WO2021142799A1 - 路径选择方法和路径选择装置 - Google Patents

路径选择方法和路径选择装置 Download PDF

Info

Publication number
WO2021142799A1
WO2021142799A1 PCT/CN2020/072850 CN2020072850W WO2021142799A1 WO 2021142799 A1 WO2021142799 A1 WO 2021142799A1 CN 2020072850 W CN2020072850 W CN 2020072850W WO 2021142799 A1 WO2021142799 A1 WO 2021142799A1
Authority
WO
WIPO (PCT)
Prior art keywords
obstacle
path
risk
road
candidate
Prior art date
Application number
PCT/CN2020/072850
Other languages
English (en)
French (fr)
Inventor
刘亚林
曹昊天
宋晓琳
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
Priority to EP20914201.7A priority Critical patent/EP4089369A4/en
Priority to CN202080004767.5A priority patent/CN112639849A/zh
Priority to PCT/CN2020/072850 priority patent/WO2021142799A1/zh
Publication of WO2021142799A1 publication Critical patent/WO2021142799A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G06Q10/047—Optimisation of routes or paths, e.g. travelling salesman problem
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W60/00—Drive control systems specially adapted for autonomous road vehicles
    • B60W60/001—Planning or execution of driving tasks
    • B60W60/0011—Planning or execution of driving tasks involving control alternatives for a single driving scenario, e.g. planning several paths to avoid obstacles
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0635—Risk analysis of enterprise or organisation activities
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/40—Business processes related to the transportation industry
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W2554/00—Input parameters relating to objects
    • B60W2554/40—Dynamic objects, e.g. animals, windblown objects
    • B60W2554/404—Characteristics
    • B60W2554/4041—Position
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W2554/00—Input parameters relating to objects
    • B60W2554/40—Dynamic objects, e.g. animals, windblown objects
    • B60W2554/404—Characteristics
    • B60W2554/4042—Longitudinal speed
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W2554/00—Input parameters relating to objects
    • B60W2554/80—Spatial relation or speed relative to objects
    • B60W2554/805—Azimuth angle

Definitions

  • This application relates to the field of unmanned driving technology, and in particular to a route selection method and route selection device.
  • Unmanned driving is an important part of intelligent transportation system. After the unmanned vehicle (hereinafter referred to as the unmanned vehicle) receives various perception information from the sensors, it analyzes the current environment and then issues instructions to the underlying control module. This process is the main task of the decision-making and planning module.
  • Typical unmanned vehicle decision-making planning includes: global path planning, behavioral decision-making and path planning. Among them, during path planning, a planning algorithm is usually used to generate multiple candidate paths on the road surface, and then the best obstacle avoidance path is selected according to the geometric properties of the candidate paths and the location of obstacles. Please refer to the schematic diagram of path selection in FIG. 1.
  • the existing obstacle avoidance path selection mechanism judges whether the candidate path has an intersection with the obstacle by constructing an inequality, and performs collision detection on the candidate paths one by one. If there is an intersection between the path and the obstacle, then discard the path.
  • the embodiment of the present application provides a path selection method for determining an optimal obstacle avoidance path from multiple candidate paths, which can improve path selection efficiency.
  • the first aspect of the embodiments of the present application provides a route selection method, including: obtaining basic road information of a target road area, location information of obstacles in the target road area, and the number of target objects in the target road area.
  • Candidate paths determine the obstacle risk relationship of the target road area according to the basic information of the road and the location information of the obstacle, and the obstacle risk relationship is used to obtain the information of any position in the target road area Obstacle risk; determine the comprehensive obstacle risk of each candidate path according to the obstacle risk relationship and the position distribution of each candidate path in the multiple candidate paths in the target road area; according to each candidate path
  • the comprehensive obstacle risk of the path, and the target path is determined from the multiple candidate paths.
  • the multiple candidate paths are determined based on the position of the target object on the target road area.
  • the obstacle risk relationship in the target road area can be constructed through the basic road information and the position information of the obstacles. Therefore, the comprehensive obstacle risk of multiple candidate paths can be determined, thereby Directly obtaining the target path with lower obstacle risk from multiple candidate paths can improve the efficiency of path selection.
  • the multiple candidate paths include a first candidate path; the obstacle risk relationship and each candidate path in the multiple candidate paths are at the target
  • the location distribution in the road area to determine the comprehensive obstacle risk of each candidate path includes: determining a plurality of path points on the first candidate path; determining the obstacle risk of the plurality of path points according to the obstacle risk relationship ; Determine the comprehensive obstacle risk of the first candidate path according to the obstacle risk of the multiple path points.
  • the path selection method provided by the embodiments of the present application can determine the comprehensive obstacle risk of the entire candidate path based on the obstacle risk of the multiple path points by sampling multiple path points on the candidate path, thereby reducing the amount of calculation and improving Path selection speed.
  • the obstacle risk of the first waypoint is determined according to the lateral obstacle risk relationship of the first waypoint and the longitudinal obstacle risk relationship of the first waypoint
  • the horizontal obstacle risk relationship of the first waypoint is determined according to the position information of the obstacle and the obstacle horizontal risk magnification coefficient; the longitudinal obstacle risk relationship of the first waypoint is based on the position information of the obstacle And the longitudinal risk magnification factor of obstacles is determined.
  • the lateral risk magnification coefficient and the longitudinal risk magnification coefficient are preset parameters, which are used to respectively represent the influence of the lateral obstacle and the longitudinal obstacle on the risk of each point in the road.
  • the lateral obstacle risk of the first waypoint is positively correlated with the lateral risk amplification factor, and is negatively correlated with the distance between the first waypoint and the obstacle in the lateral direction of the road perpendicular to the road reference line;
  • the longitudinal obstacle risk of the first waypoint is determined according to the preset obstacle longitudinal risk amplification coefficient, and the longitudinal obstacle risk of the first waypoint is the same as that of the The longitudinal risk magnification coefficient is positively correlated with the distance between the first way point and the obstacle in the longitudinal direction of the road along the road reference line direction, and the obstacle lateral risk magnification coefficient is smaller than the obstacle longitudinal risk Enlargement factor;
  • the obstacle risk of the first waypoint is determined according to the lateral obstacle risk of the first waypoint and the longitudinal obstacle risk of the first waypoint.
  • the obstacle risk relationship when constructing the obstacle risk relationship, according to the obstacles in the horizontal and vertical directions of the road, the threat to the target is different, and the horizontal risk amplification factor and the vertical risk amplification factor are set, where, The longitudinal risk magnification factor of the obstacle is greater than the lateral risk magnification factor of the obstacle. Therefore, the obstacle risk corresponding to each position in the determined obstacle risk relationship can be closer to the actual scene.
  • the method further includes: obtaining a heading angle of the obstacle, the heading angle of the obstacle is obtained according to the direction of movement of the obstacle; the first path The obstacle risk of the point is determined according to the heading angle of the obstacle, the horizontal obstacle risk relationship of the first waypoint and the longitudinal obstacle risk relationship of the first waypoint.
  • the obstacle risk of the first waypoint is inversely related to the magnitude of a first angle, and the first angle is a ray passing through the first waypoint with the obstacle as an endpoint and a horizontal axis of the ground coordinate system The difference between the angle of and the heading angle of the obstacle.
  • the current heading angle of the obstacle belt can also be considered.
  • the obstacle is usually a moving object, such as a moving vehicle, refer to its moving direction and consider the obstacle.
  • the dynamic risk of the obstacle, the obstacle risk relationship obtained from this can be adjusted adaptively with the attitude of the obstacle. As a result, the generated obstacle risk relationship can be closer to the risk in the actual scene, and the obstacle risk prediction effect is better.
  • the obstacle is a traveling vehicle in the target road area.
  • the target object is a vehicle in the target road area
  • the obstacle is a vehicle other than the target object in the target road area
  • the road risk relationship of the target road area is determined according to the basic road information, and the road risk relationship is used to obtain a road at any position in the target road area.
  • the basic information of the road includes the length of the road, the width of the road, and the position of the road reference line; the position distribution of each candidate path in the target road area is determined according to the road risk relationship and the position distribution of each of the multiple candidate paths.
  • the comprehensive road risk of each candidate path in the multiple candidate paths where the comprehensive road risk is used to determine a target obstacle avoidance path from the candidate paths.
  • the route selection method provided in the embodiments of the present application also considers the risk of the road boundary.
  • the obstacle risk and the road risk are used to determine the driving risk and are used to select the target route, which can predict the risk more comprehensively.
  • the multiple candidate paths are determined according to the road risk relationship and the position distribution of each candidate path in the target road area in the multiple candidate paths
  • the comprehensive road risk of each candidate path in the multiple candidate paths includes: determining multiple path points on the first candidate path in the multiple candidate paths; determining the road risk of the multiple path points according to the road risk relationship; The road risks of multiple path points determine the comprehensive road risks of the first candidate path.
  • the road risk at the third location is positively correlated with the distance difference between the third location and the left and right borders of the road.
  • the path selection method provided by the embodiment of the present application determines the comprehensive road risk of each candidate path according to multiple path points sampled in the candidate path. The closer the road is to the road boundary, the higher the road risk.
  • the method further includes: the determining a target path from the multiple candidate paths according to the comprehensive obstacle risk of each candidate path includes: The path parameter and the comprehensive obstacle risk of each candidate path determine a target path from the multiple candidate paths.
  • the route parameters are also comprehensively considered.
  • the route parameters can be used for the path quality to determine the target obstacle avoidance path.
  • the path parameter includes at least one of the following: curvature, length, path consistency, curvature change rate, heading angle error, and the change rate of the heading angle error
  • the path Consistency is used to indicate the degree of consistency between the candidate route at the current moment and the candidate route at the previous planning moment
  • the heading angle error is used to indicate the difference between the heading angle of the target traveling along the candidate path and the heading angle traveling along the road reference line. The deviation between.
  • the path selection method provided in the embodiments of the present application can specifically design different combinations of path parameters to determine the target path, and the path parameters of different combinations can measure the quality of the path and meet the diversified requirements for path quality in practical applications.
  • a second aspect of the embodiments of the present application provides a path selection device, including: an acquisition module for acquiring basic road information in a target road area, location information of obstacles in the target road area, and target objects in the target road area. Multiple candidate paths in the target road area; a determining module, configured to determine the obstacle risk relationship in the target road area according to the basic road information and the location information of the obstacle, and the obstacle risk relationship is used to obtain The obstacle risk at any position in the target road area; the determining module is further configured to, according to the obstacle risk relationship and the position of each candidate path in the multiple candidate paths in the target road area The position distribution determines the comprehensive obstacle risk of each candidate path; the determining module is further configured to determine a target path from the multiple candidate paths according to the comprehensive obstacle risk of each candidate path.
  • the multiple candidate paths include a first candidate path; the determining module is specifically configured to: determine multiple path points on the first candidate path; The obstacle risk relationship determines the obstacle risk of the multiple way points; the comprehensive obstacle risk of the first candidate path is determined according to the obstacle risks of the multiple way points.
  • the multiple waypoints include a first waypoint; the obstacle risk of the first waypoint is based on the lateral obstacle risk relationship of the first waypoint and the The longitudinal obstacle risk relationship of the first waypoint is determined; the horizontal obstacle risk relationship of the first waypoint is determined according to the position information of the obstacle and the obstacle's lateral risk magnification coefficient; the longitudinal direction of the first waypoint The obstacle risk relationship is determined according to the position information of the obstacle and the obstacle longitudinal risk amplification coefficient.
  • the lateral obstacle risk of the first waypoint is positively correlated with the lateral risk amplification coefficient, and is negatively related to the distance between the first waypoint and the obstacle in the lateral direction of the road perpendicular to the road reference line.
  • the longitudinal obstacle risk of the first waypoint is positively correlated with the longitudinal risk magnification coefficient, and is negatively correlated with the distance between the first waypoint and the obstacle in the longitudinal direction of the road along the road reference line direction .
  • the lateral risk magnification factor of the obstacle is smaller than the longitudinal risk magnification factor of the obstacle.
  • the acquisition module is further configured to: acquire the heading angle of the obstacle, and the heading angle of the obstacle is obtained according to the movement direction of the obstacle;
  • the obstacle risk of a waypoint is determined according to the heading angle of the obstacle, and the horizontal obstacle risk relationship of the first waypoint and the longitudinal obstacle risk relationship of the first waypoint.
  • the obstacle risk of the first waypoint is inversely related to the magnitude of the first angle, and the first angle is the angle between the ray passing through the first waypoint with the obstacle as the endpoint and the horizontal axis of the ground coordinate system. The difference between the heading angles of the obstacles.
  • the obstacle is a traveling vehicle in the target road area.
  • the target object is a vehicle in the target road area
  • the obstacle is a vehicle other than the target object in the target road area
  • the determining module is further configured to: the determining module is further configured to determine the road risk relationship of the target road area according to the basic road information, and the road risk The relationship is used to obtain the road risk of any position in the target road area, and the basic road information includes the road length, the road width and the position of the road reference line; according to the road risk relationship and the multiple candidate paths The position distribution of each candidate route in the target road area determines the comprehensive road risk of each candidate route in the multiple candidate routes, and the comprehensive road risk is used to determine the target obstacle avoidance route from the candidate routes .
  • the determining module is specifically configured to: determine multiple path points on the first candidate path among the multiple candidate paths; determine the multiple path points according to the road risk relationship The road risk of each path point; and the comprehensive road risk of the first candidate path is determined according to the road risk of the multiple path points.
  • the road risk of the third waypoint in the plurality of waypoints is positively correlated with the distance difference between the third waypoint and the left and right borders of the road.
  • the determining module is specifically configured to: determine from the multiple candidate paths according to the path parameters of each candidate path and the comprehensive obstacle risk Target path.
  • the path parameter includes at least one of the following: curvature, length, path consistency, curvature change rate, heading angle error, and the change rate of the heading angle error
  • the path Consistency is used to indicate the degree of consistency between the candidate path at the current moment and the candidate path at the previous planning moment
  • the heading angle error is used to indicate the heading angle of the target object traveling along the candidate path and the heading angle of traveling along the road reference line The deviation between.
  • a third aspect of the embodiments of the present application provides a path selection device, which is characterized by comprising a processor and a memory, the processor and the memory are connected to each other, wherein the memory is used to store a computer program, and the computer
  • the program includes program instructions, and the processor is used to call the program instructions to execute the method described in the first aspect and any one of various possible implementation manners.
  • the fourth aspect of the embodiments of the present application provides a computer program product containing instructions, which is characterized in that, when it runs on a computer, the computer is caused to execute any of the above-mentioned first aspect and various possible implementation manners. The method described in the item.
  • the fifth aspect of the embodiments of the present application provides a computer-readable storage medium, including instructions, which are characterized in that, when the instructions run on a computer, the computer executes the above-mentioned first aspect and various possible implementation manners. Any of the methods.
  • a sixth aspect of the embodiments of the present application provides a chip including a processor.
  • the processor is used to read and execute the computer program stored in the memory to execute the method in any possible implementation manner of any one of the foregoing aspects.
  • the chip should include a memory, and the memory and the processor are connected to the memory through a circuit or a wire.
  • the chip further includes a communication interface, and the processor is connected to the communication interface.
  • the communication interface is used to receive data and/or information that needs to be processed, and the processor obtains the data and/or information from the communication interface, processes the data and/or information, and outputs the processing result through the communication interface.
  • the communication interface can be an input and output interface.
  • the path selection method provided by the embodiment of the application constructs the obstacle risk relationship based on the artificial potential field method, and can obtain the obstacle risk at any position in the road, according to the obstacle risk relationship and the position distribution of multiple candidate paths in the target road area
  • the obstacle risk of multiple candidate paths can be determined.
  • the optimal target path can be determined. When there are more obstacle data in the target road area, compared with existing collision detection methods, it can be Significantly reduce the path selection time.
  • the route selection method provided in the embodiments of the present application can also construct a road risk relationship based on basic road information, and determine the driving risk relationship by the obstacle risk relationship and the road risk relationship.
  • the magnitude of the risk potential value reflects the driving risk of the candidate route.
  • the optimal obstacle avoidance path is selected by calculating the comprehensive evaluation value of the "path parameter" and “driving risk” of the candidate path, which is different from the "collision detection” which judges whether the candidate path overlaps with obstacles one by one through inequality.
  • the method provided in the solution of this application is more concise and efficient.
  • this solution can be adapted to optimally select candidate paths when planning and avoiding obstacles on structured roads of arbitrary shapes.
  • Figure 1 is a schematic diagram of path selection
  • FIG. 2 is a schematic diagram of the architecture of the optimal obstacle avoidance path selection system in an embodiment of the application
  • FIG. 3 is a schematic diagram of an embodiment of a path selection method in an embodiment of the application.
  • Fig. 4 is a schematic diagram of the obstacle driving risk field in the embodiment of the application.
  • FIG. 5 is a schematic diagram of an embodiment of a path selection method in an embodiment of the application.
  • Figure 6a is a contour map of the driving risk field in an embodiment of the application.
  • Fig. 6b is a contour map of driving risk in an artificial potential field in an embodiment of the application.
  • FIG. 7a is a schematic diagram of determining a target obstacle avoidance path from multiple candidate paths in an embodiment of the application
  • Figure 7b is a schematic diagram of path comprehensive evaluation values of multiple candidate paths
  • FIG. 8 is a schematic diagram of comparison between a path selection method and a collision detection method according to an embodiment of the application.
  • FIG. 9 is a schematic diagram of an embodiment of a path selection device in an embodiment of the application.
  • Fig. 10 is a schematic diagram of an embodiment of a path selection device in an embodiment of the application.
  • the embodiment of the present application provides a route selection method for route selection in intelligent driving, which can avoid selecting candidate routes one by one, reduce the amount of calculation during route selection, and improve the efficiency of route decision-making.
  • the embodiments of the present application can be applied to the movement of various moving targets (including motor vehicles, non-motor vehicles, pedestrians or robots, etc.) on various forms of paths (including highways, urban roads, rural roads or indoor paths, etc.).
  • various moving targets including motor vehicles, non-motor vehicles, pedestrians or robots, etc.
  • paths including highways, urban roads, rural roads or indoor paths, etc.
  • the following embodiments are described by taking vehicles or lanes as examples, but those skilled in the art can extend it to the path planning field of other targets, and the specifics are not limited here.
  • the unmanned vehicle is driving on a structured road.
  • the actual application scenario may include multiple lanes.
  • no matter which lane the unmanned vehicle is in it can only be driven according to traffic rules.
  • Select adjacent lanes to change lanes that is, unmanned vehicles can only change lanes to the left or to the right. Therefore, the driving of unmanned vehicles in multiple lanes can be simplified to a two-lane model.
  • the perception module is compared to human eyes and ears in an autonomous driving system, then decision-making and planning are the brains of autonomous driving. After the brain receives various perception information from the sensors, it analyzes the current environment and then issues instructions to the underlying control module. This process is the main task of the decision-making and planning module.
  • a typical unmanned vehicle decision-making planning module can be divided into three levels:
  • Global path planning After receiving a given driving destination, combine the map information to generate a global path as a reference for subsequent specific path planning;
  • Behavior decision-making layer After receiving the global path, it combines the environmental information (including other vehicles and pedestrians, obstacles, and traffic rules on the road) obtained from the perception module to make specific behavior decisions (such as choice changes). To overtake or to follow);
  • the final path planning layer According to specific behavior decisions, plan to generate a trajectory that meets specific constraints (such as the vehicle's own dynamic constraints, collision avoidance, etc.). The trajectory is used as the input of the control module to determine the final travel path of the vehicle.
  • specific constraints such as the vehicle's own dynamic constraints, collision avoidance, etc.
  • the trajectory is used as the input of the control module to determine the final travel path of the vehicle.
  • the most commonly used path planning method is to first use a certain planning algorithm to generate multiple candidate paths on the road, and then select the best obstacle avoidance path according to the geometric properties of the candidate paths and the location of obstacles.
  • the time headway is an important indicator for evaluating driving safety.
  • the time headway represents the time difference between the front ends of two vehicles passing through the same place. Generally, it can be calculated by dividing the headway distance between the front and rear vehicles by the speed of the future vehicle.
  • the headway represents the maximum reaction time that the driver of the following vehicle has when the current vehicle is braking, so it does not fluctuate with the change of speed. It is generally expressed in ht and the unit is s.
  • FIG. 2 is a schematic diagram of the architecture of the optimal obstacle avoidance path selection system in an embodiment of this application.
  • the vehicle obtains road boundary and obstacle location information through environmental perception and information fusion, and obtains a cluster of candidate paths through the path planning module.
  • the system takes road boundary, obstacle location and candidate path as input, and the path selection system comprehensively evaluates each candidate
  • the quality of the path (determined by path parameters, such as path curvature, length and path consistency) and driving risk (including the driving risk caused by road boundaries and obstacles), select the optimal obstacle avoidance path as the output.
  • FIG. 3 is a schematic diagram of an embodiment of a path selection method in an embodiment of this application.
  • the path selection device constructs the road risk relationship according to the basic road information of the target road area.
  • the basic road information includes the position of the road, including the starting position of the road, the ending position of the road, the position of the road boundary and the position of the road reference line.
  • the road reference line may be, for example, The centerline of the road.
  • the starting position of the road and the ending position of the road can be preset to determine the target road area according to the basic information of the road.
  • the coordinate origin can be the current target position or the starting point of the target road centerline, etc. The specifics are not limited here, and the Cartesian coordinate system is used as an example to introduce.
  • the road risk relationship is constructed and determined according to the lateral distance between any point (x, y) on the road and the road boundary in the Cartesian coordinate system as a parameter.
  • the road risk relationship in the target road area in the embodiment of the present application is also referred to as a road risk field.
  • k l and k r are the intensity value of the risk field at the left boundary of the road and the intensity value of the risk field at the right boundary, respectively, representing the risk assessment value of the road boundary. The larger the value, the higher the risk when it is close to the road boundary. This value It can be determined by debugging through simulation experience, and the specific value is not limited here.
  • Y bi (s), i ⁇ l,r ⁇ respectively represent the left and right boundary values of the road, and
  • f cx and f cy are functions of the Cartesian coordinates of the road centerline with respect to the corresponding centerline length s. This means that if the location is closer to the left and right boundaries of the road, the corresponding risk value will become larger and larger, thereby achieving the purpose of restricting the host vehicle from driving within the road.
  • the subscript i can take the value l (left, left) or r (right, right), which respectively represent the left boundary Y bl and the right boundary Y br .
  • the comprehensive road risk of each candidate route can be calculated.
  • the path points on the candidate path are sampled at a certain interval, and the comprehensive road risk value of the entire candidate path is determined according to the road risk value at each path point.
  • multiple candidate paths in the target road area are determined based on the location of the target object on the target road area, for example, with the target object as the starting point; or, multiple candidate paths are set based on preset starting positions and destination positions.
  • the method for obtaining candidate paths is not specifically limited here.
  • step 301 to step 302 are optional steps, and step 301 and step 302 may be performed, or step 301 and step 302 may not be performed, and the details are not limited here.
  • the obstacle risk relationship in the target road area in the embodiment of the application is also called the obstacle risk field.
  • the obstacles are vehicles other than the target vehicle; if the target road area does not include the target vehicle, the obstacles are all vehicles in the target road area .
  • the obstacle driving risk field is also constructed based on the Gaussian function form, and its general form is:
  • N is the number of obstacles, and the number of obstacles is not limited in this application.
  • K obst is the intensity factor of the obstacle driving risk field. It should be designed according to the status of the main vehicle and the obstacle. Optionally, it is expressed as the headway time T between the vehicle and the obstacle i in the embodiment of this application.
  • k obst is the reference value of the benchmark strength, a coefficient that characterizes the strength of the obstacle risk, which can be determined based on experience, and the specific value is not limited here.
  • ⁇ is the exponential part of the adjustment parameter
  • D i is the pitch of the current host vehicle and the i-th obstacle
  • v host is the forward speed of the host vehicle, which means that greater headway
  • the obstacle is the risk intensity factor Smaller, the lower the driving risk will be.
  • A is the magnification factor of the obstacle's longitudinal driving risk field
  • B is the magnification factor of the obstacle's lateral driving risk field, which can be regarded as a position and the center of the obstacle (x obst, i , y obst, i )
  • the square of the "distance" after the "zoom" of the longitudinal and lateral distances is adjusted by adjusting the obstacle driving risk field magnification factor A or B.
  • the value of A is usually greater than B, for example, A is 250 to 1000 , B is 1 or 2, etc.
  • the specific values of A and B can be set by experience, and the specific values are not limited. , “Enlarge” or “shrink” the obstacle’s longitudinal or horizontal driving risk value.
  • the longitudinal influence range of the obstacle’s driving risk field can be enlarged, so that the main The vehicle achieves the purpose of maintaining a sufficient and safe longitudinal relative distance from the obstacle.
  • Ei is related to the current heading angle ⁇ obst,i of the i-th obstacle, so the distribution of the obstacle risk field value can also be adjusted adaptively with the attitude of the obstacle, and the heading angle of the vehicle can be adjusted according to sensors such as gyroscopes.
  • the purpose of the obstacle driving danger field is to prevent the host vehicle from colliding with other obstacle vehicles, so that the host vehicle maintains a certain safe distance from it. Therefore, when selecting the obstacle driving risk field function, on the one hand, when the vehicle approaches the obstacle vehicle, The corresponding risk value should also tend to infinity; on the other hand, the design of the risk field should make the risk value of the main vehicle and the obstacle in the vertical direction higher than the risk value in the horizontal direction under the same distance.
  • the comprehensive obstacle risk of the candidate path can be calculated.
  • the path points in the candidate path are sampled at a certain interval, and the comprehensive obstacle risk value of the entire candidate path is determined according to the obstacle risk value at each path point.
  • path reference points For each candidate path, a certain number of path reference points are selected. The number of path reference points is not limited here. In this embodiment, N path reference points are taken as an example for introduction, and each path is evaluated based on one or more path parameters.
  • path parameters include at least one of the following: curvature, length, path consistency, curvature change rate, heading angle error, heading angle error change rate, etc.
  • path consistency is used to indicate the degree of consistency between the candidate path at the current moment and the candidate path determined at the previous planning moment.
  • the heading angle error is used to indicate the deviation between the heading angle of the target object traveling along the candidate path and the heading angle traveling along the road reference line.
  • the heading angle error is a certain path on the candidate path when the target object is traveling along the candidate path. The deviation between the heading angle at a point and the heading angle at the point closest to the path point when driving along the road reference line.
  • curvature, length, path consistency, curvature change rate, heading angle error, and heading angle error change rate can be determined, or it can be determined from curvature, length, path consistency, and curvature change rate.
  • heading angle error and the heading angle error rate of change are more than one of the heading angle error and the heading angle error rate of change. The specifics are not limited here. It is understandable that the smaller the curvature of the path, the lower the degree of curvature, the higher the quality of the path; the smaller the length of the path, the higher the quality of the path; the higher the consistency of the path, the smaller the difference from the planned path at the previous moment.
  • this embodiment is introduced by taking path parameters including curvature, length, and consistency as examples.
  • mean curvature the mean curvature with a path metric Q ⁇ , mean curvature Q ⁇ is the absolute value of the reference point of the path:
  • ⁇ k is the path curvature at the path reference point.
  • the length of the average length of the path metric Q s, Q s is the average path length of the reference point of the average interval length:
  • ⁇ d k is the interval between adjacent sampling points.
  • the path consistency is measured by the average lateral deviation square Q l between the path sampling point and the planned path at the previous time:
  • L k is the lateral displacement of the sampling point from the center line of the road
  • L pre is the lateral displacement of the end point at the previous planning moment. This index ensures the continuity of the path planning before and after.
  • the average value of the sum of the above three path parameter indexes is used as the index Q to measure the quality of the path, namely:
  • the driving risk of each candidate route is determined according to the road risk and the obstacle risk, which is also called a driving risk field in this embodiment.
  • the driving risk field U includes the road boundary risk field U rd and the obstacle risk field U obst , and its size is the sum of the two:
  • the path parameters can be one or more of curvature, length, path consistency, curvature change rate, heading angle error, and heading angle error change rate. If based on multiple path parameters To determine the target path, you can set weights for different path parameters, calculate the quality index Q to measure the path according to the path parameters, and then select the path according to the quality index Q and driving risk.
  • path parameters including curvature, length and path consistency as an example to introduce.
  • the index J takes into account the path quality and driving risk, so the target path corresponding to the smallest J value is the optimal obstacle avoidance path at the current planning time.
  • the road centerline is a cubic polynomial curve
  • the curvature at the beginning of the centerline is 0
  • the basic intensity factor of the obstacle risk field k obst 100, if the main vehicle is located at the beginning of the center of the right lane and the speed is 20m/s, it is the head time distance of the stationary obstacle #1 Headway with obstacle vehicle #2
  • the expression of the obstacle risk field can be obtained according to (3):
  • the size of the driving risk field at the current position can be calculated according to (18).
  • the existing virtual artificial potential field is constructed to make obstacles produce "repulsive force” on the vehicle, and produce “attractive force” to the vehicle at the target point.
  • the vehicle avoids the obstacle under the action of "attractive force” and “repulsive force”.
  • This method has simple principle, small calculation amount and easy realization.
  • the artificial potential field value under the prior art usually cannot effectively distinguish between the longitudinal and lateral of the obstacle. For example, if the main vehicle maintains a distance of 1m from the obstacle in the longitudinal direction and a distance of 1m in the lateral direction, it is dangerous.
  • the degree is not equivalent, and the corresponding potential field values should also be clearly distinguished; in addition, the existing artificial potential field design usually cannot make appropriate adjustments with the movement of obstacles, posture changes, etc.; When the potential field method is used for path planning, the lack of global information easily makes the vehicle fall into a local minimum, which makes the planned path oscillate or even stagnate.
  • FIG. 6a and FIG. 6b are respectively the contour map of the driving risk field and the contour map of the driving risk of the artificial potential field in the embodiment of the application. It can be obtained by programming in the MATLAB environment.
  • Fig. 6a It can be seen from Fig. 6a that the solution proposed in this application can generate the risk value of the current driving environment in real time. When the main vehicle is closer to the lane boundary or close to an obstacle, the driving risk value will increase significantly.
  • Figure 6b shows the driving risk contour map constructed by the general artificial potential field, that is, when constructing the obstacle driving risk, there is no distinction between the obstacles in their horizontal and vertical hazards, so that it is only small in the vertical direction.
  • a high potential field value within a certain distance means that it will be affected only when it is very close to the obstacle.
  • it has a large potential field value in the horizontal lane, and it must be far away from the obstacle in the horizontal direction.
  • a lower potential field value can be obtained, both of which are not suitable for path decision-making.
  • Figure 6b also fails to reflect the difference in the degree of risk of the obstacle in different states.
  • the risk field value of the obstacle can be determined by the current relative motion state of the main vehicle. Due to the static obstacle and the front of the main vehicle The distance is greater than the movement obstacles far away, so the driving risk value of stationary obstacles is higher than that of movement obstacles far away; in addition, we can also see that compared with Figure 6b, the driving risk value in Figure 6a can be Make corresponding adjustments according to the heading of the vehicle.
  • FIG. 7a is a schematic diagram of determining a target obstacle avoidance path from multiple candidate paths in an embodiment of this application.
  • Multiple candidate paths can be generated using a cubic polynomial method.
  • According to (1-5) and (10) the driving risk value is calculated and normalized.
  • the normalized values are shown in Table 1.
  • the final evaluation value of the path is calculated according to (11), and the minimum value corresponds to the path
  • the serial number is the best obstacle avoidance path. Please refer to FIG. 7b for a schematic diagram of the path comprehensive evaluation values of multiple candidate paths.
  • the specific implementation effect of the candidate path is shown in Figure 7a.
  • the obstacle is about 70m away from the host vehicle, and the normal displacement from the centerline of the road is -2m.
  • the generated curve cluster is the candidate path, and the bold candidate path is
  • the determined target obstacle avoidance path is the optimal obstacle avoidance path.
  • the corresponding evaluation value of each path is shown in Figure 7b. We can observe that path 13 has the smallest comprehensive evaluation value of the path and can be determined as the target obstacle avoidance path .
  • FIG. 8 is a schematic diagram of comparison between the path selection method and the collision detection method according to an embodiment of the present application.
  • this embodiment selects the commonly used method of "collision detection” by inequality discrimination for comparison.
  • the average time consumption of the path selection method of this application is about 25ms.
  • the collision detection method takes more than twice the time, about 59ms; and the optimal path selected by the two methods is also different. As shown in Figure 8, the collision detection path is closer to the road boundary than the path selected by this method.
  • the location of the path selected by the method of this application is close to the centerline of the road, which is more ideal.
  • FIG. 9 is a schematic diagram of an embodiment of the path selection device in the embodiment of the application.
  • the path selection device includes:
  • the obtaining module 901 is configured to obtain basic road information of a target road area, location information of obstacles in the target road area, and multiple candidate paths of the target object in the target road area;
  • the determining module 902 is configured to determine the obstacle risk relationship of the target road area according to the road information and the location information of the obstacle, and the obstacle risk relationship is used to obtain any position in the target road area The obstacle risk;
  • the determining module 902 is further configured to determine the comprehensive obstacle risk of each candidate path according to the obstacle risk relationship and the position distribution of each candidate path in the target road area among the multiple candidate paths;
  • the determining module 902 is further configured to determine a target path from the multiple candidate paths according to the comprehensive obstacle risk of each candidate path.
  • the multiple candidate paths include a first candidate path; the determining module 902 is specifically configured to: determine multiple way points on the first candidate path; determine the multiple path points according to the obstacle risk relationship Obstacle risks of three path points; and determine the comprehensive obstacle risk of the first candidate path according to the obstacle risks of the multiple path points.
  • the multiple waypoints include a first waypoint; the obstacle risk of the first waypoint is based on the horizontal obstacle risk relationship of the first waypoint and the longitudinal obstacle of the first waypoint The risk relationship is determined; the horizontal obstacle risk relationship of the first waypoint is determined according to the position information of the obstacle and the obstacle's horizontal risk amplification coefficient; the longitudinal obstacle risk relationship of the first waypoint is determined according to the obstacle The location information and the obstacle longitudinal risk magnification factor are determined.
  • the acquisition module 901 is further configured to: acquire the heading angle of the obstacle, the heading angle of the obstacle is obtained according to the movement direction of the obstacle; the obstacle risk of the first waypoint is based on The heading angle of the obstacle, the horizontal obstacle risk relationship of the first waypoint and the longitudinal obstacle risk relationship of the first waypoint are determined.
  • the obstacle is a driving vehicle in the target road area.
  • the target object is a vehicle in the target road area
  • the obstacle is a vehicle other than the target object in the target road area.
  • the determining module 902 is further configured to determine the road risk relationship of the target road area according to the road information, and the road risk relationship is used to obtain the road risk of any position in the target road area.
  • the road information includes the road length, road width, and the position of the road reference line; the multiple candidate paths are determined according to the road risk relationship and the location distribution of each candidate path in the target road area.
  • the determining module 902 is specifically configured to: determine multiple waypoints on the first candidate path among the multiple candidate paths; determine the road risk of the multiple waypoints according to the road risk relationship; The road risks of the multiple path points determine the comprehensive road risks of the first candidate path.
  • the determining module 902 is further configured to determine the path quality of each candidate path;
  • the determining module 902 is specifically configured to determine a target path from the multiple candidate paths according to the path quality of each candidate path and the comprehensive obstacle risk.
  • the path quality is obtained based on at least one of the following factors: curvature, length, path consistency, curvature change rate, heading angle error, and heading angle error change rate, and the path consistency is used to indicate the current time The degree of consistency between the candidate path and the candidate path at the last planning time.
  • the software or firmware includes but is not limited to computer program instructions or codes, and can be executed by a hardware processor.
  • the hardware includes, but is not limited to, various integrated circuits, such as a central processing unit (CPU), a digital signal processor (DSP), a field programmable gate array (FPGA), or an application specific integrated circuit (ASIC).
  • CPU central processing unit
  • DSP digital signal processor
  • FPGA field programmable gate array
  • ASIC application specific integrated circuit
  • FIG. 10 is a schematic diagram of an embodiment of the path selection device in the embodiment of the present application.
  • the path selection apparatus 1000 includes a memory 1001 and a processor 1002.
  • the memory 1001 stores computer program instructions, and the processor 1002 runs the computer program instructions to perform the path selection related operations described in the foregoing embodiments.
  • the processor 1002 is also connected to one or more sensors outside the path selection device 1000, and receives raw data of the surrounding environment of the vehicle detected by the sensors.
  • the sensors include, but are not limited to, cameras, lidars, ultrasonic radars, or millimeter wave radars.
  • the target path output by the path selection device 1000 is generally sent to the underlying control module of the intelligent driving vehicle to provide reference information for controlling the vehicle.
  • the underlying control module may also be a software module executed by the processor 1002 or integrated in the processor 1002, which is not limited in this embodiment.
  • the processor 1002 includes, but is not limited to, various types of CPUs, DSPs, microcontrollers, microprocessors, or artificial intelligence processors.
  • the path selection device shown in Figure 9 and Figure 10 above constructs an obstacle risk field based on the artificial potential field method, and also constructs a road risk field based on road information.
  • the obstacle risk field and the road risk field jointly determine the driving risk field.
  • the value of the value reflects the level of the driving risk of the candidate route.
  • the optimal obstacle avoidance path is selected by calculating the comprehensive evaluation value of the "path quality index" and "driving risk index" of the candidate path, which is different from judging the candidate path one by one through inequality Whether there is a "collision detection" that overlaps with an obstacle, the method provided in the solution of this application is more concise and efficient.
  • the path selection time can be significantly reduced.

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Tourism & Hospitality (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Game Theory and Decision Science (AREA)
  • Development Economics (AREA)
  • Automation & Control Theory (AREA)
  • Mechanical Engineering (AREA)
  • Educational Administration (AREA)
  • Human Computer Interaction (AREA)
  • Transportation (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Traffic Control Systems (AREA)

Abstract

本申请提供了一种路径选择方法，用于从多条候选路径中确定最优避障路径，可以提供路径选择效率，该方法包括：获取目标道路区域的道路基本信息、所述目标道路区域内的障碍物的位置信息和目标对象在所述目标道路区域内的多条候选路径；根据所述道路基本信息和所述障碍物的位置信息确定所述目标道路区域的障碍物风险关系，所述障碍物风险关系用于获取所述目标道路区域中的任意一个位置的障碍物风险；根据所述障碍物风险关系和所述多条候选路径中每条候选路径在所述目标道路区域中的位置分布确定每条候选路径的综合障碍物风险；根据所述每条候选路径的综合障碍物风险，从所述多条候选路径中确定目标路径。

Description

路径选择方法和路径选择装置 技术领域
本申请涉及无人驾驶技术领域，尤其涉及一种路径选择方法和路径选择装置。
背景技术
无人驾驶是智能交通系统中的重要组成部分。无人驾驶车辆(下面简称无人车)接收到传感器的各种感知信息之后，对当前环境做出分析，然后对底层控制模块下达指令，这一过程就是决策规划模块的主要任务。典型的无人车决策规划包括：全局路径规划、行为决策和路径规划。其中，路径规划时，通常运用规划算法在路面上生成多条候选路径，然后根据候选路径的几何性质和障碍物的位置来选择最佳的避障路径，请参阅图1的路径选择的示意图。
现有的避障路径选择机制，通过构建不等式判断候选路径是否与障碍物存在交集，对候选路径逐条进行碰撞检测。如果路径与障碍物存在交集，那么舍弃该条路经。
由于需要逐条检查候选路径是否会与障碍物发生碰撞，当道路环境复杂，障碍物数据较多时，进行碰撞检测的不等式约束式复杂度高，路径选择效率低。
发明内容
本申请实施例提供了一种路径选择方法，用于从多条候选路径中确定最优避障路径，可以提供路径选择效率。
本申请实施例的第一方面提供一种路径选择方法，包括：获取目标道路区域的道路基本信息、所述目标道路区域内的障碍物的位置信息和目标对象在所述目标道路区域内的多条候选路径；根据所述道路基本信息和所述障碍物的位置信息确定所述目标道路区域的障碍物风险关系，所述障碍物风险关系用于获取所述目标道路区域中的任意一个位置的障碍物风险；根据所述障碍物风险关系和所述多条候选路径中每条候选路径在所述目标道路区域中的位置分布确定每条候选路径的综合障碍物风险；根据所述每条候选路径的综合障碍物风险，从所述多条候选路径中确定目标路径。所述多条候选路径基于目标对象在目标道路区域上的位置确定。
本申请实施例提供的路径选择方法，通过道路基本信息和障碍物的位置信息，可以构建目标道路区域中的障碍物风险关系，由此，可以确定多条候选路径的综合障碍物风险，从而从多条候选路径中直接获取障碍物风险较低的目标路径，可以提高路径选择效率。
在第一方面的一种可能的实现方式中，所述多条候选路径包括第一候选路径；所述根据所述障碍物风险关系和所述多条候选路径中每条候选路径在所述目标道路区域中的位置分布确定每条候选路径的综合障碍物风险包括：确定所述第一候选路径上的多个路径点；根据所述障碍物风险关系确定所述多个路径点的障碍物风险；根据所述多个路径点的障碍物风险确定所述第一候选路径的综合障碍物风险。
本申请实施例提供的路径选择方法，可以通过在候选路径上采样多个路径点，根据多个路径点的障碍物风险确定整条候选路径的综合障碍物风险，由此可以减少计算量，提高路径选择速度。
在第一方面的一种可能的实现方式中，所述第一路径点的障碍物风险根据所述第一路径点的横向障碍物风险关系和所述第一路径点的纵向障碍物风险关系确定；所述第一路径点的横向障碍物风险关系根据所述障碍物的位置信息和障碍物横向风险放大系数确定；所述第一路径点的纵向障碍物风险关系根据所述障碍物的位置信息和障碍物纵向风险放大系数确定。
可选的，所述横向风险放大系数和所述纵向风险放大系数为预设参数，用于分别表示横向障碍物与纵向障碍物对道路内各点的风险的影响。
可选地，第一路径点的横向障碍物风险与所述横向风险放大系数正相关，与所述第一路径点在垂直道路参考线的道路横向方向上与所述障碍物的距离负相关；根据所述道路基本信息和所述障碍物位置，按照预设的障碍物纵向风险放大系数确定所述第一路径点的纵向障碍物风险，所述第一路径点的纵向障碍物风险与所述纵向风险放大系数正相关，与所述第一路径点在沿道路参考线方向的道路纵向方向上与所述障碍物的距离负相关，所述障碍物横向风险放大系数小于所述障碍物纵向风险放大系数；根据所述第一路径点的横向障碍物风险和所述第一路径点的纵向障碍物风险确定所述第一路径点的障碍物风险。
本申请实施例提供的路径选择方法，在构建障碍物风险关系时，针对障碍物在道路横向和纵向方向上，对于目标造成的威胁程度不同，设置横向风险放大系数和纵向风险放大系数，其中，障碍物纵向风险放大系数大于障碍物横向风险放大系数，由此，确定的障碍物风险关系中每个位置对应的障碍物风险可以更贴近实际场景。
在第一方面的一种可能的实现方式中，所述方法还包括：获取所述障碍物的航向角，所述障碍物的航向角根据所述障碍物的运动方向获得；所述第一路径点的障碍物风险根据所述障碍物的航向角、以及所述第一路径点的横向障碍物风险关系和所述第一路径点的纵向障碍物风险关系确定。可选地，所述第一路径点的障碍物风险与第一角度大小负相关，所述第一角度为以所述障碍物为端点通过所述第一路径点的射线与地面坐标系横轴的角度与所述障碍物的航向角的差值。
本申请实施例提供的路径选择方法，在构建障碍物风险关系时，还可以考虑障碍物带当前的航向角，由于障碍物通常为移动物体，例如行驶中的车辆，参考其运动方向，考虑障碍物的动态风险，由此获取的障碍物风险关系可随着障碍物的姿态进行自适应调整。由此，生成的障碍物风险关系可以更贴近实际场景中的风险大小，障碍物风险预测效果更佳。
在第一方面的一种可能的实现方式中，所述障碍物为所述目标道路区域内的行驶车辆。
在第一方面的一种可能的实现方式中，所述目标对象为所述目标道路区域的车辆，所述障碍物为所述目标道路区域内除所述目标对象之外的车辆。
在第一方面的一种可能的实现方式中，根据所述道路基本信息确定所述目标道路区域的道路风险关系，所述道路风险关系用于获取所述目标道路区域中的任意一个位置的道路风险，所述道路基本信息包括道路长度、道路宽度和道路参考线的位置；根据所述道路风险关系和所述多条候选路径中每条候选路径在所述目标道路区域中的位置分布确定所述多条候选路径中每条候选路径的综合道路风险，所述综合道路风险用于从所述条候选路径中确定目标避障路径。
本申请实施例提供的路径选择方法，还考虑了道路边界的风险，由障碍物风险和道路风险共同确定行车风险，并用于选择目标路径，能更全面地预测风险。
在第一方面的一种可能的实现方式中，所述根据所述道路风险关系和所述多条候选路径中每条候选路径在所述目标道路区域中的位置分布确定所述多条候选路径中每条候选路径的综合道路风险包括：确定所述多条候选路径中第一候选路径上的多个路径点；根据所述道路风险关系确定所述多个路径点的道路风险；根据所述多个路径点的道路风险确定所述第一候选路径的综合道路风险。可选地，所述第三位置的道路风险与所述第三位置与道路左右边界的的距离差值正相关。
本申请实施例提供的路径选择方法，根据候选路径中采样的多个路径点确定每条候选路径的综合道路风险，道路中越靠近道路边界，则道路风险越高。
在第一方面的一种可能的实现方式中，所述方法还包括：所述根据所述每条候选路径的综合障碍物风险，从所述多条候选路径中确定目标路径包括：根据所述每条候选路径的所述路径参数和所述综合障碍物风险，从所述多条候选路径中确定目标路径。
本申请实施例提供的路径选择方法，除了考虑行车风险，还综合考虑路径参数，路径参数可以用于路径质量以确定目标避障路径。
在第一方面的一种可能的实现方式中，所述路径参数包括以下至少一个：曲率、长度、路径一致性、曲率变化率、航向角误差和所述航向角误差的变化率，所述路径一致性用于指示当前时刻的候选路径与上一规划时刻的所述候选路径的一致程度，所述航向角误差用于指示目标沿候选路径行驶的航向角与沿道路参考线行驶的航向角之间的偏差。
本申请实施例提供的路径选择方法，可以针对性设计不同组合的路径参数以确定目标路径，不同组合的路径参数可以衡量路径的质量，满足实际应用中对路径质量的多样化需求。
本申请实施例的第二方面提供了一种路径选择装置，包括：获取模块，用于获取目标道路区域的道路基本信息、所述目标道路区域内的障碍物的位置信息和目标对象在所述目标道路区域内的多条候选路径；确定模块，用于根据所述道路基本信息和所述障碍物的位置信息确定所述目标道路区域的障碍物风险关系，所述障碍物风险关系用于获取所述目标道路区域中的任意一个位置的障碍物风险；所述确定模块还用于，根据所述障碍物风险关系和所述多条候选路径中每条候选路径在所述目标道路区域中的位置分布确定每条候选路径的综合障碍物风险；所述确定模块还用于，根据所述每条候选路径的综合障碍物风险，从所述多条候选路径中确定目标路径。
在第二方面的一种可能的实现方式中，所述多条候选路径包括第一候选路径；所述确定模块具体用于：确定所述第一候选路径上的多个路径点；根据所述障碍物风险关系确定所述多个路径点的障碍物风险；根据所述多个路径点的障碍物风险确定所述第一候选路径的综合障碍物风险。
在第二方面的一种可能的实现方式中，所述多个路径点包括第一路径点；所述第一路径点的障碍物风险根据所述第一路径点的横向障碍物风险关系和所述第一路径点的纵向障碍物风险关系确定；所述第一路径点的横向障碍物风险关系根据所述障碍物的位置信息和障碍物横向风险放大系数确定；所述第一路径点的纵向障碍物风险关系根据所述障碍物的位置信息和障碍物纵向风险放大系数确定。
可选地，所述第一路径点的横向障碍物风险与所述横向风险放大系数正相关，与所述第一路径点在垂直道路参考线的道路横向方向上与所述障碍物的距离负相关，所述第一路径点 的纵向障碍物风险与所述纵向风险放大系数正相关，与所述第一路径点在沿道路参考线方向的道路纵向方向上与所述障碍物的距离负相关，所述障碍物横向风险放大系数小于所述障碍物纵向风险放大系数。
在第二方面的一种可能的实现方式中，所述获取模块还用于：获取所述障碍物的航向角，所述障碍物的航向角根据所述障碍物的运动方向获得；所述第一路径点的障碍物风险根据所述障碍物的航向角、以及所述第一路径点的横向障碍物风险关系和所述第一路径点的纵向障碍物风险关系确定。
所述第一路径点的障碍物风险与第一角度大小负相关，所述第一角度为以所述障碍物为端点通过所述第一路径点的射线与地面坐标系横轴的角度与所述障碍物的航向角的差值。
在第二方面的一种可能的实现方式中，所述障碍物为所述目标道路区域内的行驶车辆。
在第二方面的一种可能的实现方式中，所述目标对象为所述目标道路区域内的车辆，所述障碍物为所述目标道路区域内除所述目标对象之外的车辆。
在第二方面的一种可能的实现方式中，所述确定模块还用于：所述确定模块还用于：根据所述道路基本信息确定所述目标道路区域的道路风险关系，所述道路风险关系用于获取所述目标道路区域中的任意一个位置的道路风险，所述道路基本信息包括道路长度、道路宽度和道路参考线的位置；根据所述道路风险关系和所述多条候选路径中每条候选路径在所述目标道路区域中的位置分布确定所述多条候选路径中每条候选路径的综合道路风险，所述综合道路风险用于从所述条候选路径中确定目标避障路径。
在第二方面的一种可能的实现方式中，所述确定模块具体用于：确定所述多条候选路径中第一候选路径上的多个路径点；根据所述道路风险关系确定所述多个路径点的道路风险；根据所述多个路径点的道路风险确定所述第一候选路径的综合道路风险。
可选地，所述多个路径点中的第三路径点的道路风险与所述第三路径点与道路左右边界的的距离差值正相关。
在第二方面的一种可能的实现方式中，所述确定模块具体用于：根据所述每条候选路径的所述路径参数和所述综合障碍物风险，从所述多条候选路径中确定目标路径。
在第二方面的一种可能的实现方式中，所述路径参数包括以下至少一个：曲率、长度、路径一致性、曲率变化率、航向角误差和所述航向角误差的变化率，所述路径一致性用于指示当前时刻的候选路径与上一规划时刻的所述候选路径的一致程度，所述航向角误差用于指示目标物体沿候选路径行驶的航向角与沿道路参考线行驶的航向角之间的偏差。
本申请实施例第三方面提供了一种路径选择装置，其特征在于，包括处理器和存储器，所述处理器和所述存储器相互连接，其中，所述存储器用于存储计算机程序，所述计算机程序包括程序指令，所述处理器用于调用所述程序指令，执行如上述第一方面以及各种可能的实现方式中任一项所述的方法。
本申请实施例第四方面提供了一种包含指令的计算机程序产品，其特征在于，当其在计算机上运行时，使得所述计算机执行如上述第一方面以及各种可能的实现方式中任一项所述的方法。
本申请实施例第五方面提供了一种计算机可读存储介质，包括指令，其特征在于，当所述指令在计算机上运行时，使得计算机执行如上述第一方面以及各种可能的实现方式中任一 项所述的方法。
本申请实施例第六方面提供了一种一种芯片，包括处理器。处理器用于读取并执行存储器中存储的计算机程序，以执行上述任一方面任意可能的实现方式中的方法。可选地，该芯片该包括存储器，该存储器与该处理器通过电路或电线与存储器连接。进一步可选地，该芯片还包括通信接口，处理器与该通信接口连接。通信接口用于接收需要处理的数据和/或信息，处理器从该通信接口获取该数据和/或信息，并对该数据和/或信息进行处理，并通过该通信接口输出处理结果。该通信接口可以是输入输出接口。
从以上技术方案可以看出，本申请实施例具有以下优点：
本申请实施例提供的路径选择方法，基于人工势场法构建障碍物风险关系，可以获取道路中任意位置的障碍物风险，根据障碍物风险关系和多条候选路径在目标道路区域中的位置分布可以确定多条候选路径的障碍物风险，通过比较多条候选路径的障碍物风险，确定最优的目标路径，当目标道路区域的障碍物数据较多时，相较现有的碰撞检测方法，可以显著降低路径选择时间。
此外，本申请实施例提供的路径选择方法，还可以根据道路基本信息构建道路风险关系，由障碍物风险关系和道路风险关系共同确定行车风险关系，风险势值的大小体现了候选路径的行车风险的高低，然后，通过计算候选路径的“路径参数”与“行车风险”的综合评估值选择最优避障路径，不同于通过不等式逐条判断候选路径是否与障碍物存在交集的“碰撞检测”，本申请方案中提供的方法更为简洁高效。
此外，本方案可适应在任意形状的结构化道路上进行路径规划避障时，对候选路径进行最优地选择。
附图说明
图1为路径选择的示意图；
图2为本申请实施例中最优避障路径选择系统的架构示意图；
图3为本申请实施例中路径选择方法的实施例示意图；
图4为本申请实施例中障碍物行车风险场的示意图；
图5为本申请实施例中路径选择方法的实施例示意图；
图6a为本申请实施例中行车风险场等高图；
图6b为本申请实施例中人工势场行车风险等高图；
图7a为本申请实施例中从多条候选路径中确定目标避障路径的示意图；
图7b为多条候选路径的路径综合评估值的示意图；
图8为本申请实施例的路径选择方法与碰撞检测方法的对比示意图；
图9为本申请实施例中路径选择装置的一个实施例示意图；
图10为本申请实施例中路径选择装置的一个实施例示意图。
具体实施方式
本申请实施例提供了一种路径选择方法，用于智能驾驶中的路径选择，可以避免逐条选择候选路径，减少路径选择时的计算量，提升路径决策的效率。
本申请实施例可应用于各种运动目标(包括机动车、非机动车、行人或机器人等)在各 种形式的路径(包括高速公路、城市道路、乡村道路或室内路径等)上的运动，后续实施例以车辆或车道为例做描述，但本领域技术人员可以将其扩展至其他目标的路径规划领域，具体此处不做限定。
本申请实施例中，假设无人车在结构化道路上行驶，实际应用场景可能包括多车道，在多车道场景中，由于无论无人车处于哪条车道，根据交通规则行驶过程中，只能选择相邻车道变道，即无人车只能向左变道或者向右变道，由此，可以将无人车在多车道中的行驶简化为双车道模型，本申请实施例中，将以双车道模型为例进行介绍，可以理解的是，这并不会对方案实施的场景构成限定。
自动驾驶系统中如果将感知模块比作人的眼睛和耳朵，那么决策规划就是自动驾驶的大脑。大脑在接收到传感器的各种感知信息之后，对当前环境做出分析，然后对底层控制模块下达指令，这一过程就是决策规划模块的主要任务。
典型的无人车决策规划模块可以分为三个层次：
1、全局路径规划：在接收到一个给定的行驶目的地之后，结合地图信息，生成一条全局的路径，作为为后续具体路径规划的参考；
2、行为决策层：在接收到全局路径后，结合从感知模块得到的环境信息(包括其他车辆与行人，障碍物，以及道路上的交通规则信息)，做出具体的行为决策(例如选择变道超车还是跟随)；
3、最后路径规划层：根据具体的行为决策，规划生成一条满足特定约束条件(例如车辆本身的动力学约束、避免碰撞等)的轨迹，该轨迹作为控制模块的输入决定车辆最终行驶路径。目前一种运用较多的路径规划方法思路是：首先运用某种规划算法在路面上生成多条候选路径，然后根据候选路径的几何性质和障碍物的位置来选择最佳的避障路径。
为了便于理解，下面对本申请涉及的部分技术术语进行简要介绍：
车头时距(time headway，TH)是评价驾驶安全性的重要指标，车头时距代表着前后两辆车的前端通过同一地点的时间差，一般可使用前后车的车头间距除以后车速度来计算。车头时距代表当前车刹车时，后车驾驶员所具有的最大反应时间，因此它不随速度的变化而波动一般用ht表示，单位s。
下面结合附图，对本申请的实施例进行描述，显然，所描述的实施例仅仅是本申请一部分的实施例，而不是全部的实施例。本领域普通技术人员可知，随着技术的发展和新场景的出现，本申请实施例提供的技术方案对于类似的技术问题，同样适用。
本申请的说明书和权利要求书及上述附图中的术语“第一”、“第二”等是用于区别类似的对象，而不必用于描述特定的顺序或先后次序。应该理解这样使用的数据在适当情况下可以互换，以便这里描述的实施例能够以除了在这里图示或描述的内容以外的顺序实施。此外，术语“包括”和“具有”以及他们的任何变形，意图在于覆盖不排他的包含，例如，包含了一系列步骤或模块的过程、方法、系统、产品或设备不必限于清楚地列出的那些步骤或模块，而是可包括没有清楚地列出的或对于这些过程、方法、产品或设备固有的其它步骤或模块。在本申请中出现的对步骤进行的命名或者编号，并不意味着必须按照命名或者编号所指示的时间/逻辑先后顺序执行方法流程中的步骤，已经命名或者编号的流程步骤可以根据要实现的技术目的变更执行次序，只要能达到相同或者相类似的技术效果即可。
下面对本申请实施例提供的路径选择方法进行介绍。
请参阅图2，为本申请实施例中最优避障路径选择系统的架构示意图；
车辆经环境感知和信息融合得到道路边界、障碍物位置信息，并经路径规划模块得到一簇候选路径，该系统以道路边界、障碍物位置以及候选路径作为输入，路径选择系统综合评估每一条候选路径的质量(通过路径参数确定，例如路径曲率、长度和路径一致性)和行车风险(包括道路边界和障碍物引起的行车风险)，选择出最优的避障路径作为输出。
请参阅图3，为本申请实施例中路径选择方法的实施例示意图；
301、根据道路基本信息构建道路风险关系；
路径选择装置根据目标道路区域的道路基本信息构建道路风险关系，道路基本信息包括道路的位置，包括道路起始位置、道路终止位置、道路边界位置和道路参考线的位置，道路参考线例如可以是道路中心线。其中道路起始位置和道路终止位置可以预设，从而根据道路基本信息确定目标道路区域。在构建道路风险关系时可以基于不同的坐标系构建，坐标原点可以是当前的目标位置或者是目标道路中心线起点位置等，具体此处不做限定，下面以笛卡尔坐标系为例进行介绍。
根据笛卡尔坐标系中道路上任一点(x，y)与道路边界的横向距离为参数构建确定道路风险关系，本申请实施例中目标道路区域中的道路风险关系也称作道路风险场。
道路风险场的计算方法如下：
Figure PCTCN2020072850-appb-000001
其中k l、k r分别为道路左边界风险场的强度值和右边界风险场的强度值，代表道路边界所具有的风险评估值，值越大意味着靠近道路边界时风险越高，该数值可以由仿真经验调试确定，具体数值此处不做限定。Y bi(s),i∈{l,r}分别表示道路左右边界值，且
Figure PCTCN2020072850-appb-000002
其中D为道路宽度，f cx和f cy是道路中心线笛卡尔坐标关于其对应中心线长度s的函数。这意味着如果该位置越靠近道路左右边界，其对应的风险值也会越来越大，进而达到约束主车在道路范围内行驶的目的。下标i可以取值l(left，左)或者r(right，右)，分别表示左边界Y bl，和右边界Y br。
302、根据道路风险场确定候选路径的综合道路风险；
根据步骤301获取的道路风险场，可以计算每条候选路径的综合道路风险。可选的，以一定的间距采样候选路径上的路径点，根据各个路径点处的道路风险值确定整条候选路径的综合道路风险的值。
类似地，确定每条候选路径的综合道路风险的值。
可选地，目标道路区域中的多条候选路径基于目标对象在目标道路区域上的位置确定，例如以目标对象为起始点；或者，多条候选路径基于预设的起始位置和目的位置设定，对于候选路径的获取方式，此处具体不做限定。
需要说明的是，步骤301至步骤302为可选步骤，可以执行步骤301和步骤302，也可以不执行步骤301和步骤302，具体此处不做限定。
303、根据障碍物位置构建障碍物风险关系；
本申请实施例中目标道路区域中的障碍物风险关系也称作障碍物风险场，请参阅图4，为本申请实施例中障碍物行车风险场的示意图；
根据障碍物位置确定道路中的障碍物风险场。可以理解的是，若目标道路区域内包括目标车辆，则障碍物即为除目标车辆之外的其他车辆；若目标道路区域内不包括目标车辆，则障碍物为该目标道路区域内的所有车辆。
具体的，障碍物行车风险场同样基于高斯函数形式构建，其一般形式为：
Figure PCTCN2020072850-appb-000003
其中，N是障碍物个数，本申请中对于障碍物的数量不做限定。
K obst是障碍物行车风险场的强度因子，它应根据主车与障碍物的状态进行设计，可选的，本申请实施例中将其表示为关于本车与障碍物i的车头时距T h,i的非线性函数：
Figure PCTCN2020072850-appb-000004
上式中k obst是基准强度参考值，表征障碍物风险强度的一个系数，可以根据经验确定，具体数值此处不做限定。μ是指数部分的调节参数，d i是当前主车与第i个障碍物的间距，v host是主车的前进速度，这就意味着车头时距越大，该障碍物风险场强度因子越小，引起的行车风险也会越低。
E i的具体表达式为：
Figure PCTCN2020072850-appb-000005
其中，A为障碍物纵向行车风险场放大系数，B为障碍物横向行车风险场放大系数，可以看作是在障碍物坐标系下，某位置与障碍物中心(x obst,i，y obst,i)的纵向、横向距离经“缩放”后的“距离”平方量，通过调节障碍物行车风险场放大系数A或B，需要说明的是，通常A的数值大于B，例如A为250至1000，B为1或2等，A和B的具体数值可以由经验设定，具体数值不做限定。，“放大”或“缩小”障碍物在其纵向或横向上的行车风险值，例如通过增大系数A(保持B不变)可以扩大障碍物行车风险场在纵向上的影响范围，从而使主车达到与障碍保持足够的安全的纵向相对距离的目的。此外，Ei与第i个障碍物当前的航向角θ obst,i有关，因此障碍物风险场数值的分布也能随着障碍物的姿态进行自适应调整，车辆的航向角可以根据陀螺仪等传感器获取，具体获取方式此处不做限定。
障碍物行车危险场的目的是为了防止主车与其他障碍物车辆发生碰撞，使主车与其保持一定的安全距离，因此在选取障碍物行车风险场函数时，一方面当车辆越接近障碍车辆，对应的风险值也应趋向于无穷大；另一方面，风险场的设计应使得在同样的间距情况下，主车与障碍物在纵向上的风险值要高于其在横向上的风险值。
304、根据障碍物风险场确定候选路径的综合障碍物风险；
根据步骤303获取的障碍物风险场，可以计算该候选路径的综合障碍物风险。可选的，以一定的间距采样候选路径中的路径点，根据各个路径点处的障碍物风险值确定整条候选路 径的综合障碍物风险的值。
类似地，确定每条候选路径的综合障碍物风险的值。
305、确定候选路径的路径参数；
针对每条候选路径，取一定数目的路径参考点，路径参考点的数量此处不做限定，本实施例中以取N个路径参考点为例进行介绍，根据一个或多个路径参数评估每条候选路径本身的质量，路径参数包括以下至少一个：曲率、长度、路径一致性、曲率变化率、航向角误差和航向角误差变化率等。其中，路径一致性用于指示在当前时刻候选路径与上一规划时刻确定的候选路径之间的一致程度。航向角误差用于指示目标物体沿候选路径行驶的航向角与沿道路参考线行驶的航向角之间的偏差，具体的，航向角误差是目标物体沿候选路径行驶时在候选路径上某一路径点处的航向角，与沿道路参考线行驶时在距离该路径点最近的点处的航向角之间的偏差。
可选地，本步骤中可以仅确定曲率、长度、路径一致性、曲率变化率、航向角误差和航向角误差变化率中的任意一个，也可以从曲率、长度、路径一致性、曲率变化率、航向角误差和航向角误差变化率中确定多个，具体此处不做限定。可以理解的是，路径的曲率越小，弯曲程度越低，路径的质量越高；路径长度越小，路径的质量越高；路径一致性越高，与上一时刻规划路径的差异越小，路径规划的稳定性好则路径的质量越高；路径的曲率变化率越小，路径的质量越高；路径的航向角误差越小，与道路参考线的走势越接近，路径的质量越高，航向角误差变化率越小，路径的质量越高。
示例性的，本实施例以路径参数包括曲率、长度和一致性为例进行介绍。
可选的，曲率用路径的平均曲率Q κ度量，平均曲率Q κ为路径参考点处的平均曲率绝对值：
Figure PCTCN2020072850-appb-000006
其中κ k为路径参考点处的路径曲率。
可选的，长度用路径的平均长度Q s度量，平均长度Q s为 路径参考点的平均间隔长度：
Figure PCTCN2020072850-appb-000007
其中Δd k为相邻采样点的间隔。
可选的，路径一致性用路径采样点与前一时刻规划路径的平均横向偏差平方Q l度量：
Figure PCTCN2020072850-appb-000008
其中L k为采样点的与道路中心线的横向位移，L pre为前一规划时刻的终点的横向位移，该指标保证前后路径规划的连续性。
由此，将上述三个路径参数指标之和的均值作为衡量路径的质量的指标Q，即：
Figure PCTCN2020072850-appb-000009
其中
Figure PCTCN2020072850-appb-000010
分别是上述每个指标的归一化值。
需要说明的是，由于离障碍物比较近的路径通常会有更大的风险，质量最好的路径并不一定是最安全的。
306、根据候选路径的综合道路风险、综合障碍物风险和路径参数确定目标路径；
根据道路风险和障碍物风险确定每条候选路径的行车风险，本实施例中也称为行车风险场。
行车风险场U包括道路边界风险场U rd和障碍物风险场U obst，它的大小为两者之和：
U＝U rd+U obst             (10)
需要说明的是，若不执行步骤301和步骤302，则U＝U obst。
根据行程风险和路径参数确定目标路径，其中，路径参数可以是曲率、长度、路径一致性、曲率变化率、航向角误差和航向角误差变化率中的一个或多个，若根据多个路径参数确定目标路径，可以为不同的路径参数设定权值，根据路径参数计算得到衡量路径的质量指标Q，进而根据质量指标Q和行车风险选择路径。下面以路径参数包括曲率、长度和路径一致性为例进行介绍。
将所有候选路径的行车风险场U归一化后得到
Figure PCTCN2020072850-appb-000011
再根据步骤303获取的路径的质量指标Q，将每条路径的路径质量指标与行车风险指标的乘积作为最终的路径选择指标J，候选路径k的路径选择指标为J k：
Figure PCTCN2020072850-appb-000012
由(11)可知，指标J兼顾了路径质量和行车风险，因此J值最小对应的目标路径即为当前规划时刻下最优的避障路径。
示例性的，下面介绍本申请实施例中路径选择方法的一个具体实施例：
假设道路中心线为三次多项式曲线，道路长度约L＝250m，中心线开始处的曲率为0，终止曲率为C f＝1/800，道路曲率变化率C＝(C f-C 0)/L＝1/200000，道路中心线侧向位移方程设为
Figure PCTCN2020072850-appb-000013
在中心线上以间距为1采样笛卡尔横向X坐标，即x＝0:1:250，共计n＝251个采样点，根据式(12)计算得到笛卡尔纵向Y坐标的值y，通过勾股定理近似计算出相邻采样点之间的距离，并累计得到路径向量s(初始值s[0]＝0)
s[i]＝s[i-1]+sqrt((x[i]-x[i-1]) 2+(y[i]-y[i-1]) 2),i＝1,…,n    (13)
构造道路中心线笛卡尔坐标关于路径向量s的函数：在道路中心线起始处建立笛卡尔坐标系，可选的，采用三次样条曲线插值方法获得笛卡尔横向X坐标与路径向量s的函数关系式f cx，笛卡尔纵向Y坐标与路径向量s的函数关系式f cy，如(14)所示
Figure PCTCN2020072850-appb-000014
构建车道边界风险场：以单向双车道为例，取道路宽度D＝7.5m，车道宽度即为3.75m，依据(2)和(14)获得道路左右边界的表达式Y bl,Y br，即
Figure PCTCN2020072850-appb-000015
取道路边界风险场强度值k l＝100,k r＝100，依据(1)，道路边界风险场表达式为：
Figure PCTCN2020072850-appb-000016
其中s为任意道路中心线长度。
构建障碍物风险场：假定该车道上存在两个障碍物，其中左车道正中有一静止障碍物#1，距离起始位置约50m，在笛卡尔坐标系中的位置为(x obst,1，y obst,1)＝(50,1.90)，在此刻的航向角为θ obst,1＝0；右车道有一运动障碍物#2，位置为距离道路开始处80m，在笛卡尔坐标系中的位置为(x obst,2，y obst,2)＝(100,0.14)，航向角θ obst,2＝0.08rad，速度为15m/s，且运动方向与中心线平行。障碍物风险场基本强度因子k obst＝100，假若主车位于右车道中心起始处，速度为20m/s，它与静止障碍#1的车头时距
Figure PCTCN2020072850-appb-000017
与障碍车辆#2的车头时距
Figure PCTCN2020072850-appb-000018
依据式(4)，取调节参数μ＝10，这样静止障碍物#1风险场的强度因子K obst,1＝100×e -2.5/10＝77.88，移动障碍物#2的风险场的强度因子K obst,2＝100×e -5/10＝60.65。取行车风险场的放大系数A＝250，B＝1，根据(3)可以得到障碍物风险场的表达式：
Figure PCTCN2020072850-appb-000019
其中，
Figure PCTCN2020072850-appb-000020
Figure PCTCN2020072850-appb-000021
结合道路边界风险场和障碍物风险场，依据式(10)最终得到当前障碍物行车风险场的表达式为：
Figure PCTCN2020072850-appb-000022
这样如果知道任意笛卡尔坐标的位置(x,y)，根据(18)即可计算出当前位置的行车风险场的大小。
现有的通过构建虚拟人工势场，使得障碍物对车辆产生“排斥力”，在目标点对车辆产生“吸引力”，车辆在“吸引力”和“排斥力”的作用下避开障碍并向目标点运动，该方法原理简单、计算量小、易实现。现有技术下的人工势场值通常不能在障碍物的纵向和横向做出有效区分，举例来说，主车在纵向上保持与障碍物1m的距离与在横向上保持1m的距离，其危险程度是不可等同而言的，其对应的势场值也应该有着明显的区分；此外现有的人工势场设计通常不能随着障碍物运动、姿态变化等做出适当的调整；而在单纯采用势场法进行路径规划时，由于缺乏全局信息而容易使得车辆陷入局部极小值，使得规划的路径产生振荡甚至停滞。
请参阅图6a和图6b，分别为本申请实施例中行车风险场等高图和人工势场行车风险等高图。可以通过MATLAB环境下进行编程获取。
从图6a可以看出，本申请提出的方案可以实时生成当前行车环境的风险值，当主车越靠近车道边界或者接近障碍物，行车风险值将会显著增加。
而图6b展示的是一般人工势场构建的行车风险等高线图，即在构建障碍物行车风险时，没有区分障碍物在其横向和纵向上的危险程度，这样它在纵向上只有很小一段距离内有较高的势场值，意味着只有在非常靠近障碍的时候才会受到影响，反观它在横向车道范围内有很大的势场值，在横向上要距障碍物较远才可以得到较低的势场值，这两点都不适合于进行路径决策。
此外，图6b中也没能体现出障碍物在不同状态下其风险程度的差异，障碍物的风险场值能随当前与主车的相对运动状态来决定，由于静止障碍与主车的车头时距要大于离得较远的运动障碍，故静止障碍的行车风险值要高于离得较远的运动障碍；此外我们还能看出与图图6b相比，图6a中的行车风险值能根据车辆的航向做出对应的调整。
请参阅图7a，为本申请实施例中从多条候选路径中确定目标避障路径的示意图；
多条候选路径可以采用三次多项式方法生成。以距离间距为1生成路径采样点，即S p＝0:1:70。根据式(6-8)分别计算每条路径的平均曲率，平均长度以及路径一致性指标，将这三个指标归一化后再依据(9)计算路径质量指标。依据(1-5)和(10)计算得到行车风险值并进行归一化，各归一化值如表1所示，最终依据(11)计算路径最终的评估值，其最小值对应的路径序号即为最佳的避障路径。请参阅图7b为多条候选路径的路径综合评估值的示意图。
表1 候选路径评估指标
Figure PCTCN2020072850-appb-000023
候选路径的具体实现效果如图7a所示，规划初始，障碍物距离主车约70m，与道路中心线的法向位移为-2m，生成的曲线簇即为候选路径，加粗的候选路径为确定的目标避障路径，即最优避障路径，各路径对应的评估值如图7b所示，我们可以观察到，路径13具有最小的路经综合评估值，可以被确定为目标避障路径。
请参阅图8，为本申请实施例的路径选择方法与碰撞检测方法的对比示意图；
为了区别于现有技术的技术效果，本实施例选取了目前常用的采取不等式判别来进行“碰撞检测”的方法进行对比，经仿真验证，本申请的路径选择方法的平均耗时约为25ms，而碰撞检测方法所用的时间多出一倍有余，约59ms；并且两种方法的选择的最优路径也有差异，如图8所示，碰撞检测的路径比本方法选择的路径更接近道路边界，而本申请方法选择的路径位置接近道路中心线，更为理想。
上面介绍了本申请提供的路径选择方法，下面对实现该路径选择方法的路径选择装置进行介绍，请参阅图9，为本申请实施例中路径选择装置的一个实施例示意图。
该路径选择装置，包括：
获取模块901，用于获取目标道路区域的道路基本信息、所述目标道路区域内的障碍物的位置信息和目标对象在所述目标道路区域内的多条候选路径；
确定模块902，用于根据所述道路信息和所述障碍物的位置信息确定所述目标道路区域的障碍物风险关系，所述障碍物风险关系用于获取所述目标道路区域中的任意一个位置的障碍物风险；
所述确定模块902还用于，根据所述障碍物风险关系和所述多条候选路径中每条候选路径在所述目标道路区域中的位置分布确定每条候选路径的综合障碍物风险；
所述确定模块902还用于，根据所述每条候选路径的综合障碍物风险，从所述多条候选路径中确定目标路径。
可选地，所述多条候选路径包括第一候选路径；所述确定模块902具体用于：确定所述第一候选路径上的多个路径点；根据所述障碍物风险关系确定所述多个路径点的障碍物风险；根据所述多个路径点的障碍物风险确定所述第一候选路径的综合障碍物风险。
可选地，所述多个路径点包括第一路径点；所述第一路径点的障碍物风险根据所述第一路径点的横向障碍物风险关系和所述第一路径点的纵向障碍物风险关系确定；所述第一路径点的横向障碍物风险关系根据所述障碍物的位置信息和障碍物横向风险放大系数确定；所述第一路径点的纵向障碍物风险关系根据所述障碍物的位置信息和障碍物纵向风险放大系数确定。
可选地，所述获取模块901还用于：获取所述障碍物的航向角，所述障碍物的航向角根据所述障碍物的运动方向获得；所述第一路径点的障碍物风险根据所述障碍物的航向角、以及所述第一路径点的横向障碍物风险关系和所述第一路径点的纵向障碍物风险关系确定。
可选地，所述障碍物为所述目标道路区域内的行驶车辆。
可选地，所述目标对象为所述目标道路区域内的车辆，所述障碍物为所述目标道路区域内除所述目标对象之外的车辆。
可选地，所述确定模块902还用于：根据所述道路信息确定所述目标道路区域的道路风险关系，所述道路风险关系用于获取所述目标道路区域中的任意一个位置的道路风险，所述道路信息包括道路长度、道路宽度和道路参考线的位置；根据所述道路风险关系和所述多条候选路径中每条候选路径在所述目标道路区域中的位置分布确定所述多条候选路径中每条候选路径的综合道路风险，所述综合道路风险用于从所述条候选路径中确定目标避障路径。
可选地，所述确定模块902具体用于：确定所述多条候选路径中第一候选路径上的多个路径点；根据所述道路风险关系确定所述多个路径点的道路风险；根据所述多个路径点的道路风险确定所述第一候选路径的综合道路风险。
可选地，所述确定模块902还用于：确定所述每条候选路径的路径质量；
所述确定模块902具体用于：根据所述每条候选路径的所述路径质量和所述综合障碍物风险，从所述多条候选路径中确定目标路径。
可选地，所述路径质量基于以下因素中的至少一个获取：曲率、长度、路径一致性、曲 率变化率、航向角误差和航向角误差变化率，所述路径一致性用于指示当前时刻的候选路径与上一规划时刻的所述候选路径的一致程度。
图9中的各个模块的只一个或多个可以软件、硬件、固件或其结合实现。所述软件或固件包括但不限于计算机程序指令或代码，并可以被硬件处理器所执行。所述硬件包括但不限于各类集成电路，如中央处理单元(CPU)、数字信号处理器(DSP)、现场可编程门阵列(FPGA)或专用集成电路(ASIC)。
请参阅图10，本申请实施例中路径选择装置的一个实施例示意图。
如图10所示，路径选择装置1000包括存储器1001和处理器1002。所述存储器1001存储计算机程序指令，所述处理器1002运行所述计算机程序指令以执行上述实施例描述的路径选择相关操作。所述处理器1002还与路径选择装置1000外界的一个或多个传感器相连接，接收所述传感器探测的自车周围环境的原始数据。所述传感器包括但不限于如摄像头、激光雷达、超声波雷达或毫米波雷达。路径选择装置1000输出的目标路径一般发送给智能驾驶车辆的底层控制模块，提供控车参考信息。底层控制模块也可以是由处理器1002执行的一个软件模块或集成于处理器1002中，本实施例不做限定。处理器1002包括但不限于各类CPU、DSP、微控制器、微处理器或人工智能处理器。
上述图9、图10所示的路径选择装置，基于人工势场法构建障碍物风险场，还根据道路信息构建道路风险场，由障碍物风险场和道路风险场共同确定行车风险场，风险势值的大小体现了候选路径的行车风险的高低，然后，通过计算候选路径的“路径质量指标”与“行车风险指标”的综合评估值选择最优避障路径，不同于通过不等式逐条判断候选路径是否与障碍物存在交集的“碰撞检测”，本申请方案中提供的方法更为简洁高效。当目标道路区域的障碍物数据较多时，相较现有的碰撞检测方法，可以显著降低路径选择时间。

Claims (21)

  1. 一种路径选择方法，其特征在于，包括：
    获取目标道路区域的道路基本信息、所述目标道路区域内的障碍物的位置信息和目标对象的在所述目标道路区域内的多条候选路径；
    根据所述道路基本信息和所述障碍物的位置信息确定所述目标道路区域的障碍物风险关系，所述障碍物风险关系用于获取所述目标道路区域中的任意一个位置的障碍物风险；
    根据所述障碍物风险关系和所述多条候选路径中每条候选路径在所述目标道路区域中的位置分布确定每条候选路径的综合障碍物风险；
    根据所述每条候选路径的综合障碍物风险，从所述多条候选路径中确定目标路径。
  2. 根据权利要求1所述的方法，其特征在于，所述多条候选路径包括第一候选路径；
    所述根据所述障碍物风险关系和所述多条候选路径中每条候选路径在所述目标道路区域中的位置分布确定每条候选路径的综合障碍物风险包括：
    确定所述第一候选路径上的多个路径点；
    根据所述障碍物风险关系确定所述多个路径点的障碍物风险；
    根据所述多个路径点的障碍物风险确定所述第一候选路径的综合障碍物风险。
  3. 根据权利要求2所述的方法，其特征在于，所述多个路径点包括第一路径点；
    所述第一路径点的障碍物风险根据所述第一路径点的横向障碍物风险关系和所述第一路径点的纵向障碍物风险关系确定；
    所述第一路径点的横向障碍物风险关系根据所述障碍物的位置信息和障碍物横向风险放大系数确定；
    所述第一路径点的纵向障碍物风险关系根据所述障碍物的位置信息和障碍物纵向风险放大系数确定。
  4. 根据权利要求3所述的方法，其特征在于，所述方法还包括：
    获取所述障碍物的航向角，所述障碍物的航向角根据所述障碍物的运动方向获得；
    所述第一路径点的障碍物风险根据所述障碍物的航向角、以及所述第一路径点的横向障碍物风险关系和所述第一路径点的纵向障碍物风险关系确定。
  5. 根据权利要求1至4中任一项所述的方法，其特征在于，所述障碍物为所述目标道路区域内的行驶车辆。
  6. 根据权利要求1至5中任一项所述的方法，其特征在于，所述方法还包括：
    根据所述道路基本信息确定所述目标道路区域的道路风险关系，所述道路风险关系用于获取所述目标道路区域中的任意一个位置点的道路风险，所述道路基本信息包括道路长度、道路宽度和道路参考线的位置；
    根据所述道路风险关系和所述多条候选路径中每条候选路径在所述目标道路区域中的位置分布确定所述多条候选路径中每条候选路径的综合道路风险，所述综合道路风险用于从所述条候选路径中确定目标避障路径。
  7. 根据权利要求6所述的方法，其特征在于，
    所述根据所述道路风险关系和所述多条候选路径中每条候选路径在所述目标道路区域中的位置分布确定所述多条候选路径中每条候选路径的综合道路风险包括：
    确定所述多条候选路径中第一候选路径上的多个路径点；
    根据所述道路风险关系确定所述多个路径点的道路风险；
    根据所述多个路径点的道路风险确定所述第一候选路径的综合道路风险。
  8. 根据权利要求1至7中任一项所述的方法，其特征在于，
    所述根据所述每条候选路径的综合障碍物风险，从所述多条候选路径中确定目标路径包括：
    根据所述每条候选路径的所述路径参数和所述综合障碍物风险，从所述多条候选路径中确定目标路径。
  9. 根据权利要求8所述的方法，其特征在于，
    所述路径参数包括以下至少一个：曲率、长度、路径一致性、曲率变化率、航向角误差和所述航向角误差的变化率，所述路径一致性用于指示当前时刻的候选路径与上一规划时刻的所述候选路径的一致程度，所述航向角误差用于指示目标沿候选路径行驶的航向角与沿道路参考线行驶的航向角之间的偏差。
  10. 一种路径选择装置，其特征在于，包括：
    获取模块，用于获取目标道路区域的道路基本信息、所述目标道路区域内的障碍物的位置信息和目标对象在所述目标道路区域内的多条候选路径；
    确定模块，用于根据所述道路基本信息和所述障碍物的位置信息确定所述目标道路区域的障碍物风险关系，所述障碍物风险关系用于获取所述目标道路区域中的任意一个位置的障碍物风险；
    所述确定模块还用于，根据所述障碍物风险关系和所述多条候选路径中每条候选路径在所述目标道路区域中的位置分布确定每条候选路径的综合障碍物风险；
    所述确定模块还用于，根据所述每条候选路径的综合障碍物风险，从所述多条候选路径中确定目标路径。
  11. 根据权利要求10所述的装置，其特征在于，所述多条候选路径包括第一候选路径；
    所述确定模块具体用于：
    确定所述第一候选路径上的多个路径点；
    根据所述障碍物风险关系确定所述多个路径点的障碍物风险；
    根据所述多个路径点的障碍物风险确定所述第一候选路径的综合障碍物风险。
  12. 根据权利要求11所述的装置，其特征在于，所述多个路径点包括第一路径点；
    所述第一路径点的障碍物风险根据所述第一路径点的横向障碍物风险关系和所述第一路径点的纵向障碍物风险关系确定；
    所述第一路径点的横向障碍物风险关系根据所述障碍物的位置信息和障碍物横向风险放大系数确定；
    所述第一路径点的纵向障碍物风险关系根据所述障碍物的位置信息和障碍物纵向风险放大系数确定。
  13. 根据权利要求11或12中任一项所述的装置，其特征在于，
    所述获取模块还用于：获取所述障碍物的航向角，所述障碍物的航向角根据所述障碍物的运动方向获得；
    所述第一路径点的障碍物风险根据所述障碍物的航向角、以及所述第一路径点的横向障碍物风险关系和所述第一路径点的纵向障碍物风险关系确定。
  14. 根据权利要求10至13中任一项所述的装置，其特征在于，所述障碍物为所述目标道路区域内的行驶车辆。
  15. 根据权利要求10至14中任一项所述的装置，其特征在于，所述确定模块还用于：
    根据所述道路基本信息确定所述目标道路区域的道路风险关系，所述道路风险关系用于获取所述目标道路区域中的任意一个位置的道路风险，所述道路基本信息包括道路长度、道路宽度和道路参考线的位置；
    根据所述道路风险关系和所述多条候选路径中每条候选路径在所述目标道路区域中的位置分布确定所述多条候选路径中每条候选路径的综合道路风险，所述综合道路风险用于从所述条候选路径中确定目标避障路径。
  16. 根据权利要求15所述的装置，其特征在于，所述确定模块具体用于：
    确定所述多条候选路径中第一候选路径上的多个路径点；
    根据所述道路风险关系确定所述多个路径点的道路风险；
    根据所述多个路径点的道路风险确定所述第一候选路径的综合道路风险。
  17. 根据权利要求10至16中任一项所述的装置，其特征在于，所述确定模块具体用于：
    根据所述每条候选路径的所述路径参数和所述综合障碍物风险，从所述多条候选路径中确定目标路径。
  18. 根据权利要求17所述的装置，其特征在于，
    所述路径参数包括以下至少一个：曲率、长度、路径一致性、曲率变化率、航向角误差和所述航向角误差的变化率，所述路径一致性用于指示当前时刻的候选路径与上一规划时刻的所述候选路径的一致程度，所述航向角误差用于指示目标物体沿候选路径行驶的航向角与沿道路参考线行驶的航向角之间的偏差。
  19. 一种路径选择装置，其特征在于，包括处理器和存储器，所述处理器和所述存储器相互连接，其中，所述存储器用于存储计算机程序，所述计算机程序包括程序指令，所述处理器用于调用所述程序指令，执行如权利要求1至9中任一项所述的方法。
  20. 一种包含指令的计算机程序产品，其特征在于，当其在计算机上运行时，使得所述计算机执行如权利要求1至9中任一项所述的方法。
  21. 一种计算机可读存储介质，包括指令，其特征在于，当所述指令在计算机上运行时，使得计算机执行如权利要求1至9中任一项所述的方法。
PCT/CN2020/072850 2020-01-17 2020-01-17 路径选择方法和路径选择装置 WO2021142799A1 (zh)

Priority Applications (3)

Application Number Priority Date Filing Date Title
EP20914201.7A EP4089369A4 (en) 2020-01-17 2020-01-17 PATH SELECTION METHOD AND PATH SELECTION DEVICE
CN202080004767.5A CN112639849A (zh) 2020-01-17 2020-01-17 路径选择方法和路径选择装置
PCT/CN2020/072850 WO2021142799A1 (zh) 2020-01-17 2020-01-17 路径选择方法和路径选择装置

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2020/072850 WO2021142799A1 (zh) 2020-01-17 2020-01-17 路径选择方法和路径选择装置

Publications (1)

Publication Number Publication Date
WO2021142799A1 true WO2021142799A1 (zh) 2021-07-22

Family

ID=75291269

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/072850 WO2021142799A1 (zh) 2020-01-17 2020-01-17 路径选择方法和路径选择装置

Country Status (3)

Country Link
EP (1) EP4089369A4 (zh)
CN (1) CN112639849A (zh)
WO (1) WO2021142799A1 (zh)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114030483A (zh) * 2021-12-16 2022-02-11 阿波罗智联(北京)科技有限公司 车辆控制方法、装置、电子设备和介质
CN114077249A (zh) * 2021-10-22 2022-02-22 陕西欧卡电子智能科技有限公司 一种作业方法、作业设备、装置、存储介质
CN115327914A (zh) * 2022-08-24 2022-11-11 安徽机电职业技术学院 一种基于人造引力场运动模拟的机器人运动规划方法
CN116817948A (zh) * 2023-06-07 2023-09-29 重庆大学 一种基于顶点风险评估的车辆分段路径规划方法
CN117073709A (zh) * 2023-10-17 2023-11-17 福瑞泰克智能系统有限公司 路径规划方法、装置、计算机设备以及存储介质
CN117705123A (zh) * 2024-02-01 2024-03-15 戴盟(深圳)机器人科技有限公司 一种轨迹规划方法、装置、设备及存储介质
CN118010055A (zh) * 2024-02-02 2024-05-10 广州小鹏自动驾驶科技有限公司 行车规划方法、装置、终端设备以及存储介质

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113124891B (zh) * 2021-04-20 2023-05-16 东软睿驰汽车技术（沈阳）有限公司 一种行驶路径规划方法及相关装置
CN113859267B (zh) * 2021-10-27 2023-08-25 广州小鹏自动驾驶科技有限公司 路径决策方法、装置及车辆
CN114200943A (zh) * 2021-12-13 2022-03-18 哈尔滨工业大学芜湖机器人产业技术研究院 一种动态避让方法及移动机器人
CN114184195B (zh) * 2021-12-14 2024-04-26 广州极飞科技股份有限公司 路径搜索方法、装置、无人设备及存储介质
CN114987498B (zh) * 2022-06-10 2023-01-20 清华大学 自动驾驶车辆的拟人化轨迹规划方法、装置、车辆及介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104897168A (zh) * 2015-06-24 2015-09-09 清华大学 基于道路危险评估的智能车路径搜索方法及系统
US20170166204A1 (en) * 2015-12-11 2017-06-15 Hyundai Motor Company Method and apparatus for controlling path of autonomous driving system
CN107702716A (zh) * 2017-08-31 2018-02-16 广州小鹏汽车科技有限公司 一种无人驾驶路径规划方法、系统和装置
CN109855637A (zh) * 2018-12-24 2019-06-07 北京新能源汽车股份有限公司 一种车辆的自动驾驶路径规划方法、装置及设备

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
AT507035B1 (de) * 2008-07-15 2020-07-15 Airbus Defence & Space Gmbh System und verfahren zur kollisionsvermeidung
JP5526717B2 (ja) * 2009-02-27 2014-06-18 日産自動車株式会社 車両用運転操作補助装置、車両用運転操作補助方法および自動車
US8849803B2 (en) * 2011-10-31 2014-09-30 International Business Machines Corporation Data collection for usage based insurance
CN105549597B (zh) * 2016-02-04 2018-06-26 同济大学 一种基于环境不确定性的无人车动态路径规划方法
KR101795250B1 (ko) * 2016-05-03 2017-11-07 현대자동차주식회사 자율주행차량의 주행경로 계획장치 및 방법
CN106371439B (zh) * 2016-09-13 2020-11-20 同济大学 一种统一的自动驾驶横向规划方法与系统
CN107346611B (zh) * 2017-07-20 2021-03-23 北京纵目安驰智能科技有限公司 一种自主驾驶的车辆的避障方法以及避障系统
US10860023B2 (en) * 2018-06-25 2020-12-08 Mitsubishi Electric Research Laboratories, Inc. Systems and methods for safe decision making of autonomous vehicles

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104897168A (zh) * 2015-06-24 2015-09-09 清华大学 基于道路危险评估的智能车路径搜索方法及系统
US20170166204A1 (en) * 2015-12-11 2017-06-15 Hyundai Motor Company Method and apparatus for controlling path of autonomous driving system
CN107702716A (zh) * 2017-08-31 2018-02-16 广州小鹏汽车科技有限公司 一种无人驾驶路径规划方法、系统和装置
CN109855637A (zh) * 2018-12-24 2019-06-07 北京新能源汽车股份有限公司 一种车辆的自动驾驶路径规划方法、装置及设备

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
JIANG JING, ZHU SHUYUN: "Research on Improvement of Path Planning Algorithm for Industrial AGV Forklift", ELECTRONICS WORLD, 15 November 2019 (2019-11-15), pages 19 - 22, XP055828429, DOI: 10.19353/j.cnki.dzsj.2019.21.004 *
See also references of EP4089369A4 *

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114077249A (zh) * 2021-10-22 2022-02-22 陕西欧卡电子智能科技有限公司 一种作业方法、作业设备、装置、存储介质
CN114077249B (zh) * 2021-10-22 2024-03-15 陕西欧卡电子智能科技有限公司 一种作业方法、作业设备、装置、存储介质
CN114030483A (zh) * 2021-12-16 2022-02-11 阿波罗智联(北京)科技有限公司 车辆控制方法、装置、电子设备和介质
CN114030483B (zh) * 2021-12-16 2024-06-07 阿波罗智联(北京)科技有限公司 车辆控制方法、装置、电子设备和介质
CN115327914A (zh) * 2022-08-24 2022-11-11 安徽机电职业技术学院 一种基于人造引力场运动模拟的机器人运动规划方法
CN116817948A (zh) * 2023-06-07 2023-09-29 重庆大学 一种基于顶点风险评估的车辆分段路径规划方法
CN116817948B (zh) * 2023-06-07 2024-05-07 重庆大学 一种基于顶点风险评估的车辆分段路径规划方法
CN117073709A (zh) * 2023-10-17 2023-11-17 福瑞泰克智能系统有限公司 路径规划方法、装置、计算机设备以及存储介质
CN117073709B (zh) * 2023-10-17 2024-01-16 福瑞泰克智能系统有限公司 路径规划方法、装置、计算机设备以及存储介质
CN117705123A (zh) * 2024-02-01 2024-03-15 戴盟(深圳)机器人科技有限公司 一种轨迹规划方法、装置、设备及存储介质
CN117705123B (zh) * 2024-02-01 2024-04-09 戴盟(深圳)机器人科技有限公司 一种轨迹规划方法、装置、设备及存储介质
CN118010055A (zh) * 2024-02-02 2024-05-10 广州小鹏自动驾驶科技有限公司 行车规划方法、装置、终端设备以及存储介质

Also Published As

Publication number Publication date
CN112639849A (zh) 2021-04-09
EP4089369A1 (en) 2022-11-16
EP4089369A4 (en) 2023-02-01

Similar Documents

Publication Publication Date Title
WO2021142799A1 (zh) 路径选择方法和路径选择装置
US11465619B2 (en) Vehicle collision avoidance based on perturbed object trajectories
US11390300B2 (en) Method for using lateral motion to optimize trajectories for autonomous vehicles
WO2021142793A1 (zh) 路径规划方法和路径规划装置
US11225247B2 (en) Collision prediction and avoidance for vehicles
US10606277B2 (en) Speed optimization based on constrained smoothing spline for autonomous driving vehicles
US10816990B2 (en) Non-blocking boundary for autonomous vehicle planning
JP6622148B2 (ja) 周辺環境認識装置
US10591926B2 (en) Smooth road reference for autonomous driving vehicles based on 2D constrained smoothing spline
CN109829351B (zh) 车道信息的检测方法、装置及计算机可读存储介质
EP3517893A1 (en) Path and speed optimization fallback mechanism for autonomous vehicles
US11360480B2 (en) Collision zone detection for vehicles
US20220410939A1 (en) Collision detection method, electronic device, and medium
KR102399963B1 (ko) 측방향 경사가 있는 도로 평면 출력
JP7200371B2 (ja) 車両速度を決定する方法及び装置
US11433885B1 (en) Collision detection for vehicles
JP2021531208A (ja) 車両のための衝突予測及び回避
US20220274625A1 (en) Graph neural networks with vectorized object representations in autonomous vehicle systems
WO2022016941A1 (zh) 行驶装置的避障路径的规划方法和装置
KR102671660B1 (ko) 교통 혼잡 감지 방법, 장치, 전자 기기 및 저장 매체
US11414096B2 (en) QP spline path and spiral path based reference line smoothing method for autonomous driving
WO2023092451A1 (zh) 预测可行驶车道的方法和装置
JP2022522298A (ja) 速度および位置情報を使用するレーダ反射の認識
US11685379B2 (en) Vehicle control device and storage medium storing computer program for vehicle control
CN117485370A (zh) 一种轨迹规划方法、电子设备和存储介质

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: 20914201

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2020914201

Country of ref document: EP

Effective date: 20220812