A vehicle-following target decision method, vehicle and storage medium

By using global path information and obstacle data to establish a motion model in an autonomous vehicle, the problems of high sensor costs and complex algorithms are solved, and the safety of autonomous vehicles is improved without increasing costs and complexity. Comfortable decision-making method for following the car.

CN114655206BActive Publication Date: 2024-11-12CHONGQING CHANGAN AUTOMOBILE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210466400.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-04-29
Publication Date
2024-11-12
Estimated Expiration
2042-04-29

AI Technical Summary

Technical Problem

In the prior art, the vehicle-following target decision-making method has problems such as high cost of sensors and computing equipment, complex algorithms, and the stability and feasibility of vehicle driving are not guaranteed. Especially in curved scenarios, target vehicle selection is prone to missed selection and missed selection.

Method used

Using global path information combined with bicycle positioning information and status information to establish a motion model, and using obstacle data to describe the lane model through grid space without requiring additional lane line detection, filter out the following targets, and realize lane and Initial modeling of the environment, longitudinal velocity planning is performed.

Benefits of technology

Without increasing costs and complexity, the safety and ride comfort of autonomous vehicles are improved, ensuring safety and smooth following the car in low-speed driving environments such as parking.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114655206B_ABST
    Figure CN114655206B_ABST
Patent Text Reader

Abstract

This invention relates to a method for determining a following target, a vehicle, and a storage medium. The invention uses global path information as prior knowledge, combines it with the vehicle's localization and state information to establish a motion model, and incorporates obstacle perception data. Without requiring additional lane line detection, it uses a grid space to more accurately describe the lane model than a cubic fitting curve obtained from lane line detection, thus completing a preliminary modeling of the lane and the vehicle's environment. Following targets are then selected based on spatial relationships. This scheme ensures safe and stable following of the vehicle when the preceding vehicle cuts in or out of its lane in low-speed driving environments such as parking. This method, while considering cost, is unaffected by scene complexity and real-time requirements, thereby effectively improving the safety and ride comfort of autonomous vehicles.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of target decision making, and in particular relates to a vehicle following target decision making method, a vehicle and a storage medium. Background Art

[0002] Following target decision is an important part of autonomous driving environment cognition, and can provide a relatively accurate and stable following target for longitudinal speed planning. Its main function is to complete the generation of the virtual own lane of the intelligent driving vehicle by combining the perceived measurement obstacle data and the global planning path information during the driving process of the autonomous driving vehicle, and to identify the scene of the vehicle changing lanes and further adjust the position of the virtual own lane by combining the own vehicle information. Based on the generated virtual own lane, the perceived measured obstacles are processed, the obstacle vehicles in front of the own lane are screened out, and the obstacle vehicles closest to the own vehicle are selected as the acceleration and deceleration planning targets of the longitudinal speed planning module, so as to achieve safe and smooth following in the process of autonomous driving.

[0003] There have been a lot of research and mass production solutions for forward following target decision in the field of autonomous driving. For example, CN110696828A describes a forward target selection method that mainly uses a forward target selection model and real-time perception data to determine the real-time forward target; the model is trained by an integrated learning method. In order to address the problems of missed selection and misselection of target vehicles in curved scenarios due to the different performance of different sensors, CN111469841A combines the first motion information of the intelligent driving vehicle and the target selection section to determine the target vehicle. Both methods involved in the above two patents require perception data to complete lane line target measurement, and perform forward target preliminary selection and determine the target selection section based on the measured lane line target data. In applications where there is no perception data to measure lane line targets, CN105631217A - "System and method for selecting effective targets ahead based on the vehicle's adaptive virtual lane" first initializes the generation of the vehicle's adaptive virtual lane, and then adaptively adjusts the vehicle's virtual lane based on the vehicle's motion state information and the vehicle-mounted radar perception data, and calculates the probability of the target being in the vehicle's lane based on the position of the radar-measured target in the vehicle's virtual lane, and selects effective targets. Although this method can select and release the cut-in and cut-out targets in a timely manner, it has high requirements for the accuracy of the virtual lane and the calibration of adaptive parameters. Summary of the invention

[0004] In view of the above-mentioned deficiencies in the prior art, the technical problem to be solved by the present invention is to provide a vehicle following target decision method, a vehicle and a storage medium to avoid the problems of excessively high costs of sensors and computing equipment, overly complex algorithms and inability to ensure the stability and feasibility of vehicle driving.

[0005] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0006] A vehicle following target decision method comprises the following steps:

[0007] S1: Calculate the distance between the lane and the global path based on the global path, the real-time positioning information and status information of the vehicle, and determine whether the vehicle should change lanes to complete the lane decision of the vehicle;

[0008] At the same time, according to the position of the global path, the lane space corresponding to the global path at the current position of the vehicle is calculated and generated;

[0009] S2: Determine whether the vehicle is in the lane where the global path is located. If so, enter the obstacle projection. If not, enter the adaptively adjusted lane first and then enter the obstacle projection.

[0010] S3: If the vehicle is in the lane where the global path is located, the direct obstacle is projected into the lane space where the global path is located based on the obstacle positioning data;

[0011] If the vehicle is not in the lane where the global path is located, the offset adjustment of the corresponding lane position is completed according to the decision result of the lane where the vehicle is located, and the final lane where the vehicle is located is obtained. Then, based on the obstacle positioning data, the direct obstacle is projected into the lane space where the global path is located;

[0012] S4: Filter out obstacles that are not in the lane where the final vehicle is located;

[0013] S5: For obstacles projected in the lane where the vehicle is finally located, the obstacle screening in front of the vehicle is completed according to the real-time longitudinal distance between the obstacle and the vehicle;

[0014] S6: According to the real-time longitudinal distance between the obstacle selected in step S5 and the vehicle, select the obstacle closest to the vehicle in the longitudinal direction, and perform longitudinal planning to calculate the acceleration of the vehicle to control the acceleration and deceleration of the vehicle.

[0015] To further improve the above technical solution, the step S1 of calculating the distance between the vehicle and the global path to determine whether the vehicle changes lanes includes:

[0016] If the distance is within half the width of the calibrated lane, the vehicle is determined to be in the lane where the global path is located;

[0017] If the distance exceeds the left boundary of the half-width range of the calibrated lane, it is determined that the vehicle performs left lane change adaptive processing to obtain the final lane where the vehicle is located;

[0018] If the distance exceeds the right boundary of the half-width range of the calibrated lane, it is determined that the vehicle is adaptively changing lanes to the right, and the final lane where the vehicle is located is obtained;

[0019] Complete the lane decision of the vehicle.

[0020] Furthermore, in step S1:

[0021] After receiving the real-time positioning information of the vehicle, a sampling space is generated in the coordinate system of the vehicle according to the calibrated horizontal and vertical resolutions, and the lane space corresponding to the global path at the current position of the vehicle is calculated and generated based on the position and angle information of the global path. The lane space is used for subsequent lane modeling.

[0022] Furthermore, when obstacle projection is performed in step S3, obstacle perception and positioning data are acquired in real time, obstacle expansion boundary calibration is performed for different scenarios, and the obstacle is projected into the sampling space.

[0023] The present invention also relates to a vehicle, which adopts the steps of the vehicle following target decision method as described above.

[0024] The present invention also relates to a storage medium, wherein the storage medium stores one or more programs, and when the one or more programs are executed by a processor, the steps of the vehicle following target decision method as described above are executed.

[0025] Compared with the prior art, the present invention has the following beneficial effects:

[0026] A method for following a vehicle target decision of the present invention uses global path information as a priori, combines the positioning information and state information of the vehicle to establish a motion model, and perceives obstacle data information. Without the need for additional lane line detection, the grid space is used to more accurately describe the lane model than the cubic fitting curve of the lane line detection, and the lane and the environment in which the vehicle is located are roughly modeled. The following vehicle target is further selected based on the spatial position relationship. This solution ensures the safety and stability of the vehicle when the front vehicle cuts in and out in low-speed driving environments such as parking. This method can be used without being affected by the complexity and real-time performance of the scene on the basis of cost considerations, thereby effectively improving the safety and ride comfort of the autonomous driving vehicle. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 The present invention is a flow chart of a vehicle following target decision method according to an embodiment. DETAILED DESCRIPTION

[0028] The specific implementation modes of the present invention will be further described in detail below in conjunction with the accompanying drawings.

[0029] Referring to the figure, a vehicle following target decision method according to a specific embodiment includes the following steps:

[0030] S1: Calculate the distance between the lane and the global path based on the global path, the real-time positioning information and status information of the vehicle, and determine whether the vehicle should change lanes to complete the lane decision of the vehicle;

[0031] At the same time, according to the position of the global path, the lane space corresponding to the global path at the current position of the vehicle is calculated and generated;

[0032] S2: Determine whether the vehicle is in the lane where the global path is located. If so, enter the obstacle projection. If not, enter the adaptively adjusted lane first and then enter the obstacle projection.

[0033] S3: If the vehicle is in the lane where the global path is located, the direct obstacle is projected into the lane space where the global path is located based on the obstacle positioning data;

[0034] If the vehicle is not in the lane where the global path is located, the offset adjustment of the corresponding lane position is completed according to the decision result of the lane where the vehicle is located, and the final lane where the vehicle is located is obtained. Then, based on the obstacle positioning data, the direct obstacle is projected into the lane space where the global path is located;

[0035] S4: Filter out obstacles that are not in the lane where the final vehicle is located;

[0036] S5: For obstacles projected in the lane where the vehicle is finally located, the obstacle screening in front of the vehicle is completed according to the real-time longitudinal distance between the obstacle and the vehicle;

[0037] S6: According to the real-time longitudinal distance between the obstacle selected in step S5 and the vehicle, select the obstacle closest to the vehicle in the longitudinal direction, and perform longitudinal planning to calculate the acceleration of the vehicle to control the acceleration and deceleration of the vehicle.

[0038] A method for following a vehicle in an embodiment adopts global path information as a priori, combines the positioning information and state information of the vehicle to establish a motion model, and perceives obstacle data information. Without the need for additional lane line detection, the grid space is used to more accurately describe the lane model than the cubic fitting curve of the lane line detection, and the lane and the environment in which the vehicle is located are roughly modeled. The following vehicle target is further selected based on the spatial position relationship. In low-speed driving environments such as parking, this solution ensures the safety and stability of the vehicle when the front vehicle cuts in and out. This method can be used without being affected by the complexity and real-time performance of the scene on the basis of cost considerations, thereby effectively improving the safety and ride comfort of autonomous driving vehicles.

[0039] Wherein, calculating the distance between the vehicle and the global path in step S1 to determine whether the vehicle changes lanes includes:

[0040] If the distance is within half the width of the calibrated lane, the vehicle is determined to be in the lane where the global path is located;

[0041] If the distance exceeds the left boundary of the half-width range of the calibrated lane, it is determined that the vehicle changes lanes to the left lane;

[0042] If the distance exceeds the right boundary of the half-width range of the calibrated lane, it is determined that the vehicle changes lanes to the right lane;

[0043] Complete the lane decision of the vehicle.

[0044] The calibrated lane is the lane where the global path is located.

[0045] When implemented, the step S1 further includes: obtaining the real-time positioning information, speed, steering wheel angle and yaw angular velocity of the vehicle to form a motion model of the vehicle, and calibrating the lane ID of the vehicle based on the motion model and in combination with the spatial position relationship between the real-time positioning information of the vehicle and the global path, so as to determine whether the vehicle changes lanes;

[0046] If the calibration is 0, the vehicle is in the lane where the global path is located and does not need to change lanes;

[0047] If the calibration is 1, the vehicle is in the left adjacent lane of the global path, and it is determined that the vehicle performs left lane change adaptive processing to obtain the final lane where the vehicle is located;

[0048] If the calibration is 2, the vehicle is in the right adjacent lane of the global path, and it is determined that the vehicle performs right lane change adaptive processing to obtain the final lane where the vehicle is located.

[0049] Among them, in the step S1: after receiving the real-time positioning information of the vehicle, a sampling space is generated in the coordinate system of the vehicle according to the calibrated horizontal and vertical resolutions, and the lane space corresponding to the global path at the current position of the vehicle is calculated and generated according to the position and angle information of the global path. The lane space is used for subsequent lane modeling.

[0050] During implementation, since the vehicle positioning information is obtained in real time, the grid space (i.e., the sampling space) of the lane where the global path ahead is located will be continuously generated in the coordinate system where the vehicle is located according to the calibrated horizontal and vertical resolutions to express the lane where the global path is located.

[0051] When obstacle projection is performed in step S3, obstacle perception and positioning data are acquired in real time, obstacle expansion boundary calibration is performed for different scenarios, and the obstacle is projected into the sampling space.

[0052] When implemented, different scenarios include following a vehicle, meeting another vehicle, and crossing.

[0053] During implementation, obstacle perception and positioning data is acquired in real time, environment modeling is performed, obstacles are projected into the sampling space, and then it is determined whether the obstacle is in the final lane where the vehicle is located (i.e., the lane where the vehicle is currently located), and the lane mark position where the obstacle is located is obtained:

[0054] If the flag is 1, the obstacle is in the lane where the vehicle is finally located;

[0055] If the flag is 0, the obstacle is not in the lane where the vehicle is finally located, and the target is filtered out.

[0056] For obstacles in the lane where the vehicle is located and whose flag position is 1, the relative front and rear position relationship between the obstacle and the vehicle is determined based on the real-time position information of the obstacle in the vehicle coordinate system, and the corresponding flag position is obtained:

[0057] If the flag is 1, the obstacle is in front of the vehicle;

[0058] If the flag is 0, the obstacle is behind the vehicle and the target is filtered out.

[0059] Furthermore, for the obstacle in front of the vehicle whose flag is 1 and is in the final lane of the vehicle, based on the real-time longitudinal distance information in the vehicle coordinate system, the obstacle closest to the vehicle in longitudinal distance is selected as the final following target of the vehicle.

[0060] The present invention also provides a vehicle, which adopts the steps of the above-mentioned vehicle following target decision method.

[0061] The present invention also provides a storage medium, which stores one or more programs. When the one or more programs are executed by a processor, the steps of the vehicle following target decision method as described above are executed.

[0062] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0063] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A vehicle following target decision method, characterized by: The following steps are involved: S1: Calculate the distance between the lane and the global path based on the global path, the real-time positioning information and status information of the vehicle, and determine whether the vehicle should change lanes to complete the lane decision of the vehicle; At the same time, according to the position of the global path, the lane space corresponding to the global path at the current position of the vehicle is calculated and generated; S2: Determine whether the vehicle is in the lane where the global path is located. If it is in the lane where the global path is located, enter the obstacle projection; If the lane is not the one where the global path is located, the system will first enter the adaptively adjusted lane and then enter the obstacle projection lane. S3: If the vehicle is in the lane where the global path is located, the direct obstacle is projected into the lane space where the global path is located based on the obstacle positioning data; If the vehicle is not in the lane where the global path is located, the offset adjustment of the corresponding lane position is completed according to the decision result of the lane where the vehicle is located, and the final lane where the vehicle is located is obtained. Then, based on the obstacle positioning data, the direct obstacle is projected into the lane space where the global path is located; S4: Filter out obstacles that are not in the lane where the final vehicle is located; S5: For obstacles projected in the lane where the vehicle is finally located, the obstacle screening in front of the vehicle is completed according to the real-time longitudinal distance between the obstacle and the vehicle; S6: According to the real-time longitudinal distance between the obstacles selected in step S5 and the vehicle, select the obstacle closest to the vehicle in the longitudinal direction, and perform longitudinal planning to calculate the acceleration of the vehicle to control the acceleration and deceleration of the vehicle; The step S1 in which the distance between the vehicle and the global path is calculated to determine whether the vehicle changes lanes includes: If the distance is within half the width of the calibrated lane, the vehicle is determined to be in the lane where the global path is located; If the distance exceeds the left boundary of the half-width range of the calibrated lane, it is determined that the vehicle performs left lane change adaptive processing to obtain the final lane where the vehicle is located; If the distance exceeds the right boundary of the half-width range of the calibrated lane, it is determined that the vehicle is adaptively changing lanes to the right, and the final lane where the vehicle is located is obtained; Complete the lane decision of the vehicle; The step S1 further includes: obtaining the real-time positioning information, speed, steering wheel angle and yaw angular velocity of the vehicle to form a motion model of the vehicle, and calibrating the lane ID of the vehicle based on the motion model and in combination with the spatial position relationship between the real-time positioning information of the vehicle and the global path, so as to determine whether the vehicle changes lanes; If the calibration is 0, the vehicle is in the lane where the global path is located and does not need to change lanes; If the calibration is 1, the vehicle is in the left adjacent lane of the global path, and it is determined that the vehicle performs left lane change adaptive processing to obtain the final lane where the vehicle is located; If the calibration is 2, the vehicle is in the right adjacent lane of the global path, and it is determined that the vehicle performs right lane change adaptive processing to obtain the final lane where the vehicle is located.

2. A vehicle following target decision method according to claim 1, characterized in that: In step S1: After receiving the real-time positioning information of the vehicle, a sampling space is generated in the coordinate system of the vehicle according to the calibrated horizontal and vertical resolutions, and the lane space corresponding to the global path at the current position of the vehicle is calculated and generated based on the position and angle information of the global path. The lane space is used for subsequent lane modeling.

3. A vehicle following target decision method according to claim 2, characterized in that: When obstacle projection is performed in step S3, obstacle perception and positioning data are acquired in real time, obstacle expansion boundary calibration is performed for different scenarios, and the obstacle is projected into the sampling space.

4. A vehicle, characterized in that: The steps of a vehicle following target decision method as described in any one of claims 1 to 3 are adopted.

5. A storage medium, characterized in that: The storage medium stores one or more programs, and when the one or more programs are executed by the processor, the steps of a vehicle following target decision method as described in any one of claims 1-3 are executed.