始点为端点的曲线。
其中，所述K’条车道拓扑曲线包括分别以所述驶入道路的最左侧驶入车道的末端、所述驶出道路的最左侧驶出车道的起始点为端点的曲线，还包括分别以所述驶入道路的最右侧驶入车道的末端、所述驶出道路的最右侧驶出车道的起始点为端点的曲线。
进一步地，所述K’条车道拓扑曲线包括以所述驶入道路中的驶入车道X的末端以及所述驶出道路中的驶出车道Y的左侧车道的起始点为端点的车道拓扑曲线，以及以所述驶入车道X的末端以及所述驶出车道Y的右侧车道的起始点为端点的车道拓扑曲线，还包括以所述驶入车道X的末端以及所述驶出车道Y的起始点为端点的车道拓扑曲线，其中，所述驶出车道Y的左侧和右侧均存在车道；或者，
所述K’条车道拓扑曲线包括以所述驶入道路中的驶入车道X的末端以及所述驶出道路中的驶出车道Y的左侧车道或者右侧车道的起始点为端点的车道拓扑曲线，还包括以所述驶入道路中的驶入车道X的末端、所述驶出车道Y的起始点为端点的车道拓扑曲线，其中，所述驶出道路中的驶出车道Y仅左侧或者右侧存在车道。
可选的，所述K’条车道拓扑曲线中每条车道拓扑曲线的最大曲率不大于第二预设阈值，且，所述每条车道拓扑曲线与所述车道拓扑曲线软边界约束、所述车道拓扑曲线硬边界约束之间的距离不小于第三预设距离，且，任意两条车道拓扑曲线之间的距离不小于第四预设距离。
其中，所述装置还包括评价模块，用于：
计算所述K’条车道拓扑曲线中每条车道拓扑曲线的评价值，所述评价值与所述车道拓扑曲线曲率、曲率变化率以及斜穿车道数量、以及所述车道拓扑曲线对应的车道的车道交汇信息、交通规则信息、车流量估计值、可行驶距离中的至少一项有关；
所述确定模块1603，用于：
当所述车辆位于所述驶入道路中时，根据所述K’条车道拓扑曲线中每条车道拓扑曲线的评价值确定所述目标路径。
其中，所述路口包括交叉路口、环岛、待转区路口、小S弯、高架出入口、多车道且无车道标线路段、连续转弯路口、窄道掉头路口中的至少一种。
在本实施例中，该引导车辆行驶装置是以模块的形式来呈现。这里的“模块”可以指特定应用集成电路(application-specific integrated circuit，ASIC)，执行一个或多个软件或固件程序的处理器和存储器，集成逻辑电路，和/或其他可以提供上述功能的器件。
此外，以上曲线生成模块1601、检测处理模块1602和确定模块1603，可通过图17所示的引导车辆行驶装置的处理器1702来实现。
另一方面，本方案还提供一种基于路口的地图生成装置，包括：曲线生成模块，用于根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及车道线生成M条车道拓扑曲线，所述车道拓扑曲线为以所述驶入道路的驶入车道的末端和所述驶出道路的驶出车道的起始点为端点的曲线；检测处理模块，用于对所述M条车道拓扑曲线进行合理性检测处理，以得到K’条车道拓扑曲线，其中，K’不大于M；地图生成模块，用于根据所述路口内的K’条车道拓扑曲线生成所述路口的地图。
该装置还可以包括如上所述的各模块，本方案对此不做具体限定。
图17是本申请实施例提供的引导车辆行驶装置的硬件结构示意图。图17所示的引导车辆行驶装置1700(该装置1700具体可以是一种计算机设备)包括存储器1701、处理器1702、通信接口1703以及总线1704。其中，存储器1701、处理器1702、通信接口1703通过总线1704实现彼此之间的通信连接。
存储器1701可以是只读存储器(Read Only Memory，ROM)，静态存储设备，动态存储设备或者随机存取存储器(Random Access Memory，RAM)。
存储器1701可以存储程序，当存储器1701中存储的程序被处理器1702执行时，处理器1702和通信接口1703用于执行本申请实施例的引导车辆行驶方法的各个步骤。
处理器1702可以采用通用的中央处理器(Central Processing Unit，CPU)，微处理器，应用专用集成电路(Application Specific Integrated Circuit，ASIC)，图形处理器(graphics processing unit，GPU)或者一个或多个集成电路，用于执行相关程序，以实现本申请实施例的引导车辆行驶装置中的单元所需执行的功能，或者执行本申请方法实施例的引导车辆行驶方法。
处理器1702还可以是一种集成电路芯片，具有信号的处理能力。在实现过程中，本申请的引导车辆行驶方法的各个步骤可以通过处理器1702中的硬件的集成逻辑电路或者软件形式的指令完成。上述的处理器1702还可以是通用处理器、数字信号处理器(Digital Signal Processing，DSP)、专用集成电路(ASIC)、现成可编程门阵列(Field Programmable Gate Array，FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件。可以实现或者执行本申请实施例中的公开的各方法、步骤及逻辑框图。通用处理器可以是微处理器或者 该处理器也可以是任何常规的处理器等。结合本申请实施例所公开的方法的步骤可以直接体现为硬件译码处理器执行完成，或者用译码处理器中的硬件及软件模块组合执行完成。软件模块可以位于随机存储器，闪存、只读存储器，可编程只读存储器或者电可擦写可编程存储器、寄存器等本领域成熟的存储介质中。该存储介质位于存储器1701，处理器1702读取存储器1701中的信息，结合其硬件完成本申请实施例的引导车辆行驶装置中包括的单元所需执行的功能，或者执行本申请方法实施例的引导车辆行驶方法。
通信接口1703使用例如但不限于收发器一类的收发装置，来实现装置1700与其他设备或通信网络之间的通信。例如，可以通过通信接口1703获取数据。
总线1704可包括在装置1700各个部件(例如，存储器1701、处理器1702、通信接口1703)之间传送信息的通路。
应注意，尽管图17所示的装置1700仅仅示出了存储器、处理器、通信接口，但是在具体实现过程中，本领域的技术人员应当理解，装置1700还包括实现正常运行所必须的其他器件。同时，根据具体需要，本领域的技术人员应当理解，装置1700还可包括实现其他附加功能的硬件器件。此外，本领域的技术人员应当理解，装置1700也可仅仅包括实现本申请实施例所必须的器件，而不必包括图17中所示的全部器件。
本申请还提供一种智能驾驶车辆，包括行进系统、传感系统、控制系统和计算机系统，其中，所述计算机系统用于执行上述任一个方法中的一个或多个步骤。
本申请实施例还提供了一种计算机可读存储介质，该计算机可读存储介质中存储有指令，当其在计算机或处理器上运行时，使得计算机或处理器执行上述任一个方法中的一个或多个步骤。
本申请实施例还提供了一种包含指令的计算机程序产品。当该计算机程序产品在计算机或处理器上运行时，使得计算机或处理器执行上述任一个方法中的一个或多个步骤。
所属领域的技术人员可以清楚地了解到，为描述的方便和简洁，上述描述的系统、装置和单元的具体工作过程，可以参考前述方法实施例中的对应步骤过程的具体描述，在此不再赘述。
应理解，在本申请的描述中，除非另有说明，“/”表示前后关联的对象是一种“或”的关系，例如，A/B可以表示A或B；其中A，B可以是单数或者复数。并且，在本申请的描述中，除非另有说明，“多个”是指两个或多于两个。“以下至少一项(个)”或其类似表达，是指的这些项中的任意组合，包括单项(个)或复数项(个)的任意组合。例如，a，b，或c中的至少一项(个)，可以表示：a，b，c，a-b，a-c，b-c，或a-b-c，其中a，b，c可以是单个，也可以是多个。另外，为了便于清楚描述本申请实施例的技术方案，在本申请的实施例中，采用了“第一”、“第二”等字样对功能和作用基本相同的相同项或相似项进行区分。本领域技术人员可以理解“第一”、“第二”等字样并不对数量和执行次序进行限定，并且“第一”、“第二”等字样也并不限定一定不同。同时，在本申请实施例中，“示例性的”或者“例如”等词用于表示作例子、例证或说明。本申请实施例中被描述为“示例性的”或者“例如”的任何实施例或设计方案不应被解释为比其它实施例或设计方案更优选或更具优势。确切而言，使用“示例性的”或者“例如”等词旨在以具体方式呈现相关概念，便于理解。
在本申请所提供的几个实施例中，应该理解到，所揭露的系统、装置和方法，可以通过其它的方式实现。例如，该单元的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，例如，多个单元或组件可以结合或者可以集成到另一个系统，或一些特征可以忽略，或不执行。所显示或讨论的相互之间的耦合、或直接耦合、或通信连接可以是通过一 些接口，装置或单元的间接耦合或通信连接，可以是电性，机械或其它的形式。
作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。
在上述实施例中，可以全部或部分地通过软件、硬件、固件或者其任意组合来实现。当使用软件实现时，可以全部或部分地以计算机程序产品的形式实现。该计算机程序产品包括一个或多个计算机指令。在计算机上加载和执行该计算机程序指令时，全部或部分地产生按照本申请实施例的流程或功能。该计算机可以是通用计算机、专用计算机、计算机网络、或者其他可编程装置。该计算机指令可以存储在计算机可读存储介质中，或者通过该计算机可读存储介质进行传输。该计算机指令可以从一个网站站点、计算机、服务器或数据中心通过有线(例如同轴电缆、光纤、数字用户线(digital subscriber line，DSL))或无线(例如红外、无线、微波等)方式向另一个网站站点、计算机、服务器或数据中心进行传输。该计算机可读存储介质可以是计算机能够存取的任何可用介质或者是包含一个或多个可用介质集成的服务器、数据中心等数据存储设备。该可用介质可以是只读存储器(read-only memory，ROM)，或随机存取存储器(random access memory，RAM)，或磁性介质，例如，软盘、硬盘、磁带、磁碟、或光介质，例如，数字通用光盘(digital versatile disc，DVD)、或者半导体介质，例如，固态硬盘(solid state disk，SSD)等。
以上所述，仅为本申请实施例的具体实施方式，但本申请实施例的保护范围并不局限于此，任何在本申请实施例揭露的技术范围内的变化或替换，都应涵盖在本申请实施例的保护范围之内。因此，本申请实施例的保护范围应以所述权利要求的保护范围为准。

Claims (52)

  1. 一种引导车辆行驶的方法，其特征在于，包括：
    根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及车道线生成M条车道拓扑曲线，所述车道拓扑曲线为以所述驶入道路的驶入车道的末端和所述驶出道路的驶出车道的起始点为端点的曲线；
    对所述M条车道拓扑曲线进行合理性检测处理，以得到K’条车道拓扑曲线，其中，K’不大于M；
    当所述车辆位于所述驶入道路中时，从所述K’条车道拓扑曲线中确定目标路径。
  2. 根据权利要求1所述的方法，其特征在于，所述根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及车道线生成M条车道拓扑曲线，包括：
    根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及不可跨越车道线得到车道拓扑曲线硬边界约束；
    根据所述路口、所述路口的驶入道路、所述路口的驶出道路中的可跨越车道线得到车道拓扑曲线软边界约束；
    根据所述车道拓扑曲线硬边界约束和所述车道拓扑曲线软边界约束得到K个车道拓扑曲线虚拟边界约束；
    根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束生成所述M条车道拓扑曲线。
  3. 根据权利要求2所述的方法，其特征在于，所述K个车道拓扑曲线虚拟边界约束与K个车道拓扑一一对应，所述K个车道拓扑曲线虚拟边界约束中任一车道拓扑曲线虚拟边界约束A是通过将所述路口的最左侧车道拓扑左侧的硬边界约束和/或软边界约束向右平移第一预设距离，以及将所述路口的最右侧车道拓扑右侧的硬边界约束和/或软边界约束向左平移第一预设距离得到的，其中，所述第一预设距离是根据车道拓扑A’的车道位序确定的，或者，所述第一预设距离是根据所述车辆预设通过宽度、车道宽度中的至少一项以及所述车道拓扑A’的车道位序确定的，所述车道拓扑曲线虚拟边界约束A与所述车道拓扑A’对应，所述K个车道拓扑包括所述路口的最左侧车道拓扑和所述最右侧车道拓扑。
  4. 根据权利要求2或3所述的方法，其特征在于，所述根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束生成所述M条车道拓扑曲线，包括：
    分别对所述驶入道路中每条驶入车道的末端进行角度采样得到所述驶入道路中的至少一个起始点位姿向量，并分别对所述驶出道路中每条驶出车道的起始点进行角度采样得到所述驶出道路中的至少一个终点位姿向量；
    对所述至少一个起始点位姿向量和所述至少一个终点位姿向量进行曲线采样，得到所述驶入道路和所述驶出道路之间的多条曲线；
    根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束对所述多条曲线进行筛选处理，得到所述M条车道拓扑曲线。
  5. 根据权利要求4所述的方法，其特征在于，所述对所述至少一个起始点位姿向量和所述至少一个终点位姿向量进行曲线采样，得到所述驶入道路和所述驶出道路之间的多条曲线，包括：
    在所述驶入道路的每条驶入车道的末端和所述驶出道路的每条驶出车道的起始点之间生成多个控制点；
    根据所述至少一个起始点位姿向量、所述至少一个终点位姿向量以及所述多个控制点，生成所述多条平滑曲线。
  6. 根据权利要求4或5所述的方法，其特征在于，所述驶入道路中的至少一个起始点位姿向量是通过将所述每条驶入车道的末端延伸第二预设距离并进行采样得到的。
  7. 根据权利要求1至6任一项所述的方法，其特征在于，所述对所述M条车道拓扑曲线进行合理性检测处理，以得到K’条车道拓扑曲线，包括：
    根据所述驶入道路的方向向量、驶出道路的方向向量得到所述驶入道路、驶出道路之间的投影线，所述投影线为所述驶入道路的方向向量、驶出道路的方向向量相交所得到的夹角的平分线所在的直线，或者，所述投影线为垂直于所述驶出道路的方向向量且通过所述驶出道路的起始点的直线；
    计算所述驶入道路的每条驶入车道和所述驶出道路的每条驶出车道之间的对齐系数，其中，所述每条驶入车道和所述每条驶出车道之间的对齐系数为第一参数与第二参数之间的比值，所述第一参数为所述每条驶入车道的车道边线以及所述每条驶出车道的车道边线分别延长至所述投影线所得到的两条线段之间的重叠长度，所述第二参数为所述每条驶入车道的车道边线和所述每条驶出车道的车道边线分别延长至所述投影线所得到的两条线段中最短的线段的长度；
    根据所述驶入道路的每条驶入车道和所述驶出道路的每条驶出车道之间的对齐系数，得到K’条车道拓扑曲线，其中，所述K’条车道拓扑曲线包括以对齐系数大于第一预设阈值的驶入车道的末端、驶出车道的起始点为端点的曲线。
  8. 根据权利要求1至7任一项所述的方法，其特征在于，所述K’条车道拓扑曲线包括分别以所述驶入道路的最左侧驶入车道的末端、所述驶出道路的最左侧驶出车道的起始点为端点的曲线，还包括分别以所述驶入道路的最右侧驶入车道的末端、所述驶出道路的最右侧驶出车道的起始点为端点的曲线。
  9. 根据权利要求1至8任一项所述的方法，其特征在于，所述K’条车道拓扑曲线包括以所述驶入道路中的驶入车道X的末端以及所述驶出道路中的驶出车道Y的左侧车道的起始点为端点的车道拓扑曲线，以及以所述驶入车道X的末端以及所述驶出车道Y的右侧车道的起始点为端点的车道拓扑曲线，还包括以所述驶入车道X的末端以及所述驶出车道Y的起始点为端点的车道拓扑曲线，其中，所述驶出车道Y的左侧和右侧均存在车道；或者，
    所述K’条车道拓扑曲线包括以所述驶入道路中的驶入车道X的末端以及所述驶出道路中的驶出车道Y的左侧车道或者右侧车道的起始点为端点的车道拓扑曲线，还包括以所述驶入道路中的驶入车道X的末端、所述驶出车道Y的起始点为端点的车道拓扑曲线，其中，所述驶出道路中的驶出车道Y仅左侧或者右侧存在车道。
  10. 根据权利要求2至9任一项所述的方法，其特征在于，所述K’条车道拓扑曲线中每条车道拓扑曲线的最大曲率不大于第二预设阈值，且，所述每条车道拓扑曲线与所述车道拓扑曲线软边界约束、所述车道拓扑曲线硬边界约束之间的距离不小于第三预设距离，且，任意两条车道拓扑曲线之间的距离不小于第四预设距离。
  11. 根据权利要求1至10任一项所述的方法，其特征在于，所述方法还包括：
    计算所述K’条车道拓扑曲线中每条车道拓扑曲线的评价值，所述评价值与所述车道拓扑曲线曲率、曲率变化率以及斜穿车道数量、以及所述车道拓扑曲线对应的车道的车道交汇信息、交通规则信息、车流量估计值、可行驶距离中的至少一项有关；
    所述当所述车辆位于所述驶入道路中时，从所述K’条车道拓扑曲线中确定目标路径，包括：
    当所述车辆位于所述驶入道路中时，根据所述K’条车道拓扑曲线中每条车道拓扑曲线的评价值确定所述目标路径。
  12. 根据权利要求1至11任一项所述的方法，其特征在于，所述路口包括交叉路口、环岛、待转区路口、小S弯、高架出入口、多车道且无车道标线路段、连续转弯路口、窄道掉头路口中的至少一种。
  13. 一种基于路口的地图生成方法，其特征在于，包括：
    根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及车道线生成M条车道拓扑曲线，所述车道拓扑曲线为以所述驶入道路的驶入车道的末端和所述驶出道路的驶出车道的起始点为端点的曲线；
    对所述M条车道拓扑曲线进行合理性检测处理，以得到K’条车道拓扑曲线，其中，K’不大于M；
    根据所述路口内的K’条车道拓扑曲线生成所述路口的地图。
  14. 根据权利要求13所述的方法，其特征在于，所述根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及车道线生成M条车道拓扑曲线，包括：
    根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及不可跨越车道线得到车道拓扑曲线硬边界约束；
    根据所述路口、所述路口的驶入道路、所述路口的驶出道路中的可跨越车道线得到车道拓扑曲线软边界约束；
    根据所述车道拓扑曲线硬边界约束和所述车道拓扑曲线软边界约束得到K个车道拓扑曲线虚拟边界约束；
    根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束生成所述M条车道拓扑曲线。
  15. 根据权利要求14所述的方法，其特征在于，所述K个车道拓扑曲线虚拟边界约束与K个车道拓扑对应，所述K个车道拓扑曲线虚拟边界约束中任一车道拓扑曲线虚拟边界约束A是通过将所述路口的最左侧车道拓扑左侧的硬边界约束和/或软边界约束向右平移第一预设 距离，以及将所述路口的最右侧车道拓扑右侧的硬边界约束和/或软边界约束向左平移第一预设距离得到的，其中，所述第一预设距离是根据车道拓扑A’的车道位序确定的，或者，所述第一预设距离是根据所述车辆预设通过宽度、车道宽度中的至少一项以及所述车道拓扑A’的车道位序确定的，所述车道拓扑曲线虚拟边界约束A与所述车道拓扑A’对应，所述K个车道拓扑包括所述路口的最左侧车道拓扑和所述最右侧车道拓扑。
  16. 根据权利要求14或15所述的方法，其特征在于，所述根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束生成所述M条车道拓扑曲线，包括：
    分别对所述驶入道路中每条驶入车道的末端进行角度采样得到所述驶入道路中的至少一个起始点位姿向量，并分别对所述驶出道路中每条驶出车道的起始点进行角度采样得到所述驶出道路中的至少一个终点位姿向量；
    对所述至少一个起始点位姿向量和所述至少一个终点位姿向量进行曲线采样，得到所述驶入道路和所述驶出道路之间的多条曲线；
    根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束对所述多条曲线进行筛选处理，得到所述M条车道拓扑曲线。
  17. 根据权利要求16所述的方法，其特征在于，所述对所述至少一个起始点位姿向量和所述至少一个终点位姿向量进行曲线采样，得到所述驶入道路和所述驶出道路之间的多条曲线，包括：
    在所述驶入道路的每条驶入车道的末端和所述驶出道路的每条驶出车道的起始点之间生成多个控制点；
    根据所述至少一个起始点位姿向量、所述至少一个终点位姿向量以及所述多个控制点，生成所述多条平滑曲线。
  18. 根据权利要求16或17所述的方法，其特征在于，所述驶入道路中的至少一个起始点位姿向量是通过将所述每条驶入车道的末端延伸第二预设距离并进行采样得到的。
  19. 根据权利要求13至18任一项所述的方法，其特征在于，所述对所述M条车道拓扑曲线进行合理性检测处理，以得到K’条车道拓扑曲线，包括：
    根据所述驶入道路的方向向量、驶出道路的方向向量得到所述驶入道路、驶出道路之间的投影线，所述投影线为所述驶入道路的方向向量、驶出道路的方向向量相交所得到的夹角的平分线所在的直线，或者，所述投影线为垂直于所述驶出道路的方向向量且通过所述驶出道路的起始点的直线；
    计算所述驶入道路的每条驶入车道和所述驶出道路的每条驶出车道之间的对齐系数，其中，所述每条驶入车道和所述每条驶出车道之间的对齐系数为第一参数与第二参数之间的比值，所述第一参数为所述每条驶入车道的车道边线以及所述每条驶出车道的车道边线分别延长至所述投影线所得到的两条线段之间的重叠长度，所述第二参数为所述每条驶入车道的车道边线和所述每条驶出车道的车道边线分别延长至所述投影线所得到的两条线段中最短的线段的长度；
    根据所述驶入道路的每条驶入车道和所述驶出道路的每条驶出车道之间的对齐系数，得 到K’条车道拓扑曲线，其中，所述K’条车道拓扑曲线包括以对齐系数大于第一预设阈值的驶入车道的末端、驶出车道的起始点为端点的曲线。
  20. 根据权利要求13至19任一项所述的方法，其特征在于，所述K’条车道拓扑曲线包括分别以所述驶入道路的最左侧驶入车道的末端、所述驶出道路的最左侧驶出车道的起始点为端点的曲线，还包括分别以所述驶入道路的最右侧驶入车道的末端、所述驶出道路的最右侧驶出车道的起始点为端点的曲线。
  21. 根据权利要求13至20任一项所述的方法，其特征在于，所述K’条车道拓扑曲线包括以所述驶入道路中的驶入车道X的末端以及所述驶出道路中的驶出车道Y的左侧车道的起始点为端点的车道拓扑曲线，以及以所述驶入车道X的末端以及所述驶出车道Y的右侧车道的起始点为端点的车道拓扑曲线，还包括以所述驶入车道X的末端以及所述驶出车道Y的起始点为端点的车道拓扑曲线，其中，所述驶出车道Y的左侧和右侧均存在车道；或者，
    所述K’条车道拓扑曲线包括以所述驶入道路中的驶入车道X的末端以及所述驶出道路中的驶出车道Y的左侧车道或者右侧车道的起始点为端点的车道拓扑曲线，还包括以所述驶入道路中的驶入车道X的末端、所述驶出车道Y的起始点为端点的车道拓扑曲线，其中，所述驶出道路中的驶出车道Y仅左侧或者右侧存在车道。
  22. 根据权利要求14至21任一项所述的方法，其特征在于，所述K’条车道拓扑曲线中每条车道拓扑曲线的最大曲率不大于第二预设阈值，且，所述每条车道拓扑曲线与所述车道拓扑曲线软边界约束、所述车道拓扑曲线硬边界约束之间的距离不小于第三预设距离，且，任意两条车道拓扑曲线之间的距离不小于第四预设距离。
  23. 根据权利要求13至22任一项所述的方法，其特征在于，所述路口包括交叉路口、环岛、待转区路口、小S弯、高架出入口、多车道且无车道标线路段、连续转弯路口、窄道掉头路口中的至少一种。
  24. 一种引导车辆行驶的装置，其特征在于，包括：
    曲线生成模块，用于根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及车道线生成M条车道拓扑曲线，所述车道拓扑曲线为以所述驶入道路的驶入车道的末端和所述驶出道路的驶出车道的起始点为端点的曲线；
    检测处理模块，用于对所述M条车道拓扑曲线进行合理性检测处理，以得到K’条车道拓扑曲线，其中，K’不大于M；
    确定模块，用于当所述车辆位于所述驶入道路中时，从所述K’条车道拓扑曲线中确定目标路径。
  25. 根据权利要求24所述的装置，其特征在于，所述曲线生成模块，用于：
    根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及不可跨越车道线得到车道拓扑曲线硬边界约束；
    根据所述路口、所述路口的驶入道路、所述路口的驶出道路中的可跨越车道线得到车道拓扑曲线软边界约束；
    根据所述车道拓扑曲线硬边界约束和所述车道拓扑曲线软边界约束得到K个车道拓扑曲线虚拟边界约束；
    根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束生成所述M条车道拓扑曲线。
  26. 根据权利要求25所述的装置，其特征在于，所述K个车道拓扑曲线虚拟边界约束与K个车道拓扑对应，所述K个车道拓扑曲线虚拟边界约束中任一车道拓扑曲线虚拟边界约束A是通过将所述路口的最左侧车道拓扑左侧的硬边界约束和/或软边界约束向右平移第一预设距离，以及将所述路口的最右侧车道拓扑右侧的硬边界约束和/或软边界约束向左平移第一预设距离得到的，其中，所述第一预设距离是根据车道拓扑A’的车道位序确定的，或者，所述第一预设距离是根据所述车辆预设通过宽度、车道宽度中的至少一项以及所述车道拓扑A’的车道位序确定的，所述车道拓扑曲线虚拟边界约束A与所述车道拓扑A’对应，所述K个车道拓扑包括所述路口的最左侧车道拓扑和所述最右侧车道拓扑。
  27. 根据权利要求25或26所述的装置，其特征在于，所述曲线生成模块，还用于：
    分别对所述驶入道路中每条驶入车道的末端进行角度采样得到所述驶入道路中的至少一个起始点位姿向量，并分别对所述驶出道路中每条驶出车道的起始点进行角度采样得到所述驶出道路中的至少一个终点位姿向量；
    对所述至少一个起始点位姿向量和所述至少一个终点位姿向量进行曲线采样，得到所述驶入道路和所述驶出道路之间的多条曲线；
    根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束对所述多条曲线进行筛选处理，得到所述M条车道拓扑曲线。
  28. 根据权利要求27所述的装置，其特征在于，所述曲线生成模块，还用于：
    在所述驶入道路的每条驶入车道的末端和所述驶出道路的每条驶出车道的起始点之间生成多个控制点；
    根据所述至少一个起始点位姿向量、所述至少一个终点位姿向量以及所述多个控制点，生成所述多条平滑曲线。
  29. 根据权利要求27或28所述的装置，其特征在于，所述驶入道路中的至少一个起始点位姿向量是通过将所述每条驶入车道的末端延伸第二预设距离并进行采样得到的。
  30. 根据权利要求24至29任一项所述的装置，其特征在于，所述检测处理模块，用于：
    根据所述驶入道路的方向向量、驶出道路的方向向量得到所述驶入道路、驶出道路之间的投影线，所述投影线为所述驶入道路的方向向量、驶出道路的方向向量相交所得到的夹角的平分线所在的直线，或者，所述投影线为垂直于所述驶出道路的方向向量且通过所述驶出道路的起始点的直线；
    计算所述驶入道路的每条驶入车道和所述驶出道路的每条驶出车道之间的对齐系数，其中，所述每条驶入车道和所述每条驶出车道之间的对齐系数为第一参数与第二参数之间的比值，所述第一参数为所述每条驶入车道的车道边线以及所述每条驶出车道的车道边线分别延长至所述投影线所得到的两条线段之间的重叠长度，所述第二参数为所述每条驶入车道的车 道边线和所述每条驶出车道的车道边线分别延长至所述投影线所得到的两条线段中最短的线段的长度；
    根据所述驶入道路的每条驶入车道和所述驶出道路的每条驶出车道之间的对齐系数，得到K’条车道拓扑曲线，其中，所述K’条车道拓扑曲线包括以对齐系数大于第一预设阈值的驶入车道的末端、驶出车道的起始点为端点的曲线。
  31. 根据权利要求24至30任一项所述的装置，其特征在于，所述K’条车道拓扑曲线包括分别以所述驶入道路的最左侧驶入车道的末端、所述驶出道路的最左侧驶出车道的起始点为端点的曲线，还包括分别以所述驶入道路的最右侧驶入车道的末端、所述驶出道路的最右侧驶出车道的起始点为端点的曲线。
  32. 根据权利要求24至31任一项所述的装置，其特征在于，所述K’条车道拓扑曲线包括以所述驶入道路中的驶入车道X的末端以及所述驶出道路中的驶出车道Y的左侧车道的起始点为端点的车道拓扑曲线，以及以所述驶入车道X的末端以及所述驶出车道Y的右侧车道的起始点为端点的车道拓扑曲线，还包括以所述驶入车道X的末端以及所述驶出车道Y的起始点为端点的车道拓扑曲线，其中，所述驶出车道Y的左侧和右侧均存在车道；或者，
    所述K’条车道拓扑曲线包括以所述驶入道路中的驶入车道X的末端以及所述驶出道路中的驶出车道Y的左侧车道或者右侧车道的起始点为端点的车道拓扑曲线，还包括以所述驶入道路中的驶入车道X的末端、所述驶出车道Y的起始点为端点的车道拓扑曲线，其中，所述驶出道路中的驶出车道Y仅左侧或者右侧存在车道。
  33. 根据权利要求25至32任一项所述的装置，其特征在于，所述K’条车道拓扑曲线中每条车道拓扑曲线的最大曲率不大于第二预设阈值，且，所述每条车道拓扑曲线与所述车道拓扑曲线软边界约束、所述车道拓扑曲线硬边界约束之间的距离不小于第三预设距离，且，任意两条车道拓扑曲线之间的距离不小于第四预设距离。
  34. 根据权利要求24至33任一项所述的装置，其特征在于，所述装置还包括评价模块，用于：
    计算所述K’条车道拓扑曲线中每条车道拓扑曲线的评价值，所述评价值与所述车道拓扑曲线曲率、曲率变化率以及斜穿车道数量、以及所述车道拓扑曲线对应的车道的车道交汇信息、交通规则信息、车流量估计值、可行驶距离中的至少一项有关；
    所述确定模块，用于：
    当所述车辆位于所述驶入道路中时，根据所述K’条车道拓扑曲线中每条车道拓扑曲线的评价值确定所述目标路径。
  35. 根据权利要求24至34任一项所述的装置，其特征在于，所述路口包括交叉路口、环岛、待转区路口、小S弯、高架出入口、多车道且无车道标线路段、连续转弯路口、窄道掉头路口中的至少一种。
  36. 一种基于路口的地图生成装置，其特征在于，包括：
    曲线生成模块，用于根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物 以及车道线生成M条车道拓扑曲线，所述车道拓扑曲线为以所述驶入道路的驶入车道的末端和所述驶出道路的驶出车道的起始点为端点的曲线；
    检测处理模块，用于对所述M条车道拓扑曲线进行合理性检测处理，以得到K’条车道拓扑曲线，其中，K’不大于M；
    地图生成模块，用于根据所述路口内的K’条车道拓扑曲线生成所述路口的地图。
  37. 根据权利要求36所述的装置，其特征在于，所述曲线生成模块，用于：
    根据路口、所述路口的驶入道路、所述路口的驶出道路中的障碍物以及不可跨越车道线得到车道拓扑曲线硬边界约束；
    根据所述路口、所述路口的驶入道路、所述路口的驶出道路中的可跨越车道线得到车道拓扑曲线软边界约束；
    根据所述车道拓扑曲线硬边界约束和所述车道拓扑曲线软边界约束得到K个车道拓扑曲线虚拟边界约束；
    根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束生成所述M条车道拓扑曲线。
  38. 根据权利要求37所述的装置，其特征在于，所述K个车道拓扑曲线虚拟边界约束与K个车道拓扑对应，所述K个车道拓扑曲线虚拟边界约束中任一车道拓扑曲线虚拟边界约束A是通过将所述路口的最左侧车道拓扑左侧的硬边界约束和/或软边界约束向右平移第一预设距离，以及将所述路口的最右侧车道拓扑右侧的硬边界约束和/或软边界约束向左平移第一预设距离得到的，其中，所述第一预设距离是根据车道拓扑A’的车道位序确定的，或者，所述第一预设距离是根据所述车辆预设通过宽度、车道宽度中的至少一项以及所述车道拓扑A’的车道位序确定的，所述车道拓扑曲线虚拟边界约束A与所述车道拓扑A’对应，所述K个车道拓扑包括所述路口的最左侧车道拓扑和所述最右侧车道拓扑。
  39. 根据权利要求37或38所述的装置，其特征在于，所述曲线生成模块，还用于：
    分别对所述驶入道路中每条驶入车道的末端进行角度采样得到所述驶入道路中的至少一个起始点位姿向量，并分别对所述驶出道路中每条驶出车道的起始点进行角度采样得到所述驶出道路中的至少一个终点位姿向量；
    对所述至少一个起始点位姿向量和所述至少一个终点位姿向量进行曲线采样，得到所述驶入道路和所述驶出道路之间的多条曲线；
    根据所述车道拓扑曲线硬边界约束、所述车道拓扑曲线软边界约束和所述K个车道拓扑曲线虚拟边界约束对所述多条曲线进行筛选处理，得到所述M条车道拓扑曲线。
  40. 根据权利要求39所述的装置，其特征在于，所述曲线生成模块，还用于：
    在所述驶入道路的每条驶入车道的末端和所述驶出道路的每条驶出车道的起始点之间生成多个控制点；
    根据所述至少一个起始点位姿向量、所述至少一个终点位姿向量以及所述多个控制点，生成所述多条平滑曲线。
  41. 根据权利要求39或40所述的装置，其特征在于，所述驶入道路中的至少一个起始点 位姿向量是通过将所述每条驶入车道的末端延伸第二预设距离并进行采样得到的。
  42. 根据权利要求36至41任一项所述的装置，其特征在于，所述检测处理模块，用于：
    根据所述驶入道路的方向向量、驶出道路的方向向量得到所述驶入道路、驶出道路之间的投影线，所述投影线为所述驶入道路的方向向量、驶出道路的方向向量相交所得到的夹角的平分线所在的直线，或者，所述投影线为垂直于所述驶出道路的方向向量且通过所述驶出道路的起始点的直线；
    计算所述驶入道路的每条驶入车道和所述驶出道路的每条驶出车道之间的对齐系数，其中，所述每条驶入车道和所述每条驶出车道之间的对齐系数为第一参数与第二参数之间的比值，所述第一参数为所述每条驶入车道的车道边线以及所述每条驶出车道的车道边线分别延长至所述投影线所得到的两条线段之间的重叠长度，所述第二参数为所述每条驶入车道的车道边线和所述每条驶出车道的车道边线分别延长至所述投影线所得到的两条线段中最短的线段的长度；
    根据所述驶入道路的每条驶入车道和所述驶出道路的每条驶出车道之间的对齐系数，得到K’条车道拓扑曲线，其中，所述K’条车道拓扑曲线包括以对齐系数大于第一预设阈值的驶入车道的末端、驶出车道的起始点为端点的曲线。
  43. 根据权利要求36至42任一项所述的装置，其特征在于，所述K’条车道拓扑曲线包括分别以所述驶入道路的最左侧驶入车道的末端、所述驶出道路的最左侧驶出车道的起始点为端点的曲线，还包括分别以所述驶入道路的最右侧驶入车道的末端、所述驶出道路的最右侧驶出车道的起始点为端点的曲线。
  44. 根据权利要求36至43任一项所述的装置，其特征在于，所述K’条车道拓扑曲线包括以所述驶入道路中的驶入车道X的末端以及所述驶出道路中的驶出车道Y的左侧车道的起始点为端点的车道拓扑曲线，以及以所述驶入车道X的末端以及所述驶出车道Y的右侧车道的起始点为端点的车道拓扑曲线，还包括以所述驶入车道X的末端以及所述驶出车道Y的起始点为端点的车道拓扑曲线，其中，所述驶出车道Y的左侧和右侧均存在车道；或者，
    所述K’条车道拓扑曲线包括以所述驶入道路中的驶入车道X的末端以及所述驶出道路中的驶出车道Y的左侧车道或者右侧车道的起始点为端点的车道拓扑曲线，还包括以所述驶入道路中的驶入车道X的末端、所述驶出车道Y的起始点为端点的车道拓扑曲线，其中，所述驶出道路中的驶出车道Y仅左侧或者右侧存在车道。
  45. 根据权利要求37至44任一项所述的装置，其特征在于，所述K’条车道拓扑曲线中每条车道拓扑曲线的最大曲率不大于第二预设阈值，且，所述每条车道拓扑曲线与所述车道拓扑曲线软边界约束、所述车道拓扑曲线硬边界约束之间的距离不小于第三预设距离，且，任意两条车道拓扑曲线之间的距离不小于第四预设距离。
  46. 根据权利要求36至45任一项所述的装置，其特征在于，所述路口包括交叉路口、环岛、待转区路口、小S弯、高架出入口、多车道且无车道标线路段、连续转弯路口、窄道掉头路口中的至少一种。
  47. 一种引导车辆行驶的装置，其特征在于，包括处理器和存储器；其中，所述存储器用于存储程序代码，所述处理器用于调用所述程序代码，以执行如权利要求1至12任意一项所述的方法。
  48. 一种基于路口的地图生成装置，其特征在于，包括处理器和存储器；其中，所述存储器用于存储程序代码，所述处理器用于调用所述程序代码，以执行如权利要求13至23任意一项所述的方法。
  49. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质存储有计算机程序，所述计算机程序被处理器执行以实现权利要求1至12任意一项所述的方法，和/或如权利要求13至23任意一项所述的方法。
  50. 一种计算机程序产品，其特征在于，当计算机程序产品在计算机上运行时，使得所述计算机执行如权利要求1至12任意一项所述的方法，和/或如权利要求13至23任意一项所述的方法。
  51. 一种芯片系统，其特征在于，所述芯片系统应用于电子设备；所述芯片系统包括一个或多个接口电路，以及一个或多个处理器；所述接口电路和所述处理器通过线路互联；所述接口电路用于从所述电子设备的存储器接收信号，并向所述处理器发送所述信号，所述信号包括所述存储器中存储的计算机指令；当所述处理器执行所述计算机指令时，所述电子设备执行如权利要求1至12任意一项所述的方法，和/或如权利要求13至23任意一项所述的方法。
  52. 一种智能驾驶车辆，其特征在于，包括行进系统、传感系统、控制系统和计算机系统，其中，所述计算机系统用于执行如权利要求1至12任意一项所述的方法，和/或如权利要求13至23任意一项所述的方法。
PCT/CN2022/138461 2022-01-12 2022-12-12 引导车辆行驶的方法、地图生成方法及相关系统 WO2023134364A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210046464.0A CN114543825A (zh) 2022-01-12 2022-01-12 引导车辆行驶的方法、地图生成方法及相关系统
CN202210046464.0 2022-01-12

Publications (1)

Publication Number Publication Date
WO2023134364A1 true WO2023134364A1 (zh) 2023-07-20

Family

ID=81671114

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/138461 WO2023134364A1 (zh) 2022-01-12 2022-12-12 引导车辆行驶的方法、地图生成方法及相关系统

Country Status (2)

Country Link
CN (1) CN114543825A (zh)
WO (1) WO2023134364A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117495989A (zh) * 2023-12-29 2024-02-02 腾讯科技（深圳）有限公司 数据处理方法、装置、设备及可读存储介质

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114543825A (zh) * 2022-01-12 2022-05-27 华为技术有限公司 引导车辆行驶的方法、地图生成方法及相关系统
CN114777799B (zh) * 2022-06-15 2022-10-18 中汽创智科技有限公司 一种路口高精地图生成方法、装置、电子设备及存储介质
CN114998857B (zh) * 2022-08-04 2022-10-25 山东未来网络研究院(紫金山实验室工业互联网创新应用基地) 一种基于机器视觉的车道流量分类方法及系统
CN115457765A (zh) * 2022-08-29 2022-12-09 无锡艾立德智能科技有限公司 一种融合物联网加密传输的智能交通监控系统
CN115717899B (zh) * 2022-11-25 2024-04-30 北京斯年智驾科技有限公司 一种弯道路径选择方法和装置
CN117238140A (zh) * 2023-11-13 2023-12-15 广汽埃安新能源汽车股份有限公司 一种路口通行方法、装置、电子设备及存储介质
CN117553801A (zh) * 2024-01-08 2024-02-13 深圳市普渡科技有限公司 拓扑路径图生成方法、装置、机器人及存储介质

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105651295A (zh) * 2016-01-15 2016-06-08 武汉光庭信息技术股份有限公司 基于Bezier曲线构建交叉路口驶入驶出车道Link的连接曲线算法
CN107340772A (zh) * 2017-07-11 2017-11-10 清华大学 一种面向无人驾驶的拟人化参考轨迹规划方法
WO2019142439A1 (ja) * 2018-01-18 2019-07-25 株式会社デンソー 交差点内の走行軌道データ生成装置、交差点内の走行軌道データ生成プログラム及び記憶媒体
CN112068545A (zh) * 2020-07-23 2020-12-11 哈尔滨工业大学（深圳） 一种无人驾驶车辆在十字路口的行驶轨迹规划方法、系统及存储介质
US20200393253A1 (en) * 2019-06-11 2020-12-17 WeRide Corp. Method for generating road map for autonomous vehicle navigation
CN112937598A (zh) * 2021-03-31 2021-06-11 戴姆勒股份公司 用于评估和/或调整车辆驾驶辅助或自动驾驶功能的方法和设备
CN113895456A (zh) * 2021-09-08 2022-01-07 北京汽车研究总院有限公司 自动驾驶车辆的交叉路口行驶方法、装置、车辆及介质
CN114543825A (zh) * 2022-01-12 2022-05-27 华为技术有限公司 引导车辆行驶的方法、地图生成方法及相关系统

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103235848B (zh) * 2013-04-15 2016-03-30 中国科学院软件研究所 一种基于简化路网模型的轻量级路网匹配方法
US10137896B2 (en) * 2016-12-30 2018-11-27 Baidu Usa Llc Method and system for operating autonomous driving vehicles using graph-based lane change guide
CN108871368B (zh) * 2018-08-06 2020-12-29 武汉中海庭数据技术有限公司 一种高精度地图车道横向拓扑关系的构建方法、系统及存储器
CN110599570B (zh) * 2019-09-17 2022-09-20 武汉中海庭数据技术有限公司 基于众包数据开放场景下的地图生成方法及系统
CN110634291B (zh) * 2019-09-17 2020-12-29 武汉中海庭数据技术有限公司 基于众包大数据的高精度地图拓扑自动构建方法及系统
CN112699708A (zh) * 2019-10-22 2021-04-23 北京初速度科技有限公司 一种车道级拓扑网的生成方法及装置
CN111142525B (zh) * 2019-12-31 2023-04-25 武汉中海庭数据技术有限公司 高精度地图车道拓扑构建方法及系统、服务器及介质
CN113074744A (zh) * 2021-03-18 2021-07-06 重庆数字城市科技有限公司 一种智能网联车用地图拓扑连线生成方法
CN113670308B (zh) * 2021-07-14 2024-01-02 华为技术有限公司 引导车辆行驶的方法及相关系统、存储介质

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105651295A (zh) * 2016-01-15 2016-06-08 武汉光庭信息技术股份有限公司 基于Bezier曲线构建交叉路口驶入驶出车道Link的连接曲线算法
CN107340772A (zh) * 2017-07-11 2017-11-10 清华大学 一种面向无人驾驶的拟人化参考轨迹规划方法
WO2019142439A1 (ja) * 2018-01-18 2019-07-25 株式会社デンソー 交差点内の走行軌道データ生成装置、交差点内の走行軌道データ生成プログラム及び記憶媒体
US20200393253A1 (en) * 2019-06-11 2020-12-17 WeRide Corp. Method for generating road map for autonomous vehicle navigation
CN112068545A (zh) * 2020-07-23 2020-12-11 哈尔滨工业大学（深圳） 一种无人驾驶车辆在十字路口的行驶轨迹规划方法、系统及存储介质
CN112937598A (zh) * 2021-03-31 2021-06-11 戴姆勒股份公司 用于评估和/或调整车辆驾驶辅助或自动驾驶功能的方法和设备
CN113895456A (zh) * 2021-09-08 2022-01-07 北京汽车研究总院有限公司 自动驾驶车辆的交叉路口行驶方法、装置、车辆及介质
CN114543825A (zh) * 2022-01-12 2022-05-27 华为技术有限公司 引导车辆行驶的方法、地图生成方法及相关系统

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117495989A (zh) * 2023-12-29 2024-02-02 腾讯科技（深圳）有限公司 数据处理方法、装置、设备及可读存储介质
CN117495989B (zh) * 2023-12-29 2024-04-19 腾讯科技（深圳）有限公司 数据处理方法、装置、设备及可读存储介质

Also Published As

Publication number Publication date
CN114543825A (zh) 2022-05-27

Similar Documents

Publication Publication Date Title
WO2023134364A1 (zh) 引导车辆行驶的方法、地图生成方法及相关系统
CN110487279B (zh) 一种基于改进a＊算法的路径规划方法
KR102094622B1 (ko) 씬 평가 장치, 주행 지원 장치, 씬 평가 방법
KR102092484B1 (ko) 씬 평가 장치, 주행 지원 장치, 씬 평가 방법
CN107851372A (zh) 驾驶计划装置、行驶辅助装置、驾驶计划方法
CN107851375A (zh) 驾驶计划装置、行驶辅助装置、驾驶计划方法
US20230159056A1 (en) Method and apparatus for planning obstacle avoidance path of traveling apparatus
CN108932856B (zh) 一种自动驾驶下交叉口通行权设置方法
WO2022227878A1 (zh) 一种车道线标注方法及装置
CN109084798A (zh) 网络下发带有道路属性的控制点的路径规划方法
CN104615897B (zh) 基于低频gps数据的路段行程时间估算方法
CN113619602B (zh) 引导车辆行驶的方法及相关系统、存储介质
CN109415877A (zh) 具有协同作用的交通交叉路口
CN112071068B (zh) 基于对称交叉口的行人过街效率与安全分析方法
CN113313957A (zh) 基于增强型Dijkstra算法的无信号灯交叉路口车辆调度方法
CN114715192A (zh) 一种解耦的自动驾驶车辆实时轨迹规划方法、装置及系统
CN112288854B (zh) 立交桥三维模型的构建方法
CN111551185B (zh) 行车道添加的方法
WO2020233222A1 (zh) 无人车路径控制方法、装置和系统
WO2024037262A1 (zh) 一种机器人通过窄道的导航方法、芯片及机器人
CN107851374A (zh) 场景评估装置、行驶辅助装置、场景评估方法
CN115638804B (zh) 一种无死锁的无人车辆在线路径规划方法
CN115220461B (zh) 室内复杂环境下机器人单体系统及多机器人交互协同方法
CN116839609A (zh) 全覆盖路径规划方法、装置及计算机可读存储介质
WO2023274306A1 (zh) 一种路径规划的方法、服务器和车辆

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22919997

Country of ref document: EP

Kind code of ref document: A1