CN113619602B

CN113619602B - Method for guiding vehicle to run, related system and storage medium

Info

Publication number: CN113619602B
Application number: CN202110971534.9A
Authority: CN
Inventors: 张叶青; 孙凯; 赵冲
Original assignee: Huawei Technologies Co Ltd
Current assignee: Huawei Technologies Co Ltd
Priority date: 2021-08-20
Filing date: 2021-08-20
Publication date: 2023-03-10
Anticipated expiration: 2041-08-20
Also published as: CN113619602A

Abstract

The embodiment of the application provides a method for guiding a vehicle to run, a related system and a storage medium, wherein the method comprises the following steps: acquiring K paths from a departure point to a destination point of a vehicle; constructing evaluation functions of the K paths, and calculating an evaluation value of each reference position in the K paths, wherein the evaluation functions comprise at least one of a transverse evaluation function and a longitudinal evaluation function; when the vehicle is located at a first position in the K paths, determining an evaluation value of the first position and an evaluation value of a second position according to the evaluation value of each reference position, wherein the second position is a position where the first position corresponds to a lane-changing alternative lane; and determining a target path according to the evaluation values of the first position and the second position. By adopting the method, the real-time lane change decision of the automatic driving system has transverse and longitudinal global views, so that the target lane selection and the lane change opportunity decision are more prospective and intelligent, and the reliability of path guidance is improved.

Description

Method for guiding vehicle to run, related system and storage medium

Technical Field

The present application relates to the field of vehicle technologies, and in particular, to a method for guiding a vehicle to travel, a related system, and a storage medium.

Background

In a complex urban road scene, lane-level path planning and navigation guidance directly affect the intelligence, safety and comfort of an automatic driving system. In particular, unlike human driving, which only provides rough semantic-level navigation advice, autopilot systems must rely on explicit, precise, continuously changing navigation information to accomplish the autopilot task. Particularly, if the provided lane change guide and lane change emergency information can have a global view, the automatic driving vehicle can arbitrate navigation lane change based on global tasks and multi-source intentions such as passing lane change based on local scene perception in real time, so that more reasonable and intelligent automatic driving behavior decisions can be made, and intelligent driving behaviors better than human drivers in strange road networks are shown.

The lane-level navigation function in the existing intelligent driving system can only carry out local lane-level guidance according to the information of the nearest intersection in front of the vehicle, meets the lane change requirement of the vehicle when passing through the intersection, but cannot provide global guidance, and discontinuous guidance information cannot meet the real-time lane change decision requirement of the automatic driving vehicle; or only one global guide line capable of processing traffic scenes such as straight going, turning around, intersections and the like is provided, so that global guidance from a specified starting point to a specified destination point of a navigation task is realized, but lane changing positions and the like in a route are fixed, a complete path selection space cannot be provided, and the degree of freedom of real-time dynamic decision of the automatic driving vehicle is limited; or, a plurality of lane-level navigation paths from the departure point to the destination point can be provided, the navigation path with low traffic cost is selected to provide navigation guidance for the automatic driving vehicle, but the provided navigation paths are not suitable for evaluation guidance mechanisms of different scenes, so that the vehicle does not have enough global navigation information to combine with real-time dynamic traffic information to make more prospective and more intelligent driving decisions, and the requirements of real high-real-time high-dynamic road scenes or commercial automatic driving vehicle products cannot be met.

Therefore, the lane-level navigation technology of the existing automatic driving system only provides part of lane-level planning tracks and lane information required by local planning, and lacks a reasonable evaluation and guidance mechanism for multiple alternative lane-level planning paths, so that navigation guidance information which is more complete in selection space, has a global view field, and has continuous comparability in a real-time dynamic scene cannot be provided, and the requirement of the automatic driving system for real-time behavior decision under the global view field in a high-real-time high-dynamic complex urban road scene cannot be met.

Disclosure of Invention

The application discloses a method for guiding a vehicle to run, a related system and a storage medium, which can enable a real-time behavior decision of an automatic driving system to have horizontal and longitudinal global views, provide complete selection space and continuously comparable lane evaluation indexes for real-time dynamic decision, enable target lane selection and lane change opportunity decision to be more prospective and intelligent, and improve the reliability of route guidance.

In a first aspect, an embodiment of the present application provides a method for guiding a vehicle to travel, including: acquiring K paths from a departure point to a destination point of the vehicle according to a path planning result, wherein K is a positive integer; according to the lane topological relation and the lane scene included by the K paths, constructing evaluation functions of the K paths, and calculating an evaluation value of each reference position in the K paths according to the evaluation functions of the K paths, wherein the evaluation functions include at least one of a transverse evaluation function and a longitudinal evaluation function, the transverse evaluation function is used for evaluating a factor influencing a vehicle transverse decision result at any position in the K paths, and the longitudinal evaluation function is used for evaluating a factor influencing a vehicle longitudinal decision result at any position in the K paths; when the vehicle is located at a first position in the K paths, determining an evaluation value of the first position and an evaluation value of a second position according to the evaluation value of each reference position in the K paths, wherein the second position is a position of the first position corresponding to a lane-changing alternative lane; and determining a target path according to the evaluation value of the first position and the evaluation value of the second position so as to guide the vehicle to travel according to the target path.

According to the embodiment of the application, an evaluation function of K paths is constructed according to the lane topological relation and the lane scene of the K paths, the transverse evaluation value and/or the longitudinal evaluation value of each reference position in the K paths are/is calculated according to the evaluation function of the K paths, the evaluation value of a first position is obtained according to the first position of a vehicle, the evaluation value of a second position corresponding to the first position in a lane-changing alternative lane is obtained according to the first position, and then a target path is determined according to the evaluation value of the first position and the evaluation value of the second position so as to guide the vehicle to travel according to the target path. According to the scheme, factors influencing the transverse decision result of the vehicle at any position in any one of K paths from the departure point to the destination point and/or factors influencing the longitudinal decision result of the vehicle at any position in any one path are evaluated, so that the forward-looking range of the vehicle is expanded to the global view during real-time navigation, a complete selection space and continuously comparable lane evaluation values are provided, the vehicle is guided to carry out multi-source lane change intention arbitration, a reasonable path is selected for driving, and more intelligent automatic driving decision compared with a human driver is realized.

Meanwhile, the scheme has better universality, is easy to expand the automatic driving functions such as emergency lane change, side parking and the like and various special road scenes in a mode of adjusting a transverse evaluation function and/or a longitudinal evaluation function, can be suitable for urban roads, high-speed and other structural scenes, has an important effect on navigation of an automatic driving vehicle in an actual high-real-time high-dynamic complex road traffic environment, and can be widely applied to automatic driving automobile navigation.

As an optional implementation manner, the longitudinal evaluation function is related to at least one of a target reachability evaluation function of each of the K paths and a longitudinal penalty evaluation function corresponding to each of the K paths, where the longitudinal penalty evaluation function includes at least one of: a static non-main lane penalty function, a dynamic occupation penalty function, a lane change protection penalty function and an edge parking penalty function.

That is, the longitudinal evaluation function may be related to a target reachability evaluation function for each of the K paths; alternatively, the longitudinal evaluation function may be related to longitudinal penalty evaluation functions respectively corresponding to the K paths; or the longitudinal evaluation function is related to the target reachability evaluation function of each path in the K paths and the longitudinal penalty evaluation function corresponding to each path in the K paths.

By adopting the method, the attributes of any position in the path, which influence the longitudinal decision result of the vehicle, are evaluated through the longitudinal evaluation function, and various factors related to the automatic driving task, the road characteristic, the driving strategy and the driving function design are brought into a unified longitudinal evaluation system for normalization and quantification, so that a clear, concise and continuous longitudinal evaluation standard with comparative expandability is established for the real-time navigation decision of the automatic driving vehicle.

As an alternative implementation manner, the transverse evaluation function is related to at least one of a minimum number of times of lane changing, a transverse offset of a single lane changing, and a lane changing direction change of a continuous lane changing behavior of each of the K paths to reach the destination point.

By adopting the method, the attributes of any position in the path, which influence the transverse decision result of the vehicle, are evaluated through the transverse evaluation function, and multiple factors related to the automatic driving task, the road characteristic, the driving strategy and the driving function design are brought into a unified transverse evaluation system for normalization and quantification, so that a clear, concise and continuous transverse evaluation standard with comparative expandability is established for the real-time navigation decision of the automatic driving vehicle, and a complete transverse and longitudinal evaluation frame is established together with the longitudinal evaluation standard.

As an optional implementation manner, the constructing the evaluation functions of the K paths, and calculating the evaluation value of each reference position in the K paths according to the evaluation functions of the K paths includes: constructing a transverse evaluation function and/or a longitudinal evaluation function of each node in the K paths, wherein the nodes are lane units which form each path in the K paths and have topological connection relations, and each lane unit comprises lane sections with uniformly distributed road topological structures and lane sections formed by dividing the positions of road topological structure changes, the positions of the road topological structure changes and intersections; and calculating the transverse evaluation value and/or the longitudinal evaluation value of each reference position in the K paths according to the transverse evaluation function and/or the longitudinal evaluation function of each node.

With this means, the horizontal evaluation value and/or the vertical evaluation value for each reference position are calculated on a node basis, and the calculation efficiency can be improved.

As an optional implementation manner, the method further includes: obtaining a topological connection relation between nodes according to the nodes of the K paths and the topological relation of lanes contained in the K paths; the calculating a transverse evaluation value and/or a longitudinal evaluation value of each reference position in the K paths according to the transverse evaluation function and/or the longitudinal evaluation function of each node includes: according to the topological connection relation among the nodes, sequentially calculating the transverse evaluation value and/or the longitudinal evaluation value of each node in the K paths from the destination point; and calculating the transverse evaluation value and/or the longitudinal evaluation value of each reference position in the K paths according to the transverse evaluation value and/or the longitudinal evaluation value of each node in the K paths.

The topological connection relation among the nodes can be expressed in a tree-shaped directed graph mode, edges of the tree-shaped directed graph indicate the topological connection relation among the nodes of the tree-shaped directed graph, and the direction of the edges of the tree-shaped directed graph indicates the direction from the departure point to the destination point; the calculating a transverse evaluation value and/or a longitudinal evaluation value of each reference position in the K paths according to the transverse evaluation function and/or the longitudinal evaluation function of each node includes: and sequentially calculating the horizontal evaluation value and/or the vertical evaluation value of each reference position in the K paths from the destination point according to the nodes and the edges in the tree-shaped directed graph.

By adopting the method, the horizontal evaluation value and/or the vertical evaluation value of each node are/is sequentially calculated based on the topological relation, the adjacent relation and the like among the nodes in the tree-shaped directed graph, and the calculation efficiency can be improved.

As an optional implementation manner, sequentially calculating, from the destination point, a horizontal evaluation value and/or a vertical evaluation value of each reference position in the K paths according to the nodes and the edges in the tree-like directed graph, includes: according to the nodes and the edges in the tree-shaped directed graph, sequentially calculating a transverse evaluation value and/or a longitudinal evaluation value of each node in the K paths from the destination point; and calculating the transverse evaluation value and/or the longitudinal evaluation value of each reference position in the K paths according to the transverse evaluation value and/or the longitudinal evaluation value of each node in the K paths.

As an optional implementation manner, when a node i is a non-path end point node, a node j is a node next to the node i along the direction in the tree-like directed graph, and only one node j and the node i are located on one edge, a longitudinal evaluation value of the node i is a sum of a longitudinal evaluation value between the node i and the node j and a longitudinal evaluation value of the node j; when a node i is a non-path end node, and when M nodes and the node i are located on one edge in a next node of the node i along the direction in the tree-like directed graph, determining a longitudinal evaluation value of the node i according to M numerical values, wherein a kth numerical value of the M numerical values is the sum of the longitudinal evaluation value of the kth node and the longitudinal evaluation value between the node i and the kth node, M is an integer not less than 2, i, j, and k are positive integers, k is not greater than M, and k is any one of the M numerical values; and when the node i is a path end node, the longitudinal evaluation value of the node i is a preset value.

By adopting the method, the longitudinal evaluation value of each node is sequentially calculated based on the topological relation, the adjacent relation and the like among the nodes in the tree-shaped directed graph, and the calculation efficiency can be improved.

As an optional implementation manner, the method further includes: determining a node corresponding to the lane where the destination point is located as a reference node, and calculating a transverse evaluation value of the reference node according to a transverse evaluation function of the reference node; if the reference node has a splicing node, calculating a transverse evaluation value of the splicing node according to a transverse evaluation function of the splicing node of the reference node, a transverse evaluation value of the reference node and a transverse evaluation value between the reference node and the splicing node; if the reference node has an adjacent node corresponding to the lane-changing alternative lane, calculating the transverse evaluation value of the adjacent node according to the transverse evaluation function of the adjacent node, the transverse evaluation value of the reference node and the transverse evaluation value between the reference node and the adjacent node; if the adjacent node of the reference node has no splicing node, determining a transverse evaluation value of the adjacent node of the splicing node of the reference node according to a transverse evaluation function of the adjacent node of the splicing node of the reference node, a transverse evaluation value of the splicing node of the reference node and a transverse evaluation value between the splicing node of the reference node and the adjacent node of the splicing node of the reference node; if the adjacent node of the reference node has a continuous node, calculating a first transverse evaluation value of the continuous node of the adjacent node of the reference node according to the transverse evaluation function of the continuous node of the adjacent node of the reference node, the transverse evaluation value of the adjacent node of the reference node and the transverse evaluation value between the adjacent node of the reference node and the continuous node of the adjacent node of the reference node, calculating a second transverse evaluation value of the continuous node of the adjacent node of the reference node according to the transverse evaluation function of the adjacent node of the continuous node of the reference node, the transverse evaluation value of the continuous node of the reference node and the transverse evaluation value between the continuous node of the reference node and the adjacent node of the continuous node of the reference node, determining the transverse evaluation value of the continuous node of the adjacent node of the reference node according to the first transverse evaluation value and the second transverse evaluation value, and so on to obtain the transverse evaluation value of each node in the K paths.

Specifically, if there are x nodes corresponding to the node p among the adjacent nodes in the lane-changing candidate lane, which are the continuation nodes of the reference node, and y nodes corresponding to the reference node, and there are z nodes among the next continuation nodes of the node p in the direction in the tree-like directed graph, which are the continuation nodes of the reference node, and t nodes corresponding to the adjacent nodes of the reference node, x + z first lateral evaluation values of the node p are determined based on the lateral evaluation function of the node p, the lateral evaluation value of the continuation node of the reference node, and the lateral evaluation value between the continuation node of the reference node and the node p, and determining y + t second transverse evaluation values of the node p according to a transverse evaluation function of the node p, a transverse evaluation value of an adjacent node of the reference node and a transverse evaluation value between the adjacent node of the reference node and the node p, determining a transverse evaluation value of the node p according to the x + z first transverse evaluation values and the y + t second transverse evaluation values, and so on to obtain a transverse evaluation value of each node in the K paths, wherein p is a positive integer, x, y, z and t are integers not less than 0, and if x, y, z and t are simultaneously 0, the transverse evaluation value of the node p is a set value.

By adopting the method, the longitudinal evaluation value of each node in each path is calculated iteratively according to the topological relation between the nodes in the tree-shaped directed graph, the navigation information with a longitudinal global view field is provided, and the urgency of the lane change requirement can be represented so as to be used for the lane change decision of the automatic driving system to select a proper lane change opportunity; and calculating the transverse evaluation value of each node in each path according to the topological relation and the adjacent relation among the nodes in the tree-shaped directed graph, providing navigation information with a transverse global view field, and reflecting global transverse motion trend information so as to be used for a lane change decision of an automatic driving system to select a target lane conforming to the global motion trend.

As an optional implementation manner, the reference position is any position in the K paths; or the reference position is a preset position in nodes of the K paths, the nodes are lane units which form each path in the K paths and have topological connection relations, and each lane unit comprises lane sections with uniformly distributed road topological structures and lane sections formed by dividing positions of road topological structure changes, positions of the road topological structure changes and intersections.

The preset position may be a start position, an end position, or an intermediate position of the lane unit corresponding to the node, or may be another preset position of the lane unit.

The above-mentioned reference position is only an example, and it may also be a reference position determined according to, for example, 100m or any other arbitrary distance, and the present solution is not particularly limited thereto. That is, each path gets one reference position every 100m from the start point, and so on.

As an optional implementation manner, when the reference position is a preset position in the nodes of the K paths, the determining, according to the evaluation value of each reference position in the K paths, the evaluation value of the first position and the evaluation value of the second position includes: determining a first node adjacent to the same lane corresponding to the first position and a second node adjacent to the same lane corresponding to the second position; calculating a transverse evaluation value and/or a longitudinal evaluation value of the first position according to the transverse evaluation value and/or the longitudinal evaluation value of the preset position in the first node and the transverse evaluation value and/or the longitudinal evaluation value between the preset position in the first node and the first position; and calculating the transverse evaluation value and/or the longitudinal evaluation value of the second position according to the transverse evaluation value and/or the longitudinal evaluation value of the preset position in the second node and the transverse evaluation value and/or the longitudinal evaluation value between the preset position in the second node and the second position.

That is, the present solution may first calculate a lateral evaluation value and/or a longitudinal evaluation value for each position, and then determine the lateral evaluation value and/or the longitudinal evaluation value for the first position, the second position from the lateral evaluation value and/or the longitudinal evaluation value for each position. By adopting the method, the transverse evaluation value and/or the longitudinal evaluation value of each position in the K paths are pre-calculated once after the path planning is finished, and only the first position and the second position are required to be obtained through position matching in each iteration period of the real-time navigation, so that the calculation resource in the real-time navigation process can be saved, and the calculation efficiency is improved.

As another alternative implementation, the transverse evaluation value and/or the longitudinal evaluation value of the first position and the second position may be calculated based on the transverse evaluation value and/or the longitudinal evaluation value of some specific reference positions.

As an optional implementation manner, the static non-main lane penalty function is calculated according to a lane average angle variation and a lane length of a lane splitting/merging section, where the lane average angle variation is obtained by performing smoothing filtering processing on a lane center point and according to an angle variation of the processed lane center point.

As an alternative implementation, the dynamic occupancy penalty function is determined according to the speed of the vehicle, the position of the vehicle, the special lane borrowing time and the traffic clear state.

As an alternative implementation manner, the lane change protection penalty function is determined according to the speed of the vehicle, the time required for lane change and the number of attempts after the preset lane change failure.

As an alternative implementation manner, the parking penalty function is determined according to a preset distance required for parking near the side.

In a second aspect, an embodiment of the present application provides an apparatus for guiding a vehicle to travel, including: the acquisition module is used for acquiring K paths from the departure point to the destination point of the vehicle according to the path planning result, wherein K is a positive integer; the calculation module is used for constructing evaluation functions of the K paths according to lane topological relations and lane scenes included by the K paths, and calculating an evaluation value of each reference position in the K paths according to the evaluation functions of the K paths, wherein the evaluation functions comprise at least one of a transverse evaluation function and a longitudinal evaluation function, the transverse evaluation function is used for evaluating a factor influencing a vehicle transverse decision result at any position in any path in the K paths, and the longitudinal evaluation function is used for evaluating a factor influencing a vehicle longitudinal decision result at any position in any path in the K paths; the first determining module is used for determining the evaluation value of a first position and the evaluation value of a second position according to the evaluation value of each reference position in the K paths when the vehicle is located at the first position in the K paths, wherein the second position is a position of the first position corresponding to the lane changing alternative; and the second determination module is used for determining a target path according to the evaluation value of the first position and the evaluation value of the second position so as to guide the vehicle to run along the target path.

By adopting the means, the factors influencing the transverse decision result of the vehicle at any position in the path are evaluated through the transverse evaluation function, the factors influencing the longitudinal decision result of the vehicle at any position in the path are evaluated through the longitudinal evaluation function, and various factors related to an automatic driving task, a road characteristic, a driving strategy and a driving function design are integrated into a unified transverse evaluation system and a unified longitudinal evaluation system for normalization and quantification, so that a clear, concise and continuous transverse and longitudinal evaluation standard with comparative expandability is established for the real-time navigation decision of the automatic driving vehicle.

As an optional implementation manner, the calculation module is configured to: constructing a transverse evaluation function and/or a longitudinal evaluation function of each node in the K paths, wherein the nodes are lane units which form each path in the K paths and have topological connection relations, and each lane unit comprises lane sections with uniformly distributed road topological structures and lane sections formed by dividing road topological structure change positions, lane topological structure change positions and intersections; and calculating the transverse evaluation value and/or the longitudinal evaluation value of each reference position in the K paths according to the transverse evaluation function and/or the longitudinal evaluation function of each node.

As an optional implementation manner, the apparatus further includes: the processing module is used for obtaining a topological connection relation between the nodes according to the nodes of the K paths and the topological relation between the lanes contained in the K paths; the calculation module is configured to: according to the topological connection relation among the nodes, sequentially calculating the transverse evaluation value and/or the longitudinal evaluation value of each node in the K paths from the destination point; and calculating the transverse evaluation value and/or the longitudinal evaluation value of each reference position in the K paths according to the transverse evaluation value and/or the longitudinal evaluation value of each node in the K paths.

The topological connection relation among the nodes can be expressed in a tree-shaped digraph mode, edges of the tree-shaped digraph indicate the topological connection relation among the nodes of the tree-shaped digraph, and the direction of the edges of the tree-shaped digraph indicates the direction from the departure point to the destination point; the calculation module is configured to: according to the nodes and edges in the tree-shaped directed graph, sequentially calculating the horizontal evaluation value and/or the vertical evaluation value of each node in the K paths from the destination point; and calculating the transverse evaluation value and/or the longitudinal evaluation value of each reference position in the K paths according to the transverse evaluation value and/or the longitudinal evaluation value of each node in the K paths.

As an optional implementation manner, the computing module is further configured to: determining a node corresponding to a lane where the destination point is located as a reference node, and calculating a transverse evaluation value of the reference node according to a transverse evaluation function of the reference node; if the reference node has a splicing node, calculating a transverse evaluation value of the splicing node according to a transverse evaluation function of the splicing node of the reference node, a transverse evaluation value of the reference node and a transverse evaluation value between the reference node and the splicing node; if the reference node has an adjacent node corresponding to the lane-changing alternative lane, calculating the transverse evaluation value of the adjacent node according to the transverse evaluation function of the adjacent node, the transverse evaluation value of the reference node and the transverse evaluation value between the reference node and the adjacent node; if the adjacent node of the reference node has no splicing node, determining a transverse evaluation value of the adjacent node of the splicing node of the reference node according to a transverse evaluation function of the adjacent node of the splicing node of the reference node, a transverse evaluation value of the splicing node of the reference node and a transverse evaluation value between the splicing node of the reference node and the adjacent node of the splicing node of the reference node; if the adjacent node of the reference node has a continuous node, calculating a first transverse evaluation value of the continuous node of the adjacent node of the reference node according to the transverse evaluation function of the continuous node of the adjacent node of the reference node, the transverse evaluation value of the adjacent node of the reference node and the transverse evaluation value between the adjacent node of the reference node and the continuous node of the adjacent node of the reference node, calculating a second transverse evaluation value of the continuous node of the adjacent node of the reference node according to the transverse evaluation function of the adjacent node of the continuous node of the reference node, the transverse evaluation value of the continuous node of the reference node and the transverse evaluation value between the continuous node of the reference node and the adjacent node of the continuous node of the reference node, determining the transverse evaluation value of the continuous node of the adjacent node of the reference node according to the first transverse evaluation value and the second transverse evaluation value, and so on to obtain the transverse evaluation value of each node in the K paths.

As an alternative implementation, the dynamic occupancy penalty function is determined according to the speed of the vehicle, the position of the vehicle, the special lane borrowing time and the traffic free state.

In a third aspect, an embodiment of the present application provides an apparatus for guiding a vehicle to travel, including a processor and a memory; wherein the memory is configured to store program code and the processor is configured to call the program code to perform any one of the possible embodiments according to the first aspect.

In a fourth aspect, the present application provides a computer-readable storage medium, where a computer program is stored, and the computer program is executed by a processor to perform any one of the possible implementation manners as described in the first aspect.

In a fifth aspect, the present application provides a computer program product, which when run on a computer, causes the computer to perform any one of the possible embodiments according to the first aspect.

In a sixth aspect, an embodiment of the present application provides a chip system, where the chip system is applied to an electronic device; the chip system comprises one or more interface circuits, and one or more processors; the interface circuit and the processor are interconnected through a line; the interface circuit is to receive a signal from a memory of the electronic device and to send the signal to the processor, the signal comprising computer instructions stored in the memory; when the processor executes the computer instructions, the electronic device performs any one of the possible embodiments of the first aspect.

In a seventh aspect, an embodiment of the present application provides an intelligent driving vehicle, which includes a traveling system, a sensing system, a control system, and a computer system, where the computer system is configured to execute any one of the possible implementation manners of the first aspect.

It is to be understood that the apparatus of the second aspect, the apparatus of the third aspect, the computer storage medium of the fourth aspect, or the computer program product of the fifth aspect, the chip system of the sixth aspect, and the smart driving vehicle of the seventh aspect, all provided above, are adapted to perform the method of the first aspect. Therefore, the beneficial effects achieved by the method can refer to the beneficial effects in the corresponding method, and the details are not repeated here.

Drawings

The drawings used in the embodiments of the present application are described below.

FIG. 1 is a schematic diagram of a system architecture for guiding a vehicle to run according to an embodiment of the present disclosure;

FIG. 2 is a schematic flow chart of a method for guiding a vehicle to run according to an embodiment of the present application;

fig. 3 is a schematic diagram of a path planning result according to an embodiment of the present application;

FIG. 4a is a schematic view of a lane scene provided in an embodiment of the present application;

FIG. 4b is a schematic diagram of a tree-like directed graph corresponding to FIG. 4a according to an embodiment of the present application;

fig. 5a is a schematic node connection diagram provided in an embodiment of the present application;

fig. 5b is a schematic diagram of another node connection provided in the embodiment of the present application;

FIG. 6a is a schematic diagram of a reference node connection according to an embodiment of the present application;

FIG. 6b is a schematic diagram of another reference node connection provided in an embodiment of the present application;

FIG. 6c is a schematic diagram of a reference node connection according to an embodiment of the present application;

FIG. 7 is a schematic diagram of determining a second position provided by an embodiment of the present application;

FIG. 8 is a schematic diagram of a method for guiding a vehicle to travel according to an embodiment of the present application;

FIG. 9 is a schematic view of a lane scene provided by an embodiment of the present application;

FIG. 10 is a schematic view of a scenario for guiding a vehicle to run according to an embodiment of the present application;

FIG. 11 is a schematic diagram of a tree-like directed graph corresponding to FIG. 10 according to an embodiment of the present application;

FIG. 12 is a schematic view of a vehicle according to an embodiment of the present disclosure;

FIG. 13 is a schematic view of another example of a scenario for guiding a vehicle to travel according to an embodiment of the present disclosure;

fig. 14 is a schematic structural diagram of an apparatus for guiding a vehicle to run according to an embodiment of the present application;

fig. 15 is a schematic structural diagram of another device for guiding a vehicle to run according to an embodiment of the present application.

Detailed Description

The embodiments of the present application will be described below with reference to the drawings. The terminology used in the description of the embodiments herein is for the purpose of describing particular embodiments herein only and is not intended to be limiting of the application.

Referring to fig. 1, a schematic diagram of a system architecture for guiding a vehicle to run according to an embodiment of the present application is shown. The system may include a lane-level path planning module, a lane-change guidance module for a global field of view, and a lane-change intent arbitration and lane-change decision generation module. The lane-level path planning module determines K paths according to the high-precision map and the departure point and the destination point of the driving task, wherein K is a positive integer. The lane change guiding module of the global view carries out scene classification on all lanes contained in the K paths according to high-precision map information such as lane topological relations and attributes contained in the K paths, and builds scene topological relations on scenes such as lane splitting, merging, intersections and special lanes; then, respectively constructing an evaluation function for each path based on a lane topological relation and a lane scene, wherein the evaluation function comprises at least one of a transverse evaluation function and a longitudinal evaluation function, and the longitudinal evaluation function is used for longitudinal global guidance generation, and is specifically used for evaluating factors influencing a vehicle longitudinal decision result at any position in any one of the K paths; and the transverse evaluation function is used for transverse global guide generation, and is particularly used for evaluating factors influencing the transverse decision result of the vehicle at any position in any path of the K paths. For example, the longitudinal evaluation function may represent the urgency of lane change requirements, and is used for deciding the execution time and the urgency of driving behaviors such as a car following or lane change at the current moment, an immediate lane change or a next lane change at the next moment, an urgent lane change or a gentle lane change and the like; the transverse evaluation function can represent global transverse behavior trend information, particularly a future movement trend for completing a navigation task from a current position to a destination point, and is used for deciding whether lane changing is needed at the current moment, whether lane changing is left-handed or right-handed, and the movement direction of driving behaviors and a target lane are larger in income and the like. Then, in the real-time navigation process, real-time lane matching is carried out according to the current position of the vehicle, the current lane of the vehicle and the selectable lane-changing alternative lane are obtained, and the transverse evaluation value and/or the longitudinal evaluation value of the current lane and all the alternative lanes are obtained, so that the lane-changing intention arbitration and lane-changing decision generation module guides the vehicle to run according to the target path, and the automatic driving decision which is more intelligent than a human driver is realized. The target path is a path with the maximum automatic driving income at the current moment determined according to the navigation lane-changing intention generated by the transverse evaluation value and/or the longitudinal evaluation value of the current lane and all the alternative lanes, other lane-changing intentions generated by real-time dynamic environment perception, the real-time traffic jam state of the road and the like, and comprises the target lane and the corresponding lane-changing time thereof.

The above description is only given by taking an example of the application of the embodiment of the present application to an automatic driving scenario. The method for guiding the vehicle to run provided by the application can also be applied to an auxiliary driving scene, and the scheme is not particularly limited in this respect.

The embodiment can be executed by a vehicle-mounted device (such as a vehicle-mounted device), and can also be executed by terminal equipment such as a mobile phone and a computer. The present embodiment is not particularly limited to this.

It should be noted that the method for guiding the vehicle to run provided by the present application may be executed locally or by the cloud. The cloud end can be realized by a server, the server can be a virtual server, an entity server and the like, and can also be other devices, and the scheme is not particularly limited to this.

Referring to fig. 2, a schematic flowchart of a method for guiding a vehicle to run according to an embodiment of the present application is shown. As shown in fig. 2, the method includes steps 201 to 204, which are specifically as follows:

201. acquiring K paths from the departure point to the destination point of the vehicle according to the path planning result;

preferably, K paths between the vehicle from the departure point to the destination point are obtained according to the road-level global path planning result, wherein the K paths are paths corresponding to the lane-level global path planning result, and K is a positive integer.

The above road-level global path planning result can be understood as: based on a road network model described by a standard precision or high precision map, combining a given departure point and a destination point of an automatic driving task, and planning an optimal path which can reach the destination point from the departure point and is formed by a series of road segments which are connected in a front-back topological way according to performance indexes (shortest time, shortest distance, fewest traffic light intersections, largest intelligent driving area range and the like).

The lane-level global path planning result can be understood as being formed by a plurality of lane segments contained in all road segments in the same road-level global path planning result, and all paths which are in mutual inclusion relation but cover the road-level global path planning result as much as possible along the direction of the road-level global path planning result and ensure that all the lane segments are connected in a front-back topological way.

The lane-level global path planning result can represent not only the front-back topological connection relation among the lane segments in the road-level global path planning result, but also the attribute information of the lane segments, so that richer and more accurate lane-level navigation guidance is provided, including whether the current lane can be changed to the left/right, the remaining length of the current position broken line lane change, the distance from the current position to the intersection/terminal point, and the like.

Specifically, a road-level global path planning result and a lane-level global path planning result can be obtained according to the high-precision map, the departure point, the passing point and the destination point.

For example, as shown in fig. 3, the result of the road-level global path planning from the departure point (e.g., point a in fig. 3) to the destination point (e.g., point B in fig. 3) is: road 1-1 → road 1-2 → road 1-3 → road 1-4.

The lane-level global path planning result from the departure point to the destination point is as follows:

route 1: the lane 1 longitudinal communication of 1-1, 1-2, 1-3 by the road constitutes, marks: R1-1L1 → R1-2L1 → R1-3L1;

the R1-1L1 represents lane 1 on road 1-1, correspondingly, R1-2L1 represents lane 1 on road 1-2, and R1-3L1 represents lane 1 on road 1-3.

Route 2: the lane 2 of the roads 1-1, 1-2 and 1-3 is longitudinally communicated and recorded as: R1-1L2 → R1-2L2 → R1-3L2;

similarly, R1-1L2 represents lane 2 on road 1-2, R1-2L2 represents lane 2 on road 1-2, and R1-3L2 represents lane 2 on road 1-3.

Route 3: the lane 3 of the roads 1-1, 1-2 and 1-3 is longitudinally communicated and recorded as: R1-1L3 → R1-2L3 → R1-3L3;

path 4: the lane 1 of the roads 1-1, 1-2, 1-3 and the lane 4 of the roads 1-4 are communicated and recorded as: R1-1L4 → R1-2L4 → R1-3L4 → R1-4L1.

In this case, K is 4.

202. According to the lane topological relation and the lane scene included by the K paths, constructing evaluation functions of the K paths, and calculating an evaluation value of each reference position in the K paths according to the evaluation functions of the K paths, wherein the evaluation functions include at least one of a transverse evaluation function and a longitudinal evaluation function, the transverse evaluation function is used for evaluating a factor influencing a vehicle transverse decision result at any position in the K paths, and the longitudinal evaluation function is used for evaluating a factor influencing a vehicle longitudinal decision result at any position in the K paths;

the topological relation of the lanes, namely the position relation of each lane in the physical world, comprises the relations of left-right adjacency, front-back continuous connection and the like among the lanes.

Lane scene: according to the information such as the topological relation and the attribute of the lanes, all the lanes contained in the lane-level path planning result are subjected to scene classification, and the classification can be divided into parallel lanes, lane separation/combination, to-be-turned areas, intersections, special lanes and the like.

Specifically, if the number of lanes in two sections of roads which are connected in a front-back topological manner is the same, and the lanes in each section are continuously connected in a one-to-one manner, parallel lanes are formed; if the number of lanes contained in the front road and the rear road is different and a certain lane has a plurality of subsequent lanes, the scene is a lane separation scene, otherwise, if the certain lanes only have one identical subsequent lane, the scene is a lane combination scene; if a certain section of lane in the road section is a bus lane, a stop lane and the like, a special lane scene is obtained.

The evaluation function includes at least one of a horizontal evaluation function and a vertical evaluation function.

And the longitudinal evaluation function is used for evaluating the factors influencing the longitudinal decision result of the vehicle at any position in any one of the K paths.

For example, the longitudinal evaluation function may characterize the urgency of lane change requirements, and is used to decide whether the current time is a car following or lane change, whether the current time is an immediate or next lane change, whether the current time is an urgent or gentle lane change, and the like. That is, the longitudinal evaluation function is used to evaluate the execution timing and urgency of the driving behavior decision of the vehicle.

And the transverse evaluation function is used for evaluating the factors influencing the transverse decision result of the vehicle at any position in any one of the K paths.

For example, the lateral evaluation function may characterize global lateral behavior trend information, especially a future movement trend for completing a navigation task from a current position to a destination point, and is used for deciding whether a lane change is needed at the current time, whether a lane change to the left or the right is more profitable, and the like. That is, the lateral evaluation function is used to evaluate and guide the movement direction of the driving behavior of the vehicle and the target lane.

By adopting the means, the factors influencing the transverse decision-making result of the vehicle at any position in the path are evaluated through the transverse evaluation function, the factors influencing the longitudinal decision-making result of the vehicle at any position in the path are evaluated through the longitudinal evaluation function, and various factors related to an automatic driving task, a road characteristic, a driving strategy and a driving function design are integrated into a unified transverse evaluation system and a unified longitudinal evaluation system for normalization and quantification, so that clear, concise and continuous transverse and longitudinal evaluation criteria with comparability are established for the real-time navigation decision of the automatic driving vehicle.

The reference position may be any position in the K paths. That is, the evaluation value of each position in the K paths is calculated according to the evaluation functions of the K paths.

Alternatively, the reference position is a preset position among nodes of the K paths. The nodes are lane units which form each path in the K paths and have topological connection relations, and each lane unit comprises lane sections with uniformly distributed road topological structures and lane sections formed by dividing road topological structure changing positions, lane topological structure changing positions and intersections.

The road topology change may be, for example, road merging, road splitting, or the like. The lane topology change may be, for example, lane merging, lane splitting, etc.

That is, the nodes in each path are a series of topologically connected lane units of the path. The lanes of each path in the K paths are divided into the nodes, so that subsequent calculation is simplified, and the calculation efficiency is improved.

Preferably, the constructing the evaluation functions of the K paths and calculating the evaluation value of each reference position in the K paths according to the evaluation functions of the K paths includes:

constructing a transverse evaluation function and/or a longitudinal evaluation function of each node in the K paths, wherein the nodes are lane units which form each path in the K paths and have topological connection relations, and each lane unit comprises lane sections with uniformly distributed road topological structures and lane sections formed by dividing the positions of road topological structure changes, the positions of the road topological structure changes and intersections;

and calculating the transverse evaluation value and/or the longitudinal evaluation value of each reference position in the K paths according to the transverse evaluation function and/or the longitudinal evaluation function of each node.

Specifically, the above calculating the horizontal evaluation value and/or the vertical evaluation value of each reference position in the K paths according to the horizontal evaluation function and/or the vertical evaluation function of each node may be:

calculating a transverse evaluation value and/or a longitudinal evaluation value of each node in the K paths according to the transverse evaluation function and/or the longitudinal evaluation function of each node;

and obtaining the transverse evaluation value and/or the longitudinal evaluation value of each reference position in the K paths according to the transverse evaluation value and/or the longitudinal evaluation value of each node in the K paths.

When the transverse evaluation value and/or the longitudinal evaluation value of each reference position in the K paths are obtained according to the transverse evaluation value and/or the longitudinal evaluation value of each node in the K paths, a target node close to the current reference position and in the same lane can be determined from the nodes contained in the K paths;

and determining a transverse evaluation value and/or a longitudinal evaluation value corresponding to the current reference position according to the transverse evaluation value and/or the longitudinal evaluation value between the target node and the current reference position and the transverse evaluation value and/or the longitudinal evaluation value of the target node.

By analogy, the transverse evaluation value and/or the longitudinal evaluation value of each reference position can be calculated.

The evaluation value of the node represents the total evaluation value from the node to the destination point; and the evaluation value between a node and a certain position represents the local evaluation value from the certain position to the node.

According to the scheme, each path in the K paths is divided into a plurality of nodes, a transverse evaluation value and/or a longitudinal evaluation value of each node are obtained by constructing a transverse evaluation function and/or a longitudinal evaluation function of each node, and then the transverse evaluation value and/or the longitudinal evaluation value of each reference position are obtained based on each node.

Preferably, a tree-shaped directed graph is obtained according to the topological relations between the nodes of the K paths and the lanes contained in the K paths, the edges of the tree-shaped directed graph indicate the topological connection relations between the nodes of the tree-shaped directed graph, and the direction of the edges of the tree-shaped directed graph indicates the direction from the departure point to the destination point;

the calculating a transverse evaluation value and/or a longitudinal evaluation value of each reference position in the K paths according to the transverse evaluation function and/or the longitudinal evaluation function of each node includes:

according to the nodes and edges in the tree-shaped directed graph, sequentially calculating the horizontal evaluation value and/or the vertical evaluation value of each node in the K paths from the destination point;

and calculating the transverse evaluation value and/or the longitudinal evaluation value of each reference position in the K paths according to the transverse evaluation value and/or the longitudinal evaluation value of each node in the K paths.

That is, the horizontal evaluation value and/or the vertical evaluation value of each reference position is calculated based on the constructed tree-like directed graph.

As shown in fig. 4a, a schematic view of a lane scene provided in the embodiment of the present application is shown. Fig. 4b is a schematic diagram of a tree-like directed graph corresponding to fig. 4a according to an embodiment of the present application. The lane segments with the uniformly distributed road topology structure in fig. 4a (for example, the lane segments with the uniformly distributed road topology structure in each of the path 1, the path 2, and the path 3) and the lane segments formed by the positions of the lane topology structure changes (for example, the lane merging corresponding to the path 4 in the path 2, the lane splitting corresponding to the path 5 in the path 4), and the intersection splitting (for example, the intersection splitting corresponding to the road 5 and the road 6) are all used as nodes, and the nodes are connected by directed edges, so as to construct a tree-shaped directed graph, as shown in fig. 4 b.

Where node L1-1 represents lane 1 in road 1 and node L1-2 represents lane 2 in road 1. Accordingly, node L2-1 represents lane 1 in road 2. The figure indicates the direction from the departure point to the destination point.

The above is merely an example, and other forms of division may also be possible, for example, one node is arranged at an interval of 100m, and the present solution is not limited in this respect.

And sequentially calculating the horizontal evaluation value and/or the vertical evaluation value of each node in the K paths from the destination point according to the nodes and the edges in the tree-shaped directed graph. The following may be used:

as shown in fig. 5a, when a node i is a non-path end point node, a node j is a node next to the node i in the direction of the tree-like directed graph, and only one node j and the node i are located on one edge, the longitudinal evaluation value of the node i is a sum of the longitudinal evaluation values of the node i and the node j.

When a node i is a non-path end node, and when M nodes in a next node of the node i along the direction in the tree-like directed graph and the node i are located on one edge, determining a longitudinal evaluation value of the node i according to M numerical values, wherein a kth numerical value of the M numerical values is the sum of the longitudinal evaluation value of the kth node and the longitudinal evaluation value between the node i and the kth node, M is an integer not less than 2, i, j, and k are positive integers, k is not greater than M, and k is any one of the M numerical values.

As shown in fig. 5b, the node i is a non-path end point node, and there are 3 nodes located on one edge with the node i along the direction in the tree-like directed graph, and the longitudinal evaluation value of the node i is an optimal value of the sum of the longitudinal evaluation value of the node j1 and the longitudinal evaluation values between the node j1 and the node i, the sum of the longitudinal evaluation value of the node j2 and the longitudinal evaluation values between the node j2 and the node i, and the sum of the longitudinal evaluation value of the node j3 and the longitudinal evaluation values between the node j3 and the node i.

The optimal value may be understood as a maximum value of the three and corresponding values, or a minimum value of the three and corresponding values, and the present disclosure is not particularly limited thereto.

When the node i is a path end point node, that is, the node i does not exist along the direction in the tree-shaped directed graph and the next node located on one edge, the longitudinal evaluation value of the node i is a preset value. The preset value may be, for example, 0, or any other set value, and this is not specifically limited in this embodiment.

The above description takes the calculation of the vertical evaluation value of each node as an example.

Calculating the transverse evaluation value of each node according to the transverse evaluation function of each node in the K paths, wherein the method comprises the following steps:

determining a node corresponding to the lane where the destination point is located as a reference node, and calculating a transverse evaluation value of the reference node according to a transverse evaluation function of the reference node;

if the reference node has a splicing node, calculating a transverse evaluation value of the splicing node according to a transverse evaluation function of the splicing node of the reference node, a transverse evaluation value of the reference node and a transverse evaluation value between the reference node and the splicing node;

if the reference node has an adjacent node corresponding to the lane-changing alternative lane, calculating the transverse evaluation value of the adjacent node according to the transverse evaluation function of the adjacent node, the transverse evaluation value of the reference node and the transverse evaluation value between the reference node and the adjacent node;

if the adjacent node of the reference node has no splicing node, determining a transverse evaluation value of the adjacent node of the splicing node of the reference node according to a transverse evaluation function of the adjacent node of the splicing node of the reference node, a transverse evaluation value of the splicing node of the reference node and a transverse evaluation value between the splicing node of the reference node and the adjacent node of the splicing node of the reference node;

if the adjacent node of the reference node has a splicing node, calculating a first transverse evaluation value of the splicing node of the adjacent node of the reference node according to the transverse evaluation function of the splicing node of the adjacent node of the reference node, the transverse evaluation value of the adjacent node of the reference node and the transverse evaluation value between the adjacent node of the reference node and the splicing node of the adjacent node of the reference node, calculating a second transverse evaluation value of the splicing node of the adjacent node of the reference node according to the transverse evaluation function of the adjacent node of the splicing node of the reference node, the transverse evaluation value of the splicing node of the reference node and the transverse evaluation value between the splicing node of the reference node and the adjacent node of the splicing node of the reference node, determining the transverse evaluation value of the splicing node of the adjacent node of the reference node according to the first transverse evaluation value and the second transverse evaluation value, and so on, so as to obtain the transverse evaluation value of each node in the K paths.

The adjacent node of the reference node and the successive node of the adjacent node of the reference node are the same node, or the successive node of the reference node and the successive node of the adjacent node of the reference node are the same node, or the node located on one edge with the reference node includes the adjacent node of the successive node and the successive node of the reference node.

As shown in fig. 6a, the reference node is S1, the continuation node of the reference node S1 is C1, the neighboring node of the reference node S1 is S2, and the continuation node of the neighboring node of the reference node S1 is C2.

In fig. 6a, the neighboring node C2 of the continuation node C1 of the reference node S1 and the continuation node C2 of the neighboring node S2 of the reference node S1 are the same node.

As shown in fig. 6b, the continuation node C1 of the reference node S1 is the same node as the continuation node C1 of the adjacent node S2 of the reference node S1.

As shown in fig. 6C, the nodes located on one edge with the reference node S1 include a continuous node C1 and an adjacent node C2 of the continuous node C1 of the reference node S1.

Specifically, if a node p corresponds to x consecutive nodes of the reference node and y consecutive nodes of the reference node in the lane-changing candidate lane, and z consecutive nodes of the reference node and t consecutive nodes of the node p are next consecutive nodes in the direction of the tree-like directed graph, x + z first lateral evaluation values of the node p are determined according to the lateral evaluation function of the node p, the lateral evaluation value of the consecutive node of the reference node, and the lateral evaluation value between the consecutive node of the reference node and the node p, y + t second lateral evaluation values of the node p are determined according to the lateral evaluation function of the node p, the lateral evaluation value of the neighboring node of the reference node, and the lateral evaluation value between the neighboring node of the reference node and the node p, and the lateral evaluation value of the node p is determined according to the x + z first lateral evaluation values and the y + t second lateral evaluation values, and the like, the y + t second lateral evaluation values of the node p are obtained by taking the path of the x + z first lateral evaluation values and the y + t second lateral evaluation values as an integer, wherein y, y is an integer, and y is not less than y, and z is set as a positive integer.

By adopting the method, the longitudinal evaluation value of each node in each path is calculated iteratively according to the topological relation between the nodes in the tree-shaped directed graph, and the navigation information with the longitudinal global visual field is provided, so that the urgency of the lane change requirement can be represented, and the lane change decision of the automatic driving system can select the appropriate lane change time; and calculating the transverse evaluation value of each node in each path according to the topological relation and the adjacent relation among the nodes in the tree-shaped directed graph, providing navigation information with a transverse global view field, and reflecting global transverse motion trend information so as to be used for a lane change decision of an automatic driving system to select a target lane conforming to the global motion trend.

The longitudinal evaluation function is related to at least one of a target reachability evaluation function of each of the K paths and a longitudinal penalty evaluation function corresponding to each of the K paths, wherein the longitudinal penalty evaluation function includes at least one of: a static non-main lane penalty function, a dynamic occupation penalty function, a lane change protection penalty function and an edge parking penalty function.

Specifically, the target reachability evaluation function is used to evaluate the cumulative remaining travelable distance that the vehicle can reach the destination point. The shorter the remaining travelable distance of the route is, the more urgent the lane change requirement for lane change from the current route is, and the higher priority quick response is required in the lane change arbitration decision of real-time navigation.

If a certain route cannot reach the end point without changing the route, the target reachability evaluation function is the cumulative travelable distance to the end (end point) of the route.

The longitudinal penalty evaluation function is used for representing the driving cost caused by the lane scene.

For example, a static non-dominant road penalty function characterizes additional costs due to static road scenes such as road geometric topology and attributes, such as an increase in along-the-road distance due to road tortuosity.

The influence on the vehicle behavior decision is only related to a lane scene, and a scene which is not related to the real-time driving state of the vehicle is defined as a static scene, otherwise, the scene is a dynamic scene. For example, parallel lanes, lane splitting/merging and the like related to lane topology are all static scenes and are not related to traffic time, traffic density, own vehicle state and the like; however, the traffic rules of the scenes such as the waiting area, the intersection, the bus lane and the like are related to the traffic light state, whether the traffic time is in the peak time period, the occupation duration of the special lane and the like, and are all dynamic scenes.

The static non-main lane penalty function can be obtained by calculating according to the average angle variation of the lane and the length of the lane splitting/merging section, wherein the average angle variation of the lane is obtained by performing smooth filtering processing on the center point of the lane and according to the angle variation of the processed center point of the lane.

The dynamic occupation penalty function is an extra cost caused by vehicle behaviors and dynamic scenes, such as parking beside, lane changing, temporary occupation of bus lane steering and the like.

The dynamic occupation penalty function can be determined according to the speed of the vehicle, the position of the vehicle, the special lane borrowing time and the traffic free state.

According to the functional design of the automatic driving system, the functions of lane changing protection, parking close to the side and the like can be expanded, so that the lane changing protection distance cost and the parking close to the side are introduced, and the functions of the automatic driving system such as continuous lane changing and parking close to the side can be finished by a vehicle.

The lane change protection distance cost can be determined by the average running speed of the vehicle, the average lane change time, the driving experience, the functional design of an automatic driving system and the like. The lane-change protection distance is understood to be the distance from the starting position of the last lane-change to the end of the route, which is reserved by the vehicle for reaching the destination point when approaching the end of each route in order to ensure that the last lane-change is successfully executed.

The side parking distance penalty can be understood as the distance required for side parking. The distance required by the side parking can be a preset fixed value, and can also be determined according to factors such as the type of the vehicle, the position of a lane where the terminal is located, the speed limit of a road where the terminal is located and the like.

The global longitudinal evaluation value of the path can be obtained by subtracting a static non-main channel penalty value, a dynamic occupation penalty value, a channel change protection penalty value, an edge parking penalty value and the like from the target reachability evaluation value. Of course, other calculation methods may also be adopted, for example, the global longitudinal evaluation value is calculated based on a preset weight value, and the present scheme is not particularly limited to this.

The lower the global longitudinal evaluation value of the path is, the more urgent the lane change requirement for lane change from the current path is, and the lane change arbitration decision of real-time navigation needs to obtain a quick response with higher priority, so that the vehicle should be guided to change the lane from the path with the lower global longitudinal evaluation value to the path with the higher global longitudinal evaluation value.

And the transverse evaluation function is related to at least one of the minimum channel changing times of each path in the K paths to reach the destination point, the transverse offset of single channel changing and the channel changing direction change of continuous channel changing behaviors.

The minimum lane change times of each path to the destination point can be determined according to road topological relation, lane change attributes and the like.

Specifically, the minimum number of conversion passes for each path to reach the destination point may be determined based on the following manner:

according to the road topological relation, if the current lane is connected with the subsequent lane, the lane changing times of the current lane inherits the lane changing times of the subsequent lane; if the lane change profit is a special scene such as road separation, the optimal value of the lane change profit of the subsequent lane is inherited, for example, the minimum lane change times of the subsequent lane is inherited. Wherein, the subsequent lane is the next subsequent lane from the current lane to the destination point.

If the current scene is the intersection, selecting at least one lane as a reference lane according to intersection topology selection strategies such as straightness, left alignment, right alignment and the like, and inheriting the minimum lane change times of the corresponding lane behind the intersection; and if the current lane is not connected with the subsequent lane, deducing the lane change times of the adjacent lane from the lane with the determined transverse lane change times according to the left-right adjacent relation of the lane and the lane change benefit.

The transverse offset of the single lane change can be determined based on the geometric position relationship of the lanes of the original lane and the target lane of the lane change behavior, and the transverse offset of the lane centers of the original lane and the target lane at the starting position and the ending position of the single lane change in the direction perpendicular to the driving direction of the lane is represented.

The lane change direction change of the continuous lane change behavior may be determined based on the lane change direction executed by the multiple continuous lane change behaviors, and represents whether the multiple lane change behaviors are multiple continuous lane changes in the same direction or multiple retrace lane changes in different directions, for example: and changing the tracks to the left/right for multiple times continuously, or changing the tracks to the left/right and then to the right/left for multiple times repeatedly.

The global transverse evaluation value of the path can be obtained by processing the minimum lane change times of each path to the destination point, the transverse offset of single lane change, the lane change direction change of continuous lane change behaviors and the like based on a preset algorithm. For example, the lane change is obtained by normalizing the minimum lane change times, the lateral offset of a single lane change, and the lane change direction change of the continuous lane change behavior. Specifically, the global lateral evaluation value may be calculated based on a preset weight value and the like, which is not specifically limited in this embodiment.

203. When the vehicle is located at a first position in the K paths, determining an evaluation value of the first position and an evaluation value of a second position according to the evaluation value of each reference position in the K paths, wherein the second position is a position where the first position corresponds to a lane-changing alternative lane;

specifically, the evaluation value of the first position and the evaluation value of the second position are obtained based on the evaluation value of each reference position by acquiring the lane-changeable alternative lane corresponding to the first position and determining the position of the first position corresponding to the lane-changeable alternative lane.

The first position corresponds to a position in the lane-changeable alternative lane, and it can be understood that the position, in the K paths, other than the first position intersects with a first straight line, the first straight line passes through the first position and is perpendicular to the direction of the road where the first position is located, and the driving direction of the lane where the second position is located is consistent with the driving direction of the lane where the first position is located.

As shown in fig. 7, the vehicle is located at a first position p of the lane 1, and a straight line L perpendicular to the lane direction is drawn through the position p, wherein L intersects with the lane 1, the lane 2, and the lane 3. Since the traveling direction of the lane 5 is opposite to the traveling direction of the lane 1, the lanes 4 and 5 are excluded. That is, the position q1 in the lane 2 and the position q2 in the lane 3 are both the second positions.

The second position may be one or more, and this is not specifically limited in this embodiment.

Specifically, when the reference position is any one of the K paths, the transverse evaluation value and/or the longitudinal evaluation value of the first position and the second position may be directly determined from the transverse evaluation value and/or the longitudinal evaluation value of each reference position calculated in step 202. By adopting the method, the transverse evaluation value and/or the longitudinal evaluation value of each position in the K paths are pre-calculated once after the path planning is finished, and only the first position and the second position are required to be obtained through position matching in each iteration period of the real-time navigation, so that the calculation resource in the real-time navigation process can be saved, and the calculation efficiency is improved.

When the reference position is a preset position in the nodes of the K paths, determining the evaluation value of the first position and the evaluation value of the second position according to the evaluation value of each reference position in the K paths includes:

determining a first node adjacent to the same lane corresponding to the first position and a second node adjacent to the same lane corresponding to the second position;

calculating a transverse evaluation value and/or a longitudinal evaluation value of the first position according to the transverse evaluation value and/or the longitudinal evaluation value of the preset position in the first node and the transverse evaluation value and/or the longitudinal evaluation value between the preset position in the first node and the first position;

and calculating the transverse evaluation value and/or the longitudinal evaluation value of the second position according to the transverse evaluation value and/or the longitudinal evaluation value of the preset position in the second node and the transverse evaluation value and/or the longitudinal evaluation value between the preset position in the second node and the second position.

That is, the lateral evaluation value and/or the longitudinal evaluation value of the first position and the second position are calculated based on the lateral evaluation value and/or the longitudinal evaluation value of the reference position obtained in step 202, and further, based on the lateral evaluation value and/or the longitudinal evaluation value of the specific reference position.

204. And determining a target path according to the evaluation value of the first position and the evaluation value of the second position so as to guide the vehicle to travel according to the target path.

Wherein the target path is a path where a position having an optimal evaluation value is located, from among the evaluation value of the first position and the evaluation value of the second position.

For example, the optimum may be the highest evaluation value, the lowest evaluation value, or exceeding a preset threshold, not greater than the preset threshold, and the like, which is not specifically limited in the present solution.

And if the longitudinal evaluation value of the first position is superior to that of the second position, and the transverse evaluation value of the first position is also superior to that of the second position, the target path is the path where the first position is located currently.

The target path may be a path where the second position is located if the longitudinal evaluation value of the second position is superior to the longitudinal evaluation value of the first position and the lateral evaluation value of the first position is superior to the lateral evaluation value of the second position.

That is, the target path is determined from the path of the first location and the path of the second location.

Alternatively, the target path may also be determined based on other means, such as: and arbitrating with other lane change intents such as a risk lane change intention generated by real-time dynamic environment perception and a passing lane change intention generated by a road real-time traffic jam state on the basis of a navigation lane change intention generated by a path where the position with the optimal evaluation value is located in the evaluation values of the first position and the second position, and determining the path with the maximum profit at the current moment as the target path. The present solution is not particularly limited to this.

And if the lane where the first position is located does not have the lane-changeable alternative lane, the target path is the path where the first position is located.

According to the embodiment of the application, an evaluation function of K paths is constructed according to the lane topological relation and the lane scene included by the K paths, a transverse evaluation value and/or a longitudinal evaluation value of each position in the K paths are/is calculated according to the evaluation function of the K paths, the evaluation value of the first position is obtained according to the first position of a vehicle, the evaluation value of the second position of the first position corresponding to a lane-changing alternative lane is obtained according to the first position, and then a target driving path is determined according to the evaluation value of the first position and the evaluation value of the second position so as to guide the vehicle to drive according to the target driving path. According to the scheme, factors influencing the transverse decision result of the vehicle at any position in any one of K paths from the departure point to the destination point and/or factors influencing the longitudinal decision result of the vehicle at any position in any one path are evaluated, so that the forward-looking range of the vehicle is expanded to the global view during real-time navigation, a complete selection space and continuously comparable lane evaluation values are provided, the vehicle is guided to carry out multi-source lane change intention arbitration, a reasonable path is selected for driving, and more intelligent automatic driving decision compared with a human driver is realized.

The following specifically describes examples of the present application. Referring to fig. 8, a schematic diagram of a method for guiding a vehicle to run according to an embodiment of the present application is shown. The method may include steps 801-807 as follows:

801. acquiring K paths from the departure point to the destination point of the vehicle according to the path planning result;

the K paths are lane-level path planning results from the departure point to the destination point.

802. Performing scene judgment and classification on all lanes contained in the K paths, and constructing shadow lane topology for a special scene to obtain a lane-level planning topology representation (tree digraph);

the method comprises the steps of classifying scenes of all lanes contained in lane-level path planning results, constructing shadow lane topology for special scenes such as lane splitting, merging, intersections and special lanes, and representing the topology structure of the lane-level planning results by adopting a tree-shaped directed graph.

Specifically, according to information such as a lane topological relation and attributes, all lanes contained in a lane-level path planning result are subjected to scene classification and are divided into parallel lanes, lane separation/merging, a to-be-turned area, an intersection, a special lane and the like.

The influence on the vehicle behavior decision is only related to a lane scene, and a scene which is not related to the real-time driving state of the vehicle is defined as a static scene, otherwise, the scene is a dynamic scene.

For the description of the lane scene, reference may be made to the foregoing embodiments, and details are not repeated herein.

According to the mutual overlapping relation of the paths, forward and backward topological extension is carried out on the paths of the split and combined scenes of the lane, a shadow lane is constructed, and the lane-level path planning result is updated.

The shadow lane is understood to mean that a plurality of paths pass through the same lane segment in a certain area, namely, comprise a segment of lane which is identical, the lane properties, the topology and the like of the paths in the area range are identical, therefore, the paths are called shadow lanes, and the area range of the same lane segment is called shadow area.

As shown in fig. 9, the shadow areas 1 of the path 3 and the path 4, and the shadow areas 2 of the path 3 and the path 5. The size of the forward and backward topological extension areas can be set arbitrarily, and the scheme is not particularly limited in this respect.

According to the scheme, after the paths in the scene of lane splitting and merging are topologically extended forwards or backwards to contain the shadow areas, the method and the device can be helpful for expressing newly added paths generated by lane splitting and merging and reducing the topological relation between the paths and the original paths, so that the description of the topological relation is more concise; the data volume generated in the real-time navigation guiding process is less, and the communication bandwidth is saved; meanwhile, the vehicle can be guided to change lanes between the original path and the newly added path and between the paths with less number of lanes more conveniently without introducing redundant logic and lane changing cost additionally.

Based on the above scene classification and shadow lane processing, the lane segments with the road topology structure distributed uniformly, and the lane segments formed by dividing the road topology structure change position, the lane topology structure change position, and the intersection are used as lane units to obtain nodes of each path, and the nodes are connected by directed edges, so as to construct a tree-like directed graph of the lane-level planning result, which can be referred to as the above fig. 4 b.

803. Determining a target reachability evaluation function, a static non-main road penalty function and a dynamic occupation penalty function based on lane topology, establishing a longitudinal evaluation function for each path, calculating a longitudinal evaluation value of each node in each path, and constructing a longitudinal global view for representing the urgency of lane change requirements;

and calculating the remaining travelable distance from the current node (position) to the route end point based on the lane topology, and establishing a longitudinal evaluation function of the target accessibility for each route. The shorter the remaining travelable distance of the route, the more urgent the lane change requirement correspondingly for lane change from the current route, and the higher priority of quick response is required in the lane change arbitration decision of real-time navigation.

Optionally, if there are multiple selectable paths in the current node, the path with the highest target reachability evaluation value in the multiple paths is selected to calculate the target reachability evaluation value of the current node.

The static non-main road penalty function represents additional cost caused by road scenes such as road geometric topology and attributes, such as road distance increase caused by road bending.

The example is given for calculating the geometric cost of a static non-main road for a lane merging and separating scenario.

Firstly, smoothing and filtering the center point of the lane, and reducing the influence of high-precision map drawing errors on the calculation of the road static non-main road penalty evaluation value. In this embodiment, a sliding window filtering method is adopted, and a filtering formula is as follows:

then, calculating the average angle change of the center point of the lane after smoothing, wherein the calculation formula is as follows:

the static non-main lane penalty caused by lane splitting and merging in the road can be calculated by the following formula:

D _geometry ＝α*D*(1-cos(angle))

wherein D is _geometry Is a static non-main lane penalty evaluation value caused by lane splitting and merging; α is a penalty factor, which can be determined from empirical values; d is the lane length of the lane splitting/merging section; m and n are respectively the starting and stopping serial numbers of the lane central point in the current filtering window, and the window length is n-m; x is the number of _k ，y _k The geographical coordinate value of the k-th lane central point in the current filtering window is represented, and the geographical coordinate value after the current window position is filtered is recorded as x _i ，y _i (ii) a N is the total number of the center points of the filtered lanes, delta angle _k Is the angle variation of the center point of the kth smoothed lane, and angle is the average angle variation of the center points of the N smoothed lanes.

Compared with the lane angle variation under the same scene, the lane with larger angle variation has higher static non-main road geometric cost. For example: compared with the angle variation of the main road lane, the inclination angle of the auxiliary road lane merging into the main road is larger, and the lane center point sequence is more zigzag, so that the static non-main road geometric cost is higher for the auxiliary road lane with larger angle variation merging into the main road lane.

The dynamic occupation penalty function is used for representing additional cost caused by vehicle behaviors and dynamic scenes, such as parking beside, lane changing, temporary occupation of bus lane steering and the like. Wherein the cost of the dynamic lane can be calculated according to the speed, the position and the like of the current vehicle.

Taking the calculation of the dynamic occupation penalty of the bus lane as an example, when the vehicle starts from the current path and must borrow a distance from the bus special lane to reach the lane where the target is located, the dynamic occupation penalty cost is related to the lane borrowing length of the bus lane, and the length required by lane borrowing is related to the current driving speed V of the vehicle and the lane borrowing time T of the bus lane _bus Correlation, see the following equation:

D _bus ＝β*V*T _bus ；

wherein D is _bus Is the penalty cost of dynamic occupation generated by the vehicle borrowing the bus lane; beta is a penalty factor, which can be determined according to empirical values; vehicle current running speed V and bus lane borrowing time T _bus In relation to the traffic unblocked state in the real-time dynamic scene, the more congested the traffic, the slower the vehicle speed, and the longer the time to borrow the lane.

And based on the target reachability evaluation function, the static non-main-track penalty function and the dynamic occupation penalty function of each path, recursively calculating the longitudinal evaluation values of all paths passing through each node, and determining the longitudinal evaluation value of each node.

The global longitudinal evaluation value of the path can be obtained by subtracting a static non-main channel penalty value and a dynamic occupation penalty value from the target reachability evaluation value. Of course, other calculation methods may also be adopted, for example, the global longitudinal evaluation value is calculated based on a preset weight value, and the present scheme is not particularly limited to this.

Preferably, according to the functional design of the automatic driving system, other evaluation functions can be expanded to calculate the global longitudinal evaluation value, for example, the functions of lane change protection and parking at the side are added, so that the lane change protection distance D is introduced _lc Parking distance cost D _pullover And the present embodiment is not particularly limited thereto.

The lower the global longitudinal evaluation value of the route is, the more urgent the lane change requirement for lane change from the current route is, the fast response with higher priority needs to be obtained in the lane change arbitration decision of real-time navigation, and the vehicle should be guided to change lanes from the route with the lower global longitudinal evaluation value to the route with the higher global longitudinal evaluation value.

In this embodiment, the functions are only used as examples for description, and may be any other functions, which is not specifically limited in this embodiment.

804. Selecting a reference lane based on lane topology, calculating the lane change times of adjacent lanes and a continuous lane of the reference lane based on the reference lane, further obtaining the minimum lane change time of each path reaching a destination point, establishing a transverse evaluation function for each path, calculating the transverse evaluation value of each node in each path, and constructing a transverse global view for describing the information of the global transverse motion trend;

firstly, determining the lane where the destination point in the tree-shaped directed graph is located as a reference lane, further calculating the minimum lane change times of the reference lane, deducing the minimum lane change times of adjacent lanes and continuous lanes, and so on, and finally obtaining the minimum lane change times of each lane reaching the destination point.

Specifically, according to the road topology relationship, if the current lane is connected to the subsequent lane, as shown in fig. 6a, the lane change number of the current lane (e.g., the lane where the node C1 is located) inherits the lane change number of the subsequent lane (e.g., the lane where the node S1 is located); if the lane change is a special scene such as road separation, as shown in fig. 6b, the current lane (e.g., the lane where the node C1 is located) inherits the minimum lane change times in the subsequent lanes (e.g., the lane where the node S1 is located and the lane where the node S2 is located);

if the current scene is an intersection, selecting at least one lane as a reference lane according to intersection topology selection strategies such as straightness, left alignment, right alignment and the like, and inheriting the minimum lane changing times of the corresponding lane behind the intersection;

and if the current lane is not connected with the subsequent lane, deducing the lane change times of the adjacent lane from the lane with the determined transverse lane change times according to the left-right adjacent relation of the lane and the lane change times.

The above is only an example, and it may also be another way to determine the minimum number of lane changes, and this is not specifically limited in this embodiment.

Other functions can be used to construct the horizontal global view, such as the horizontal offset of a single lane change, the lane change direction change of continuous lane change behavior, and the like. The present embodiment is not particularly limited to this.

805. Calculating to obtain a longitudinal evaluation value and a transverse evaluation value of each position according to the longitudinal evaluation value and the transverse evaluation value of each node in each path;

the vertical evaluation value and the horizontal evaluation value of each local position are calculated based on the adjacent nodes. For example, the node a is adjacent to the position a, and the longitudinal evaluation value and the lateral evaluation value of the position a can be obtained based on the longitudinal evaluation value and the lateral evaluation value between the node a and the position a, and the longitudinal evaluation value and the lateral evaluation value of the node a.

When the longitudinal evaluation value between the node a and the position a is, for example, a target reachability evaluation function, the longitudinal evaluation value between the node a and the position a is determined by a distance therebetween.

The above is only an example, and other methods may also be used for calculation, and this is not specifically limited in this embodiment.

806. When the vehicle is at a first position, determining that the first position corresponds to a second position in the lane-changing alternative lane, and determining a longitudinal evaluation value and a transverse evaluation value of the first position and a longitudinal evaluation value and a transverse evaluation value of the second position according to the longitudinal evaluation value and the transverse evaluation value of each position in the K paths;

based on the longitudinal evaluation value and the lateral evaluation value of each position obtained as described above, the longitudinal evaluation value and the lateral evaluation value of the first position, and the longitudinal evaluation value and the lateral evaluation value of the second position can be determined by position matching. By adopting the method, continuously comparable lane evaluation values can be provided for real-time decision making of the vehicle, the calculation efficiency of the longitudinal evaluation value and the transverse evaluation value of the path is improved, and the consumption of the real-time navigation process on calculation resources is reduced.

In the real-time navigation process, real-time lane matching is carried out according to the current position and the speed information of the vehicle, and the current lane of the vehicle and the path to which the optional lane-changing alternative lane belongs are obtained from all lane-level planning paths and are used as alternative paths for real-time decision-making of the vehicle. By adopting the method, a complete path selection space can be provided for the real-time decision of the vehicle.

In the embodiment of the present application, after the longitudinal evaluation value and the lateral evaluation value of each position are calculated based on the longitudinal evaluation value and the lateral evaluation value of each node, the longitudinal evaluation value and the lateral evaluation value of the first position are determined therefrom.

After step 804, when the vehicle is at a first position, it is determined that the first position corresponds to a second position in the lane-change candidate, and the longitudinal evaluation value and the lateral evaluation value of the first position, and the longitudinal evaluation value and the lateral evaluation value of the second position are determined according to the longitudinal evaluation value and the lateral evaluation value of each node in the K paths.

Specifically, the longitudinal evaluation value and the lateral evaluation value of the first position are calculated by determining a same-lane node adjacent to the first position and further based on the longitudinal evaluation value and the lateral evaluation value of the node. Likewise, the longitudinal evaluation value and the lateral evaluation value of the second position are calculated based on the same-lane node adjacent to the second position.

That is to say, in this alternative, the process of determining the longitudinal evaluation value and the lateral evaluation value of the first position and the second position according to the longitudinal evaluation value and the lateral evaluation value of each node in the K paths respectively is calculated according to the current first position in each iteration cycle of the real-time navigation, and is not obtained by position matching according to the current first position only in each iteration cycle of the real-time navigation after the completion of the pre-calculation.

807. And determining a target path according to the longitudinal evaluation value and the transverse evaluation value of the first position and the longitudinal evaluation value and the transverse evaluation value of the second position so as to guide the vehicle to travel according to the target path.

In this embodiment, both the longitudinal evaluation value and the transverse evaluation value are taken as consideration criteria to determine the target path, but only the longitudinal evaluation value or only the transverse evaluation value may be considered, which is not specifically limited in this embodiment.

The scheme provides a longitudinal evaluation value and a transverse evaluation value of the current lane and all alternative paths in real time, guides the vehicle to arbitrate according to navigation lane-changing intention generated by lane-changing emergency degree and transverse movement trend of the current lane, risk lane-changing intention generated by real-time dynamic environment perception, overtaking lane-changing intention generated by real-time traffic jam state of the road and other multi-source lane-changing intentions, selects a proper target lane and lane-changing opportunity, and realizes more intelligent automatic driving decision compared with a human driver.

Example one

Fig. 10 is a schematic view of a scenario for guiding a vehicle to run according to an embodiment of the present application, where an autonomous vehicle needs to turn right at an intersection with a complete view to ensure end-to-end accessibility through lane change guidance. The scene includes

roads

1,2, 3, 4, and 5 of four lanes, and a road 6 after a right turn at an intersection, where a lane merge occurs once in the road 2 and a lane split occurs once in the road 4. Further, the lane 3 of the road 1 to the lane 3 of the road 4 in this scene are bus-dedicated lanes. In this scenario, the autonomous vehicle needs to complete an autonomous driving task from lane 4 of road 1 to lane 3 of road 6.

The lane-level path planning module can provide high-precision maps, road-level and lane-level global path planning results.

The road-level global planned path from the departure point (e.g., point a in fig. 10) to the destination point (e.g., point B in fig. 10) is: road 1 → road 2 → road 3 → road 4 → road 5 → road 6, which may be denoted as R1R2R3R4R5R6;

the lane-level global path planning result is as follows:

route 1: the lane 1 of road 1, road 2, road 3, road 4 is vertically connected and is recorded as: R1L1 → R2L1 → R3L1 → R4L1; the R1L1 represents the lane 1 on the road 1, correspondingly, R2L1 represents the lane 1 on the road 2, R3L1 represents the lane 1 on the road 3, and R4L1 represents the lane 1 on the road 4.

Route 2: the lane 2 of road 1, road 2, road 3, road 4 is vertically connected and is recorded as: R1L2 → R2L2 → R3L2 → R4L2; the R1L2 represents the lane 2 on the road 1, correspondingly, R2L2 represents the lane 2 on the road 2, R3L2 represents the lane 2 on the road 3, and R4L2 represents the lane 2 on the road 4.

Route 3: the lane 3 of road 1, road 2, road 3, road 4 is vertically connected and is recorded as: R1L3 → R2L3 → R3L3 → R4L3;

path 4: the lane 4 of road 1, road 2 is vertically connected and is marked as: R1L4 → R2L4;

path 5: the lane 4 of the road 4, the lane 4 of the road 5 and the lane 3 of the road 6 are vertically communicated, and the communication is recorded as follows: R4L4 → R5L4 → R6L3;

path 6: constituted by lane 2 of road 6, noted: R6L2;

path 7: constituted by lane 1 of road 6, noted: R6L1.

Referring to fig. 10,

lanes

1,2, 3, 4 of

roads

1,3, and

lanes

1,2, 3 of road 6 are all parallel lane scenes, and

lanes

1,2, 3 of

roads

2, 4 are also parallel lane scenes, but lane 4 of road 2 is a lane merge scene and lane 4 of road 4 is a lane split scene. Lanes 3 of road 1, road 2, road 3 and road 4 are all special lane scenes of bus special lanes. Further, the lane 4 of the road 5 passes through the lane 3 of the intersection right-turn road 6, i.e., the inter-intersection lane between the lane 4 of the road 5 and the lane 3 of the road 6 is an intersection right-turn lane scene.

Since the lane 4 and the lane 3 merge at the road 2, and the lane 3 and the path 4 merge before the shadow area, the path 4 needs to extend forward for a certain distance to form the shadow area in the lane 3 of the road 3, and then the path 4 is extended as follows: the lane 4 of the road 1 and the lane 3 of the road 2 and the lane 3 of the road 3 are longitudinally communicated, and the communication is recorded as follows: R1L4 → R2L4 → R3L3; both paths contain the shadow area R3L3, thereby associating the topological relationship of path 4 with path 3.

Similarly, the lane 3 of the road 3 forms the path 3 and the path 5 due to the lane separation, and therefore the path 5 is extended as: the lane 3 of the road 3, the lane 4 of the road 4, the lane 4 of the road 5, and the lane 3 of the road 6 are communicated, and the following are recorded: R3L3 → R4L4 → R5L4 → R6L3; the path 5 includes the shadow area R3L3 together with the path 3, thereby relating the topological relationship between the path 5 and the path 3.

Taking a lane endpoint, an intersection, a lane separation position, a lane merging position and the like as nodes, connecting the nodes through directed edges, and constructing a tree-shaped directed graph, as shown in fig. 11, wherein the arrow direction of the edges indicates the direction from a departure point to a destination point.

The node L0-4 is positioned in a departure lane, namely the lane where the departure point is positioned; the nodes L0-3, L1-3, L2-3, L3-3 and L4-3 are positioned on a bus lane; the node Lcross is a crossing node; node L6-3 is located in the target lane, i.e., the lane in which the destination point is located.

Optionally, the intersection node Lcross is equivalent to the node L5-4.

In order to construct a longitudinal global view for representing the urgency of lane change requirements, a longitudinal evaluation function is established for each path, for example, a target accessibility evaluation function and a static non-main laneAnd the penalty function and the dynamic occupation penalty function are jointly formed. The lengths of

roads

1,2, 3, 4, and 6 are all set to 50m, the length of road 5 is set to 20m, and the actual length of lane 4 of

roads

2 and 4 is set to 55m, which is slightly longer than the road length because of the presence of lane merging/separating. For example, according to the empirical value, the geometric cost penalty factor α of the static non-main road of the road can be set to 2, the vehicle speed when the vehicle approaches the bus lane is 15km/h, the occupied time is 6s, and the dynamic occupation penalty factor β of the bus lane is set to 2. Meanwhile, in the scheme, the functions of lane changing protection, parking near the edge and the like are expanded, and the lane changing protection distance cost D is obtained _lc The translation is 25m, and the cost of the parking distance close to the edge is D _pullover The conversion is 20m.

First, according to the topological relation of each path in fig. 11, the target reachability evaluation value of each node can be obtained by the following calculation:

D(L6-1)＝D(L6-2)＝D(L6-3)＝0m；

D(Lcross)＝50m；

d (L5-4) = D (L6-3) + D (L5-4-L6-3) =0+50=50m; wherein D (L5-4-L6-3) represents a target reachability evaluation value between the node L6-3 and the node L5-4;

D(L4-4)＝D(L5-4)+D(L4-4-L5-4)＝50+20＝70m；

D(L4-1)＝D(L4-2)＝D(L4-3)＝0m；

D(L3-3)＝max(D(L4-4)+D(L3-3-L4-4)，D(L4-3)+D(L3-3-L4-3))＝max(70+55，0+50)＝125m；

D(L3-1)＝D(L3-2)＝0+50＝50m；

D(L2-3)＝D(L3-3)+D(L2-3-L3-3)＝50+125＝175m；

D(L2-1)＝D(L2-2)＝50+50＝100m；

D(L1-4)＝D(L2-3)+D(L1-4-L2-3)＝175+55＝230m；

D(L1-3)＝D(L2-3)+D(L1-3-L2-3)＝175+50＝225m；

D(L1-1)＝D(L1-2)＝100+50＝150m；

D(L0-4)＝D(L1-4)+D(L0-4-L1-4)＝230+50＝280m；

D(L0-3)＝D(L1-3)+D(L0-3-L1-3)＝225+50＝275m；

D(L0-1)＝D(L0-2)＝150+50＝200m。

then, a static non-main road penalty cost value is calculated according to road geometric topology, attributes and the like, and taking calculation of the static non-main road geometric cost of the merging scene of the lane 4 of the road 2 and the separation scene of the lane 4 as an example, in this embodiment, the average angle variation of the center point of the lane after smoothing before and after lane merging/separation is calculated after filtering by using a sliding window: angle =25 ° =0.436rad; the average angle change after smoothing filtering of other lanes is close to 0 (i.e. the lane is straight) and can be ignored.

Thus, the geometric cost of lane 4 of road 2 and lane 4 of road 4: d _geometry ＝α*D(L2-4)*cos(angle)＝α*D(L4-4)*cos(angle)＝2*55*(1-cos(0.436))≈10m。

And calculating the cost of dynamically occupying the special lane according to the speed and the position of the current vehicle. In the embodiment, the cost of calculating the occupation of the bus lane is taken as an example, when the vehicle runs from the path 1 and the path 2 in fig. 10 to reach the lane L6-3 where the destination point is located, the vehicle must borrow a distance from the bus lane of the path 3 to reach the right-turn lane; the path 3 occupies a bus lane in the whole process of driving to the lane where the destination point is located; the route 4 may be changed to the route 3 and then to the lane where the route 5 reaches the destination point (the lane 3 of the access road 4), or may be changed to the route 3 and then to the lane where the destination point is located along the route 5 (the lane 4 of the road 4 is directly reached from the lane 3 of the road 3). Therefore, the dynamic occupation punishment D of the bus lane of each path _bus Comprises the following steps:

D _bus1 ＝D _bus2 ＝β*V*T _bus ＝2*15/3.6*6＝50m；

D _bus3 ＝β*D(L0-3-L3-3)＝2*150＝300m；

D _bus4 ＝min(β*D(L2-3-L3-3)，β*D(L2-3-L4-3)+β*V*T _bus )＝2*50＝100m。

based on the target reachability evaluation function, the static non-main-track penalty function and the dynamic occupation penalty function, the highest evaluation value of all paths passing through each node is calculated in a recursion mode, namely the longitudinal evaluation value of the node is as follows:

F(Lcross-L6-3)＝50m；

F(Lcross-L6-2)＝F(Lcross)-D _pullover ＝50-20＝30m；

F(Lcross-L6-1)＝F(Lcross)-2*D _pullover ＝50-2*20＝10m；

F(L5-4)＝F(Lcross-L6-3)＝50m；

F(L4-4)＝F(L5-4)+D(L4-4-L5-4)＝50+20＝70m；

F(L4-3)＝D(L4-3)-D _lc ＝-25m；

F(L4-2)＝D(L4-2)-2*D _lc ＝-50m；

F(L4-1)＝D(L4-1)-3*D _lc ＝-75m；

F(L3_3)＝F(L4-4)+D(L3-3-L4-4)-D _geometry ＝70+55-10＝115m；

F(L3-2)＝F(L4-2)+D(L3-2-L4-2)-D _bus2 ＝-50m；

F(L3-1)＝F(L4-1)+D(L3-1-L4-1)-D _bus1 ＝-75m；

F(L2-3)＝F(L3-3)+D(L2-3-L3-3)-D _bus4 ＝115+50-100＝65m；

F(L2-2)＝F(L3-2)+D(L2-2-L3-2)＝-50+50＝0m；

F(L2-1)＝F(L3-1)+D(L2-1-L3-1)＝-75+50＝-25m；

F(L1-4)＝F(L2-3)+D(L1-4-L2-3)-D _geometry ＝65+55-10＝110m；

F(L1-3)＝F(L2-3)+D(L1-3-L2-3)-β*D(L1-3-L2-3)＝65+50-2*50＝15m；

F(L1-2)＝F(L2-2)+D(L1-2-L2-2)＝0+50＝50m；

F(L1-1)＝F(L2-1)+D(L1-1-L2-1)＝-25+50＝25m；

F(L0-4)＝F(L1-4)+D(L0-4-L1-4)＝110+50＝160m；

F(L0-3)＝F(L1-3)+D(L0-3-L1-3)-β*D(L0-3-L1-3)＝15+50-2*50＝-35m；

F(L0-2)＝F(L1-2)+D(L0-2-L1-2)＝50+50＝100m；

F(L0-1)＝F(L1-1)+D(L0-1-L1-1)＝25+50＝75m。

the lower the longitudinal evaluation value of the path is, the more urgent the lane change requirement for lane change from the current path is, and the fast response with higher priority needs to be obtained in the lane change arbitration decision of real-time navigation.

On the other hand, in order to construct a transverse global view, according to lane-level planning results, the minimum number of lane changing times of each lane from the destination point to the departure point is calculated, a transverse evaluation function is established for each path, and global transverse motion trend information is provided.

The lane where the destination point is located (i.e. the lane corresponding to the node L6-3) is taken as a reference lane, and the minimum number of conversion passes N of each node is sequentially calculated from the destination point to the departure point:

N(L6-3)＝0；

since the lanes L6-1 and L6-2 can only reach the end point by lane change, it can be obtained according to the adjacent topological relation of the left and right of the lanes: n (L6-2) = N (L6-3) +1=1;

N(L6-1)＝N(L6-2)+1＝2；

after the calculation of the number of the transverse lane changes of each lane in the road 6 is completed, the previous node Lcross of the road 6 is an intersection right-turn scene, an intersection reference lane can be selected according to a right alignment principle, and the transverse lane change number of the subsequent lane of the node is directly inherited, so that the number of the transverse lane changes of the node Lcross is as follows: ncross = N (L6-3) =0;

the node L5-4 of the road 5 has a successor lane, which inherits the transverse lane change times of the successor node Lcross, so that:

N(L5-4)＝Ncross＝N(L6-3)＝0；

in the same way, the number of transverse conversion passes of the node L4-4 is as follows: n (L4-4) = N (L5-4) =0;

however, L4-1, L4-2 and L4-3 have no subsequent lane, and are only adjacent lanes of the lane L4-4, so that the following can be deduced according to the left-right adjacent topological relation of the lanes:

N(L4-3)＝N(L4-4)+1＝1；

N(L4-2)＝N(L4-3)+1＝2；

N(L4-1)＝N(L4-2)+1＝3；

wherein, the node L3-3 has two successor nodes L4-3 and L4-4 at the same time, because the horizontal global lane change number N (L4-4) of the node L4-4 is less than the horizontal global lane change number N (L4-3) of the node L4-3, the node L3-3 inherits the more optimal horizontal global lane change number from the node L4-4, and can obtain: n (L3-3) = N (L4-4) =0;

the nodes L3-2 and L3-1 both have only one subsequent lane, so that the number of transverse change lanes of the subsequent lane is directly inherited:

N(L3-2)＝N(L4-2)＝2；

N(L3-1)＝N(L4-1)＝3；

similarly, the number of transverse lane changes of each lane in the road 2 can be obtained:

N(L2-3)＝N(L3-3)＝0；

N(L2-2)＝N(L3-2)＝2；

N(L2-1)＝N(L3-1)＝3；

nodes L1-3 and L1-4 in road 1 have the same successor lanes L2-3, so:

N(L1-3)＝N(L2-3)＝0；

N(L1-4)＝min(N(L2-3)，N(L1-3)+1)＝N(L2-3)＝0；

node L1-1 and node L1-2 each have a successor lane, so:

N(L1-2)＝N(L2-2)＝2；

N(L1-1)＝N(L2-2)＝3；

nodes L0-1, L0-2, L0-3, and L0-4 each have a successor lane, so:

N(L0-1)＝N(L1-1)＝3；

N(L0-2)＝N(L1-2)＝2；

N(L0-3)＝N(L1-3)＝0；

N(L0-4)＝N(L1-4)＝0。

in this embodiment, the number of lateral lane changes of a lane is described by taking the example of calculating the number of lateral lane changes of a subsequent lane with priority. Of course, the number of lateral lane changes of the lane may also be determined based on two aspects of the adjacent lane and the subsequent lane of the lane, and this is not particularly limited in the present embodiment.

And in the real-time navigation process, carrying out real-time lane matching according to the current position and the speed information of the vehicle, and acquiring the current lane of the vehicle and the optional lane change alternative lane.

When the vehicle is at three real-time matching positions Loc1, loc2, and Loc3 as shown in fig. 12, the set of the current lane and the candidate lane and the longitudinal evaluation value and the lateral evaluation value thereof are respectively:

for the real-time matching position Loc1 (located at the middle of the lane of road 1, 25m from the start of the lane):

route 1: f (Loc 1, 1) = (F (L0-1) + F (L1-1))/2 =50m, N (Loc 1, 1) = (N (L0-1) + N (L1-1))/2=3;

route 2: f (Loc 1, 2) = (F (L0-2) + F (L1-2))/2 =75m, N (Loc 1, 2) = (N (L0-2) + N (L1-2))/2=2;

route 3: f (Loc 1, 3) = (F (L0-3) + F (L1-3))/2 = -10m, N (Loc 1, 3) = (N (L0-3) + N (L1-3))/2=0;

path 4: f (Loc 1, 4) = (F (L0-4) + F (L1-4))/2 =135m, N (Loc 1, 4) = (N (L0-4) + N (L1-4))/2=0;

for the real-time matching position Loc2 (located at the rear position in the middle of the lane of road 3, 30m from the lane start):

route 1: f (Loc 2, 1) = F (L3-1) + Δ F (L3-1) = -55m, N (Loc 2, 1) = N (L3-1) + Δ N (L3-1) =3;

route 2: f (Loc 2, 2) = F (L3-2) + Δ F (L3-2) = -20m, N (Loc 2, 2) = N (L3-2) + Δ N (L3-2) =2;

route 3: f (Loc 2, 3) = F (L3-3) + Δ F (L3-3) =95m, N (Loc 2, 3) = N (L3-3) + Δ N (L3-3) =0;

for the real-time matching position Loc3 (located at the position forward of the middle of the lane of the road 4, 40m from the lane start):

route 1: f (Loc 3, 1) = F (L4-1) + Δ F (L4-1) = -75m, N (Loc 3, 1) = N (L4-1) + Δ N (L4-1) =3;

route 2: f (Loc 3, 2) = F (L4-2) + Δ F (L4-2) = -50m, N (Loc 3, 2) = N (L4-2) + Δ N (L4-2) =2;

route 3: f (Loc 3, 3) = F (L4-3) + Δ F (L4-3) = -35m, N (Loc 3, 3) = N (L4-3) + Δ N (L4-3) =1;

path 5: f (Loc 3, 5) = F (L4-4) + Δ F (L4-4) =80m, N (Loc 3, 5) = N (L4-4) + Δ N (L4-4) =0.

The longitudinal evaluation value and the lateral evaluation value of the vehicle in the current lane and all the alternative paths, which continuously change as described above, may be provided in real time during the vehicle traveling, such as F (Loc) and N (Loc). When F (Loc) is smaller, the lower the longitudinal global evaluation value of the lane is, the higher the emergency degree of the lane change from the current lane is; the smaller N (Loc) is, the smaller the lateral cost of the lane where the lane reaches the destination point is, and if the lane is selected as the target lane, the greater the lateral benefit is.

Taking the set of the current lane and the alternative lane of the real-time matching position Loc1 as an example, since F (Loc 1, 1) and F (Loc 1, 3) are too small, even if N (Loc 1, 3) < N (Loc 1, 2), the target route is to guide the vehicle to continue to travel forward in lane L1-2 of route 2; but as forward travel continues along path 2 into road 3, F (Loc, 2) of path 2 decreases rapidly, while F (Loc, 3) of path 3 increases rapidly and keeps N (Loc, 3) < N (Loc, 2), the lead vehicle will switch from lane L2-2 of path 2 into lane L3-3 of path 3 and continue forward travel along path 3; when traveling to the road 4, since there are F (Loc, 5) > F (Loc, 3) and N (Loc, 5) < N (Loc, 3), the guidance vehicle is switched into the path 5 along the separation lane L4-4 of the lane L3-3 and finally reaches the destination lane L6-3 of the autonomous driving mission.

In this process, before the vehicle changes the lane from the route 2 to the route 3, if a traffic jam situation exists in front of the vehicle in the route 2, F (Loc, 2) will be caused to drop sharply, so that the difference of the longitudinal evaluation values between the route 2 and the route 3 is increased in advance, and the vehicle is guided to select an earlier lane change timing, and finally the vehicle completes the lane change earlier and changes the lane into the route 3 as soon as possible.

If the lanes L1-3, L2-3, L3-3, and L4-3 included in the route 3 in this embodiment are not bus lanes, the longitudinal evaluation value of the node L1-4 will be smaller than that of the node L1-3 due to the geometric cost (10 m) of the static non-main lane, and the number of times N (L1-4) of the lateral lane change is equal to N (L1-3), so that the vehicle is guided to change the lane from the route 4 to the route 3 and keep traveling in the route 3 as much as possible, thereby achieving the characteristic of automatically driving to avoid the non-straight non-main lane.

Therefore, the method and the device establish a longitudinal comprehensive evaluation model F (Loc) and a transverse minimum transformation number evaluation model N (Loc) under the global view for each position of each path, so that transverse and longitudinal behavior decisions and income quantification of the vehicle are continuously changed evaluation values, the vehicle is more prospective in dynamic decision, and transverse and longitudinal benefits of each behavior decision are maximized. Because the longitudinal comprehensive evaluation value and the global change number of each path are continuously changed along with different positions, the scheme can provide a complete transverse and longitudinal comparable path selection space according to the result of position matching in the driving process of the vehicle, and fully ensures the degree of freedom of dynamic decision; according to a longitudinal comprehensive evaluation value and a transverse global lane change number evaluation value which are provided in real time and continuously change the current lane and all the alternative paths, namely the urgency and the transverse movement trend of the real-time lane change requirement under the global view, the forward looking range of the vehicle is expanded to the global view, and the necessary navigation lane change requirement and the urgency thereof can be sensed in advance; and then, the vehicle is guided to arbitrate according to other lane changing intentions such as the navigation lane changing emergency degree of the current lane, the risk lane changing intention generated by real-time dynamic environment perception, the overtaking lane changing intention generated by the real-time traffic jam state of the road and the like, and finally a target lane conforming to the global motion trend and a proper lane changing time are selected to carry out the optimal path comprehensive guidance, so that the vehicle can realize more intelligent automatic driving decision than a human driver in a diversified traffic scene.

In fact, in a special scene of the bus lane borrowing occupation, the scheme can guide the vehicle to avoid the bus lane to the maximum extent, and the vehicle can occupy for a short time only when the lane borrowing turning is needed, so that the time and space occupation on the special lane is reduced; in scenes such as complex main and auxiliary road intersection, bus station parking areas and the like, the scheme can guide the vehicles to always select straight main road lanes to run; in an actual continuous turning scene, the scheme can guide the vehicle to trigger a lane change request earlier, and reserve enough time and space allowance, so that the vehicle can be ensured to smoothly complete an automatic driving task to reach a destination.

Therefore, the scheme supports most scenes of complex topology of urban roads, such as parallel lanes, lane separation/convergence, main and auxiliary roads, to-be-turned areas, intersections, emergency lanes, bus lanes and the like, and is easy to expand automatic driving functions of emergency lane changing, side-by-side parking and the like, so that lane changing requirements and urgency of each position in each lane level path in the global visual field can be generally quantified into continuously-changed transverse evaluation values and longitudinal evaluation values, proper target lanes and lane changing time are ensured to be selected, problems of lane changing failure or emergency take-over and the like in front of the intersections or at dense traffic places are avoided, automatic driving tasks can be completed even in the intersections and non-intersections scenes with large traffic flow, and the method is suitable for real high-real-time high-dynamic road scenes or commercial automatic driving vehicles.

Example two

Referring to fig. 13, a schematic view of a scenario for guiding a vehicle to run according to an embodiment of the present application is shown. The scene is that the automatic driving vehicle needs to be guided to pass through a complex multi-lane special-shaped intersection by means of lane changing with a complete view field. The scene is composed of a road 1 and a road 2 with six lanes, a special-shaped intersection and a road 3 and a road 4 with three lanes after the intersection runs straight, wherein the lane 1 and the lane 3 of the road 2 are left-turn lanes, the lane 2, the lane 4 and the lane 5 are straight lanes, and the lane 6 is a right-turn lane. In this scenario, the autonomous vehicle needs to complete the autonomous driving task from road 1 to lane 3 of road 4.

The processing related to guiding the vehicle to run according to this embodiment may refer to the description of the foregoing embodiment, and is not described herein again.

It should be noted that, in this embodiment, according to a lane-level planning result, the global lateral minimum number of lane changes of a lane where each lane reaches the destination point is calculated from the destination point to the departure point, and global lateral motion trend information is provided, especially in a complex multi-lane special-shaped intersection scene.

As shown in fig. 13, the lane L4-3 where the destination point is located is used as the reference lane, so the number of horizontal lane changes is: n (L4-3) =0; and the lanes L4-1 and L4-2 can only reach the lane L4-3 where the destination point is located through lane change, so the number of transverse lane change of the lanes L4-1 and L4-2 can be deduced according to the left and right adjacent topological relation of the lanes:

N(L4-2)＝N(L4-3)+1＝1；

N(L4-1)＝N(L4-2)+1＝2。

each lane in the road 3 has a subsequent lane, so the transverse lane change times of each lane are as follows:

N(L3-3)＝N(L4-3)＝0；

N(L3-2)＝N(L4-2)＝1；

N(L3-1)＝N(L4-1)＝2。

after the derivation of the number of the transverse lane changes of each lane in the road 3 is completed, the front of the road 3 is a complex special-shaped intersection containing a plurality of lane topologies. Because the number and the positions of the lanes entering the intersection and the lanes leaving the intersection are not aligned one by one, the reference lane of the intersection can be selected according to the straightness principle. In the embodiment, the topology from the lane L2-4 to the lane L3-3 is selected as the intersection reference lane according to the straightness of the topology of the lanes in the intersection, and then the topology relationship between the front and the back of the intersection can be obtained according to the following topology relationship:

N(L2-4)＝N(L3-3)＝0；

besides the most straight road junction reference lane, the angle deviation of other lanes in the road junction to extend into and out of the road junction is large, and the lane changing behavior in the road junction is equivalent to the lane changing behavior in the road junction, so that the lane changing behavior cannot be used as the front-back topological relation of the lanes for calculating the transverse lane changing times. Thus, the other lanes in road 2 can be derived from the left and right adjacent topological relationship of the lanes:

N(L2-2)＝N(L2-4)+2＝2；

N(L2-5)＝N(L2-4)+1＝1；

each lane in the road 1 has a subsequent lane, so the transverse lane change times of each lane are as follows:

N(L1-2)＝N(L2-2)＝2；

N(L1-4)＝N(L2-4)＝0；

N(L1-5)＝N(L2-5)＝1。

in the embodiment, a complex topological connection relation exists in the special-shaped intersection, and the number, the position, the width and the like of each lane before and after the intersection are inconsistent. According to the scheme, the intersection reference lane is selected according to the intersection straightness principle, the global lane change times of all lanes in each path are obtained through reference lane inheritance and left-right adjacent topological relation deduction, and the calculation of the transverse global minimum lane change times in the special-shaped intersection complex topological scene is achieved. Therefore, the scheme not only simplifies the processing difficulty of the topological scene of the complex intersection, but also accords with the driving habit of a human driver, and can process various complex urban road scenes more universally.

Fig. 14 is a schematic structural diagram of a device for guiding a vehicle to run according to an embodiment of the present application. As shown in fig. 14, the apparatus includes: an obtaining module 1401, a calculating module 1402, a first determining module 1403, and a second determining module 1404, wherein:

an obtaining module 1401, configured to obtain K paths from a departure point to a destination point of a vehicle according to a path planning result, where K is a positive integer;

a calculating module 1402, configured to construct an evaluation function of the K paths according to a lane topological relation and a lane scene included in the K paths, and calculate an evaluation value of each reference position in the K paths according to the evaluation function of the K paths, where the evaluation function includes at least one of a transverse evaluation function and a longitudinal evaluation function, the transverse evaluation function is used to evaluate a factor affecting a vehicle transverse decision result at any position in any one of the K paths, and the longitudinal evaluation function is used to evaluate a factor affecting a vehicle longitudinal decision result at any position in any one of the K paths;

a first determining module 1403, configured to determine, when the vehicle is located at a first position in the K paths, an evaluation value of the first position and an evaluation value of a second position, where the first position corresponds to a position in a lane change alternative lane, according to the evaluation value of each reference position in the K paths;

a second determining module 1404, configured to determine a target route according to the evaluation value of the first location and the evaluation value of the second location, so as to guide the vehicle to travel along the target route.

Preferably, the longitudinal evaluation function is related to at least one of a target reachability evaluation function of each of the K paths and a longitudinal penalty evaluation function corresponding to each of the K paths, where the longitudinal penalty evaluation function includes at least one of: a static non-main lane penalty function, a dynamic occupation penalty function, a lane change protection penalty function and an edge parking penalty function.

The transverse evaluation function is related to at least one of the minimum lane change times of each path in the K paths to reach the destination point, the transverse offset of single lane change and the lane change direction change of continuous lane change behaviors.

Wherein the calculating module 1402 is configured to: constructing a transverse evaluation function and/or a longitudinal evaluation function of each node in the K paths, wherein the nodes are lane units which form each path in the K paths and have topological connection relations, and each lane unit comprises lane sections with uniformly distributed road topological structures and lane sections formed by dividing the positions of road topological structure changes, the positions of the road topological structure changes and intersections; and calculating the transverse evaluation value and/or the longitudinal evaluation value of each reference position in the K paths according to the transverse evaluation function and/or the longitudinal evaluation function of each node.

Preferably, the apparatus further comprises: the processing module is used for obtaining a topological connection relation between the nodes according to the nodes of the K paths and the topological relation between the lanes contained in the K paths; the calculating module 1402 is configured to: and according to the topological connection relation among the nodes, sequentially calculating the transverse evaluation value and/or the longitudinal evaluation value of each reference position in the K paths from the destination point.

When a node i is a non-path end point node, a node j is a node next to the node i, and only one node j is topologically connected with the node i, the longitudinal evaluation value of the node i is the sum of the longitudinal evaluation value between the node i and the node j and the longitudinal evaluation value of the node j; when a node i is a non-path end point node, and when M nodes in a next node of the node i are topologically connected with the node i, determining a longitudinal evaluation value of the node i according to M numbers, wherein a kth number of the M numbers is a sum of the longitudinal evaluation value of the kth node and a longitudinal evaluation value between the node i and the kth node, M is an integer not less than 2, i, j, and k are positive integers, k is not greater than M, and k is any one of the M numbers; and when the node i is a path end node, the longitudinal evaluation value of the node i is a preset value.

Wherein the calculating module 1402 is configured to: determining a node corresponding to the lane where the destination point is located as a reference node, and calculating a transverse evaluation value of the reference node according to a transverse evaluation function of the reference node; if the reference node has a splicing node, calculating a transverse evaluation value of the splicing node according to a transverse evaluation function of the splicing node of the reference node, a transverse evaluation value of the reference node and a transverse evaluation value between the reference node and the splicing node; if the reference node has an adjacent node corresponding to the lane-changing alternative lane, calculating the transverse evaluation value of the adjacent node according to the transverse evaluation function of the adjacent node, the transverse evaluation value of the reference node and the transverse evaluation value between the reference node and the adjacent node; if the adjacent node of the reference node has no splicing node, determining a transverse evaluation value of the adjacent node of the splicing node of the reference node according to a transverse evaluation function of the adjacent node of the splicing node of the reference node, a transverse evaluation value of the splicing node of the reference node and a transverse evaluation value between the splicing node of the reference node and the adjacent node of the splicing node of the reference node; if the adjacent node of the reference node has a continuous node, calculating a first transverse evaluation value of the continuous node of the adjacent node of the reference node according to the transverse evaluation function of the continuous node of the adjacent node of the reference node, the transverse evaluation value of the adjacent node of the reference node and the transverse evaluation value between the adjacent node of the reference node and the continuous node of the adjacent node of the reference node, calculating a second transverse evaluation value of the continuous node of the adjacent node of the reference node according to the transverse evaluation function of the adjacent node of the continuous node of the reference node, the transverse evaluation value of the continuous node of the reference node and the transverse evaluation value between the continuous node of the reference node and the adjacent node of the continuous node of the reference node, determining the transverse evaluation value of the continuous node of the adjacent node of the reference node according to the first transverse evaluation value and the second transverse evaluation value, and so on to obtain the transverse evaluation value of each node in the K paths.

Wherein the reference position is any one position in the K paths; or the reference position is a preset position in nodes of the K paths, the nodes are lane units which form each path in the K paths and have topological connection relations, and each lane unit comprises lane sections with uniformly distributed road topological structures and lane sections formed by dividing positions of road topological structure changes, positions of the road topological structure changes and intersections.

When the reference position is a preset position in the nodes of the K paths, the first determining module 1403 is configured to: determining a first node adjacent to the same lane corresponding to the first position and a second node adjacent to the same lane corresponding to the second position; calculating a transverse evaluation value and/or a longitudinal evaluation value of the first position according to the transverse evaluation value and/or the longitudinal evaluation value of the preset position in the first node and the transverse evaluation value and/or the longitudinal evaluation value between the preset position in the first node and the first position; and calculating the transverse evaluation value and/or the longitudinal evaluation value of the second position according to the transverse evaluation value and/or the longitudinal evaluation value of the preset position in the second node and the transverse evaluation value and/or the longitudinal evaluation value between the preset position in the second node and the second position.

The static non-main lane penalty function is obtained by calculating according to the average angle variation of the lane and the length of the lane splitting/merging section, wherein the average angle variation of the lane is obtained by performing smooth filtering processing on the center point of the lane and according to the angle variation of the processed center point of the lane.

The dynamic occupancy penalty function is determined based on the speed of the vehicle, the location of the vehicle, the special lane borrowing time, and the traffic clear status.

The lane change protection penalty function is determined according to the speed of the vehicle, the time required by lane change and the number of attempts after the lane change failure is preset.

And the limit parking penalty function is determined according to the preset distance required by limit parking.

It should be noted that the specific implementation of each module shown in fig. 14 may refer to the description of the relevant steps of the method in the foregoing embodiment, and is not described herein again.

In the present embodiment, the means for guiding the vehicle to travel is presented in the form of a module. A "module" herein may refer to an application-specific integrated circuit (ASIC), a processor and memory that execute one or more software or firmware programs, an integrated logic circuit, and/or other devices that may provide the described functionality.

Further, the above acquisition module 1401, calculation module 1402, first determination module 1403, and second determination module 1404 may be implemented by the processor 1502 of the apparatus for guiding the travel of a vehicle shown in fig. 15.

Fig. 15 is a schematic hardware structure diagram of an apparatus for guiding vehicle driving according to an embodiment of the present application. The apparatus 1500 for guiding vehicle travel shown in fig. 15 (the apparatus 1500 may specifically be a computer device) includes a memory 1501, a processor 1502, a communication interface 1503, and a bus 1504. The memory 1501, the processor 1502, and the communication interface 1503 are communicatively connected to each other via a bus 1504.

The Memory 1501 may be a Read Only Memory (ROM), a static Memory device, a dynamic Memory device, or a Random Access Memory (RAM).

The memory 1501 may store a program, and the processor 1502 and the communication interface 1503 are used to execute the steps of the method of guiding the vehicle to travel according to the embodiment of the present application when the program stored in the memory 1501 is executed by the processor 1502.

The processor 1502 may be a general-purpose Central Processing Unit (CPU), a microprocessor, an Application Specific Integrated Circuit (ASIC), a Graphics Processing Unit (GPU) or one or more Integrated circuits, and is configured to execute related programs to implement functions required to be executed by units in the device for guiding vehicle driving according to the embodiment of the present Application, or to execute the method for guiding vehicle driving according to the embodiment of the present Application.

The processor 1502 may also be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the method for guiding vehicle driving of the present application can be implemented by instructions in the form of hardware integrated logic circuits or software in the processor 1502. The processor 1502 may also be a general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic device, discrete hardware component. The various methods, steps, and logic blocks disclosed in the embodiments of the present application may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or 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 may be located in ram, flash memory, rom, prom, or eprom, registers, etc. storage media as is well known in the art. The storage medium is located in the memory 1501, and the processor 1502 reads information in the memory 1501, and in combination with hardware thereof, performs functions required to be performed by units included in the device for guiding vehicle travel according to the embodiment of the present application, or performs the method for guiding vehicle travel according to the embodiment of the method of the present application.

The communication interface 1503 enables communication between the apparatus 1500 and other devices or communication networks using transceiver means such as, but not limited to, transceivers. For example, data may be acquired via communications interface 1503.

The bus 1504 may include pathways to transfer information between components of the device 1500 (e.g., memory 1501, processor 1502, communication interface 1503).

It should be noted that although the apparatus 1500 shown in fig. 15 shows only memories, processors, and communication interfaces, in a particular implementation, those skilled in the art will appreciate that the apparatus 1500 also includes other components necessary to achieve proper operation. Also, those skilled in the art will appreciate that the apparatus 1500 may also include hardware components for performing other additional functions, according to particular needs. Furthermore, those skilled in the art will appreciate that apparatus 1500 may also include only those components necessary to implement embodiments of the present application, and need not include all of the components shown in FIG. 15.

The embodiment of the application provides a chip system, which is applied to electronic equipment; the chip system comprises one or more interface circuits, and one or more processors; the interface circuit and the processor are interconnected through a line; the interface circuit is to receive a signal from a memory of the electronic device and to send the signal to the processor, the signal comprising computer instructions stored in the memory; when the processor executes the computer instructions, the electronic device performs the above method.

The embodiment of the application provides an intelligent driving vehicle which comprises a traveling system, a sensing system, a control system and a computer system, wherein the computer system is used for executing the method.

Embodiments of the present application also provide a computer-readable storage medium having stored therein instructions, which when executed on a computer or processor, cause the computer or processor to perform one or more steps of any one of the methods described above.

The embodiment of the application also provides a computer program product containing instructions. The computer program product, when run on a computer or processor, causes the computer or processor to perform one or more steps of any of the methods described above.

It can be clearly understood by those skilled in the art that, for convenience and simplicity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the specific descriptions of the corresponding step processes in the foregoing method embodiments, and are not described herein again.

It should be understood that in the description of the present application, unless otherwise indicated, "/" indicates a relationship where the objects associated before and after are an "or", e.g., a/B may indicate a or B; wherein A and B can be singular or plural. Also, in the description of the present application, "a plurality" means two or more than two unless otherwise specified. "at least one of the following" or similar expressions refer to any combination of these items, including any combination of the singular or plural items. For example, at least one (one) of a, b, or c, may represent: a, b, c, a-b, a-c, b-c, or a-b-c, wherein a, b, c may be single or multiple. In addition, in order to facilitate clear description of technical solutions of the embodiments of the present application, in the embodiments of the present application, terms such as "first" and "second" are used to distinguish the same items or similar items having substantially the same functions and actions. Those skilled in the art will appreciate that the terms "first," "second," etc. do not denote any order or quantity, nor do the terms "first," "second," etc. denote any order or importance. Also, in the embodiments of the present application, the words "exemplary" or "such as" are used herein to mean serving as an example, instance, or illustration. Any embodiment or design described herein as "exemplary" or "e.g.," is not necessarily to be construed as preferred or advantageous over other embodiments or designs. Rather, use of the word "exemplary" or "such as" is intended to present relevant concepts in a concrete fashion for ease of understanding.

In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other ways. For example, the division of the unit is only one logical function division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. The shown or discussed mutual coupling, direct coupling or communication connection may be an indirect coupling or communication connection of devices or units through some interfaces, and may be in an electrical, mechanical or other form.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one position, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In the above embodiments, the implementation may be wholly or partially realized by software, hardware, firmware, or any combination thereof. When implemented in software, 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. The procedures or functions according to the embodiments of the present application are wholly or partially generated when the computer program instructions are loaded and executed on a computer. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored on or transmitted over a computer-readable storage medium. The computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire (e.g., coaxial cable, fiber optic, digital Subscriber Line (DSL)), or wirelessly (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device, such as a server, a data center, etc., that includes one or more of the available media. The usable medium may be a read-only memory (ROM), or a Random Access Memory (RAM), or a magnetic medium, such as a floppy disk, a hard disk, a magnetic tape, a magnetic disk, or an optical medium, such as a Digital Versatile Disk (DVD), or a semiconductor medium, such as a Solid State Disk (SSD).

The above description is only a specific implementation of the embodiments of the present application, but the scope of the embodiments of the present application is not limited thereto, and any changes or substitutions within the technical scope disclosed in the embodiments of the present application should be covered by the scope of the embodiments of the present application. Therefore, the protection scope of the embodiments of the present application shall be subject to the protection scope of the claims.

Claims

1. A method of guiding a vehicle for travel, comprising:

acquiring K paths from a departure point to a destination point of the vehicle according to a path planning result, wherein K is a positive integer;

according to the lane topological relation and the lane scene included by the K paths, constructing evaluation functions of the K paths, and calculating an evaluation value of each reference position in the K paths according to the evaluation functions of the K paths, wherein the evaluation functions include at least one of a transverse evaluation function and a longitudinal evaluation function, the transverse evaluation function is used for evaluating a factor influencing a vehicle transverse decision result at any position in any path of the K paths, and the transverse evaluation function is related to at least one of the minimum lane change times of each path reaching a destination point, the transverse offset of a single lane change and the lane change direction change of continuous lane change behaviors; the longitudinal evaluation function is used for evaluating factors influencing a vehicle longitudinal decision result at any position in any one of the K paths, and the longitudinal evaluation function is related to at least one of a target reachability evaluation function of each of the K paths and a longitudinal penalty evaluation function corresponding to each of the K paths;

when the vehicle is located at a first position in the K paths and K is an integer not less than 2, determining an evaluation value of the first position and an evaluation value of a second position according to the evaluation value of each reference position in the K paths, wherein the second position is a position where the first position corresponds to a lane-changing alternative lane;

determining a target path according to the evaluation value of the first position and the evaluation value of the second position so as to guide the vehicle to travel according to the target path;

when K is 1, determining the evaluation value of the first position according to the evaluation value of each reference position in the K paths;

and determining a target path according to the evaluation value of the first position so as to guide the vehicle to travel according to the target path.

2. The method of claim 1, wherein the longitudinal penalty evaluation function comprises at least one of: a static non-main lane penalty function, a dynamic occupation penalty function, a lane change protection penalty function and an edge parking penalty function.

3. The method according to claim 1 or 2, wherein the constructing the evaluation functions of the K paths and calculating the evaluation value of each reference position in the K paths according to the evaluation functions of the K paths comprises:

4. The method of claim 3, further comprising:

obtaining a topological connection relation between nodes according to the nodes of the K paths and the topological relation of lanes contained in the K paths;

according to the topological connection relation among the nodes, sequentially calculating the transverse evaluation value and/or the longitudinal evaluation value of each node in the K paths from the destination point;

5. The method according to claim 4, wherein when node i is a non-path destination node, node j is a node next to node i, and only one node j is topologically connected to node i, the longitudinal evaluation value of node i is the sum of the longitudinal evaluation value of node i and node j and the longitudinal evaluation value of node j;

when a node i is a non-path end point node, and when M nodes in a next node of the node i are topologically connected with the node i, determining a longitudinal evaluation value of the node i according to M numbers, wherein a kth number of the M numbers is a sum of the longitudinal evaluation value of the kth node and a longitudinal evaluation value between the node i and the kth node, M is an integer not less than 2, i, j, and k are positive integers, k is not greater than M, and k is any one of the M numbers;

and when the node i is a path end node, the longitudinal evaluation value of the node i is a preset value.

6. The method according to claim 4 or 5, characterized in that the method further comprises:

determining a node corresponding to a lane where the destination point is located as a reference node, and calculating a transverse evaluation value of the reference node according to a transverse evaluation function of the reference node;

if the adjacent node of the reference node has a continuous node, calculating a first transverse evaluation value of the continuous node of the adjacent node of the reference node according to the transverse evaluation function of the continuous node of the adjacent node of the reference node, the transverse evaluation value of the adjacent node of the reference node and the transverse evaluation value between the adjacent node of the reference node and the continuous node of the adjacent node of the reference node, calculating a second transverse evaluation value of the continuous node of the adjacent node of the reference node according to the transverse evaluation function of the adjacent node of the continuous node of the reference node, the transverse evaluation value of the continuous node of the reference node and the transverse evaluation value between the continuous node of the reference node and the adjacent node of the continuous node of the reference node, determining the transverse evaluation value of the continuous node of the adjacent node of the reference node according to the first transverse evaluation value and the second transverse evaluation value, and so on to obtain the transverse evaluation value of each node in the K paths.

7. The method of claim 1, wherein the reference position is any one of the K paths;

or the reference position is a preset position in nodes of the K paths, the nodes are lane units which form each path in the K paths and have topological connection relations, and each lane unit comprises lane sections with uniformly distributed road topological structures and lane sections formed by dividing positions of road topological structure changes, positions of the road topological structure changes and intersections.

8. The method according to claim 7, wherein when the reference position is a preset position in the nodes of the K paths, the determining the evaluation value of the first position and the evaluation value of the second position according to the evaluation value of each reference position in the K paths comprises:

9. The method of claim 2, wherein the static non-main lane penalty function is calculated according to a lane average angle variation obtained by performing a smoothing filtering process on a center point of a lane according to an angle variation of the processed center point of the lane and a lane length of a lane splitting/merging section of the lane.

10. The method according to claim 2 or 9, characterized in that the dynamic occupancy penalty function is determined from the speed of the vehicle, the position of the vehicle, the special lane borrowing time and the traffic clear status.

11. The method of claim 2 or 9, wherein the lane-change protection penalty function is determined based on the speed of the vehicle, the time required to change lanes, and the number of attempts after a preset lane-change failure.

12. The method according to claim 2 or 9, wherein the edgewise parking penalty function is determined according to a preset distance required for edgewise parking.

13. An apparatus for guiding a vehicle, comprising:

the acquisition module is used for acquiring K paths from the departure point to the destination point of the vehicle according to the path planning result, wherein K is a positive integer;

the calculation module is used for constructing evaluation functions of the K paths according to lane topological relations and lane scenes included by the K paths, and calculating an evaluation value of each reference position in the K paths according to the evaluation functions of the K paths, wherein the evaluation functions comprise at least one of a transverse evaluation function and a longitudinal evaluation function, the transverse evaluation function is used for evaluating factors influencing a vehicle transverse decision result at any position in any one of the K paths, and the transverse evaluation function is related to at least one of the minimum lane change times of each path in the K paths reaching a destination point, the transverse offset of a single lane change and the lane change direction change of continuous lane change behaviors; the longitudinal evaluation function is used for evaluating factors influencing a vehicle longitudinal decision result at any position in any one of the K paths, and the longitudinal evaluation function is related to at least one of a target reachability evaluation function of each of the K paths and a longitudinal penalty evaluation function corresponding to each of the K paths;

a first determination module, configured to determine, when the vehicle is located at a first position in the K paths and when K is an integer not less than 2, an evaluation value of the first position according to an evaluation value of each reference position in the K paths, and an evaluation value of a second position, where the first position corresponds to a position in a lane-changing alternative lane;

the second determination module is used for determining a target path according to the evaluation value of the first position and the evaluation value of the second position so as to guide the vehicle to run along the target path;

when K is 1, the first determining module is further configured to determine an evaluation value of the first position according to the evaluation value of each reference position in the K paths;

the second determining module is further configured to determine a target path according to the evaluation value of the first position, so as to guide the vehicle to travel along the target path.

14. The apparatus of claim 13, wherein the longitudinal penalty evaluation function comprises at least one of: a static non-main lane penalty function, a dynamic occupation penalty function, a lane change protection penalty function and an edge parking penalty function.

15. The apparatus of claim 13 or 14, wherein the computing module is configured to:

16. The apparatus of claim 15, further comprising:

the processing module is used for obtaining a topological connection relation between the nodes according to the nodes of the K paths and the topological relation between the lanes contained in the K paths;

the calculation module is configured to:

17. The apparatus according to claim 16, wherein when node i is a non-path destination node, node j is a node next to node i, and only one node j is topologically connected to node i, the longitudinal evaluation value of node i is a sum of the longitudinal evaluation values of node i and node j and the longitudinal evaluation value of node j;

when a node i is a non-path end point node and M nodes in the next node of the node i are topologically connected with the node i, determining a longitudinal evaluation value of the node i according to M values, wherein the kth value of the M values is the sum of the longitudinal evaluation value of the kth node and the longitudinal evaluation value between the node i and the kth node, M is an integer not less than 2, i, j and k are positive integers, k is not more than M, and k is any one of the M values;

18. The apparatus of claim 16 or 17, wherein the computing module is further configured to:

19. The apparatus of claim 13, wherein the reference position is any one of the K paths;

20. The apparatus of claim 19, wherein when the reference position is a preset position among nodes of the K paths, the first determining module is configured to:

21. The apparatus of claim 14, wherein the static non-main lane penalty function is calculated according to a lane average angle variation obtained by performing a smoothing filtering process on a center point of a lane according to an angle variation of the processed center point of the lane and a lane length of a lane splitting/merging section of the lane.

22. The apparatus of claim 14 or 21, wherein the dynamic occupancy penalty function is determined as a function of the speed of the vehicle, the position of the vehicle, a special lane borrowing time, and a traffic clear status.

23. The apparatus of claim 14 or 21, wherein the lane-change protection penalty function is determined according to the speed of the vehicle, the time required for lane change, and the number of attempts after a preset lane change failure.

24. The apparatus of claim 14 or 21, wherein the parking penalty function is determined according to a preset distance required for parking alongside.

25. An apparatus for guiding the travel of a vehicle, comprising a processor and a memory; wherein the memory is configured to store program code and the processor is configured to invoke the program code to perform the method of any of claims 1 to 12.

26. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program which is executed by a processor to implement the method of any one of claims 1 to 12.

27. A chip system is applied to electronic equipment; the chip system comprises one or more interface circuits, and one or more processors; the interface circuit and the processor are interconnected through a line; the interface circuit is to receive a signal from a memory of the electronic device and to send the signal to the processor, the signal comprising computer instructions stored in the memory; the electronic device performs the method of any one of claims 1-12 when the processor executes the computer instructions.

28. A smart driving vehicle comprising a travel system, a sensing system, a control system and a computer system, wherein the computer system is configured to perform the method of any one of claims 1 to 12.