Stable obstacle avoidance method

Abstract

The invention discloses a stable obstacle avoidance method. The stable obstacle avoidance method comprises the following steps: acquiring multi-frame obstacle transverse distance measurement data; wherein the obstacle transverse distance measurement and calculation data is used for expressing a transverse distance of an obstacle relative to a reference line in a Frenet coordinate system; a reference line in the Frenet coordinate system is a center line of a lane line; based on the obstacle transverse distance measurement and calculation data, the degree of the obstacle invading the lane where the vehicle is located is judged, when the obstacle transverse distance measurement and calculation data exceeds a preset distance threshold value, it is judged that the degree of the obstacle invading the lane where the vehicle is located exceeds a preset invading range, and an obstacle avoidance request is generated; and recording the total frame number corresponding to the generated obstacle avoidance request, and activating the transverse obstacle avoidance trajectory planning and the longitudinal obstacle avoidance trajectory planning when the total frame number corresponding to the generated obstacle avoidance request exceeds a preset frame number threshold.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving, and in particular to a method for stabilizing and avoiding obstacles. Background Art

[0002] In recent years, with the continuous improvement of the level of automobile automation and intelligence, autonomous driving has become an important development direction for the development of automobile intelligence. Vehicles in autonomous driving mode need to be able to accurately identify and avoid obstacles during driving to ensure driving safety and the effectiveness of path planning. At present, commonly used obstacle detection and avoidance technologies include LiDAR (Light detection and Ranging), cameras, ultrasonic sensors, and millimeter-wave radars. However, in order to create more cost-effective products, autonomous driving functions may encounter challenges in pure visual perception scenarios, such as insufficient light, severe weather conditions, or jumps in obstacle contours, which can cause sensors to fail to accurately detect obstacles, thereby affecting the vehicle's obstacle avoidance capabilities.

[0003] Specifically, in autonomous driving technology, although existing obstacle detection and avoidance technologies have made significant progress, they still face some challenges. In pure visual perception scenarios, the limitations of currently used sensors, such as LiDAR, cameras, ultrasonic sensors, and millimeter-wave radars, are particularly prominent.

[0004] First, the camera's field of view and image quality will be seriously affected in bad weather conditions such as rain, snow or haze, resulting in a significant decrease in the accuracy of obstacle detection. Secondly, due to the camera's viewing angle and imaging method, the obstacle contour may jump in the identification of fast-moving obstacles or extremely low-speed obstacles with complex shapes, resulting in unstable detection results. In addition, the camera's viewing angle is limited, and when there are other vehicles or objects blocking the front, it may not be able to accurately detect obstacles, which in turn affects path planning. Therefore, how to solve the jitter problem caused by perception contour jumps, deal with the problem of obstacle loss during obstacle avoidance, and design safe, stable and comfortable lateral and longitudinal obstacle avoidance strategies have become key issues that need to be urgently resolved in current autonomous driving technology. These key issues that need to be resolved affect the driving safety of autonomous vehicles in complex environments and the effectiveness of path planning, and further technological breakthroughs are needed to improve the performance of existing systems. Summary of the invention

[0005] An advantage of the present application is that it provides a stable obstacle avoidance method, wherein the stable obstacle avoidance method proposes a horizontal and vertical obstacle avoidance strategy, which can improve the safety and stability of obstacle avoidance.

[0006] An advantage of the present application is that it provides a stable obstacle avoidance method, wherein the stable obstacle avoidance method can reduce the influence of the jitter caused by contour jump on the obstacle avoidance trajectory to a certain extent.

[0007] One advantage of the present application is that it provides a stable obstacle avoidance method, wherein the stable obstacle avoidance method does not need to rely on lidar, millimeter wave radar, etc., and can achieve stable obstacle avoidance based only on the image obtained by the camera module, which can reduce costs and computing resource requirements to a certain extent.

[0008] According to one aspect of the present application, a method for stabilizing obstacle avoidance is provided, which comprises the steps of:

[0009] Acquire multiple frames of obstacle lateral distance measurement data; wherein the obstacle lateral distance measurement data is used to express the lateral distance of the obstacle relative to a reference line in a Frenet coordinate system; the reference line in the Frenet coordinate system is the center line of the lane line;

[0010] Determining the degree of intrusion of the obstacle into the lane where the vehicle is located based on the obstacle lateral distance measurement data, and determining that the degree of intrusion of the obstacle into the lane where the vehicle is located exceeds a preset intrusion range when the obstacle lateral distance measurement data exceeds a preset distance threshold, and generating an obstacle avoidance request;

[0011] Record the total number of frames corresponding to the generated obstacle avoidance requests, and activate lateral obstacle avoidance trajectory planning and longitudinal obstacle avoidance trajectory planning when the total number of frames corresponding to the generated obstacle avoidance requests exceeds a preset frame number threshold;

[0012] Wherein, in the process of activating the lateral obstacle avoidance trajectory planning, starting from the planning starting point, a plurality of reference points are formed at intervals of a preset longitudinal distance in the longitudinal direction on the reference path, the entire reference path is traversed, and the lateral distance of each reference point on the reference path relative to the reference line in the Frenet coordinate system is calculated to obtain the lateral distance measurement data of the vehicle; based on the lateral distance measurement data of the vehicle and the lateral distance measurement data of the obstacle corresponding to each reference point, the lateral distance upper boundary value and the lateral distance lower boundary value of the drivable area of ​​the vehicle on the entire reference path are determined; based on the lateral distance upper boundary value and the lateral distance lower boundary value of the drivable area of ​​the vehicle, a reference line of the obstacle avoidance trajectory is generated to generate the obstacle avoidance trajectory;

[0013] During the activation of the longitudinal obstacle avoidance trajectory planning, the speed limit of the ego vehicle is generated based on the distance between the ego vehicle and the obstacle in the longitudinal direction, the degree of the obstacle intrusion into the lane where the ego vehicle is located, and the curvature of the obstacle avoidance trajectory.

[0014] In one embodiment of the obstacle avoidance method according to the present application, multiple frames of obstacle boundary data are obtained, including the steps of: projecting the obstacle onto a reference line of a Frenet coordinate system to obtain an arc length of the obstacle from the vehicle and a lateral distance of the obstacle relative to the reference line of the Frenet coordinate system, wherein the lateral distance of the obstacle relative to the reference line of the Frenet coordinate system is used as initial obstacle lateral distance data; filtering the initial obstacle lateral distance data to obtain filtered obstacle lateral distance data, wherein the filtered obstacle lateral distance data is used as the obstacle lateral distance measurement data.

[0015] In one embodiment of the obstacle stable avoidance method described in the present application, in the process of filtering the initial obstacle lateral distance data, the filtered obstacle lateral distance data is calculated based on the initial obstacle lateral distance data and the corresponding weight value.

[0016] In one embodiment of the obstacle stable avoidance method described in the present application, the initial obstacle lateral distance data is filtered by a weighted moving average filter, including the steps of: creating a lateral distance array, the lateral distance array including multiple frames for storing the initial obstacle lateral distance data; creating a weight array, the weight array including multiple frames for storing initial weight values; normalizing the initial weight values ​​to obtain normalized weight values, and using the normalized weight values ​​as the final weight values; and calculating the filtered obstacle lateral distance data based on the initial obstacle lateral distance data and the normalized weight values.

[0017] In one embodiment of the obstacle avoidance method according to the present application, the normalized weight value is: Wherein, i represents the order of the frame; N represents the number of frames in the weight array; ψ(i) represents the normalized weight value; represents the initial weight value; the filtered obstacle lateral distance data is: in, represents the filtered obstacle lateral distance data; l(i) represents the initial obstacle lateral distance data.

[0018] In one embodiment of the obstacle avoidance method according to the present application, in the lateral distance array, the larger the value of the frame order is, the larger the initial weight value corresponding to the initial obstacle lateral distance data is.

[0019] In one embodiment of the obstacle stabilization and obstacle avoidance method described in the present application, the obstacle stabilization and obstacle avoidance method also includes the step of determining the left and right boundaries of the vehicle based on the lane type, lane line type, and lane width of the lane in which the vehicle is located.

[0020] In one embodiment of the obstacle avoidance method for stabilization described in the present application, in the process of generating an obstacle avoidance request and recording the total number of frames corresponding to the generated obstacle avoidance request, when an oncoming vehicle is detected, the originally recorded total number of frames corresponding to the generated obstacle avoidance request is cleared, and the obstacle avoidance request is regenerated and the total number of frames corresponding to the generated obstacle avoidance request is re-recorded.

[0021] In one embodiment of the obstacle stabilization and avoidance method described in the present application, the maximum value of the obstacle lateral distance measurement data is used as the lower boundary value of the lateral distance of the drivable area of ​​the vehicle on the entire reference path, and half of the sum of the upper boundary value and the lower boundary value of the lateral distance of the drivable area of ​​the vehicle is used as the reference line of the obstacle avoidance trajectory.

[0022] In one embodiment of the obstacle avoidance method for stabilization described in the present application, during the lateral obstacle avoidance trajectory planning process, a preliminary obstacle avoidance trajectory is generated based on the lower boundary value of the lateral distance of the drivable area of ​​the vehicle on the reference path, the upper boundary value of the lateral distance of the drivable area of ​​the vehicle on the reference path, and the reference line of the obstacle avoidance trajectory; the preliminary obstacle avoidance trajectory is optimized so that the smoothness of the optimized obstacle avoidance trajectory is higher than the smoothness of the preliminary obstacle avoidance trajectory; wherein the optimized obstacle avoidance trajectory is used as the final obstacle avoidance trajectory.

[0023] Further objectives and advantages of the present application will be fully reflected through understanding of the following description and drawings.

[0024] These and other objects, features and advantages of the present application are fully reflected in the following detailed description, drawings and claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the drawings, the same reference numerals generally represent the same components or.

[0026] Figure 1 A flowchart schematic diagram of an obstacle stabilization and avoidance method according to an embodiment of the present application is illustrated.

[0027] Figure 2 A flowchart schematic diagram of a specific example of an obstacle stabilization and avoidance method according to an embodiment of the present application is illustrated.

[0028] Figure 3A schematic diagram of a drivable area of ​​a vehicle generated during the lateral obstacle avoidance trajectory planning process of the obstacle stabilization and obstacle avoidance method according to an embodiment of the present application is illustrated. DETAILED DESCRIPTION

[0029] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0030] It is understood that the term "one" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element can be one, while in another embodiment, the number of the element can be multiple, and the term "one" cannot be understood as a limitation on the number. "Multiple" means greater than or equal to two.

[0031] Although ordinals such as "first," "second," and the like will be used to describe various components, those components are not limited herein. The term is used only to distinguish one component from another. For example, a first component may be referred to as a second component, and likewise, a second component may be referred to as a first component without departing from the teachings of the present application. The term "and / or" as used herein includes any and all combinations of one or more associated listed items.

[0032] The terms used herein are only used for the purpose of describing various embodiments and are not intended to be limiting. As used herein, the singular form is intended to include the plural form, unless the context clearly indicates an exception. It will also be understood that the terms "including" and / or "having" when used in this specification specify the presence of the described features, numbers, operations, components, elements, or combinations thereof, without excluding the presence or addition of one or more other features, numbers, operations, components, elements, or combinations thereof.

[0033] Schematic obstacle avoidance method

[0034] like Figures 1 to 3 As shown, the obstacle stabilization and avoidance method according to the embodiment of the present application is explained. Figure 1As shown, the obstacle avoidance method according to the embodiment of the present application includes the following steps: S110, obtaining multiple frames of obstacle lateral distance measurement data; wherein the obstacle lateral distance measurement data is used to express the lateral distance of the obstacle relative to a reference line in the Frenet coordinate system; the reference line in the Frenet coordinate system is the center line of the lane line; S120, judging the degree of obstacle invasion into the lane where the vehicle is located based on the obstacle lateral distance measurement data, and judging that the degree of obstacle invasion into the lane where the vehicle is located exceeds a preset invasion range when the obstacle lateral distance measurement data exceeds a preset distance threshold, and generating an obstacle avoidance request; S130, recording the total number of frames corresponding to the generated obstacle avoidance request, and activating the lateral obstacle avoidance trajectory when the total number of frames corresponding to the generated obstacle avoidance request exceeds a preset frame number threshold. Planning and longitudinal obstacle avoidance trajectory planning; wherein, in the process of lateral obstacle avoidance trajectory planning, starting from the planning starting point, the lateral distances of multiple reference points on the reference path relative to the reference line in the Frenet coordinate system are calculated to obtain the lateral distance measurement data of the vehicle; based on the lateral distance measurement data of the vehicle and the lateral distance measurement data of the obstacle corresponding to each reference point, the lateral distance upper boundary value and the lateral distance lower boundary value of the drivable area of ​​the vehicle on the entire reference path are determined; based on the lateral distance upper boundary value and the lateral distance lower boundary value of the drivable area of ​​the vehicle, a reference line of the obstacle avoidance trajectory is generated to generate the obstacle avoidance trajectory; in the longitudinal obstacle avoidance trajectory planning, the speed limit of the vehicle is generated based on the distance between the vehicle and the obstacle in the longitudinal direction, the degree of obstacle invasion into the lane where the vehicle is located, and the curvature of the obstacle avoidance trajectory.

[0035] In step S110, multiple frames of obstacle lateral distance measurement data are obtained. Specifically, the obstacle (represented by "obs" in the figure) is projected onto the reference line of the Frenet coordinate system to obtain the arc length of the obstacle from the ego vehicle (represented by "ego" in the figure) and the lateral distance of the obstacle relative to the reference line of the Frenet coordinate system, wherein the center line of the lane line is used as the reference line in the Frenet coordinate system. The ego vehicle refers to an autonomous driving vehicle. The lateral distance of the obstacle relative to the reference line of the Frenet coordinate system can be used as the obstacle lateral distance measurement data; the lateral distance of the obstacle relative to the reference line of the Frenet coordinate system can also be used as the initial obstacle lateral distance data, and then the initial obstacle lateral distance data is filtered to obtain the filtered obstacle lateral distance data, and the filtered obstacle lateral distance data is used as the obstacle lateral distance measurement data.

[0036] Accordingly, in a specific example of the present application, step S110 includes the steps of: S111, projecting the obstacle onto the reference line of the Frenet coordinate system to obtain the arc length of the obstacle from the vehicle and the lateral distance of the obstacle relative to the reference line of the Frenet coordinate system, wherein the lateral distance of the obstacle relative to the reference line of the Frenet coordinate system is used as the initial obstacle lateral distance data; S112, filtering the initial obstacle lateral distance data to obtain filtered obstacle lateral distance data, wherein the filtered obstacle lateral distance data is used as the obstacle lateral distance measurement data.

[0037] More specifically, the image of the environment in which the vehicle is located can be obtained through the camera module, and then the obstacles in the image can be identified. The lateral distances of multiple feature points of the identified obstacle relative to the reference line of the Frenet coordinate system can be measured, and the lateral distances of multiple feature points of the obstacle relative to the reference line of the Frenet coordinate system are used as the initial obstacle lateral distance data; wherein, at least some of the multiple feature points of the obstacle are feature points on the boundary of the obstacle, that is, the contour points of the obstacle; the lateral distances of the feature points on the boundary of the obstacle relative to the reference line of the Frenet coordinate system form the initial obstacle boundary lateral distance in the initial obstacle lateral distance data.

[0038] In the process of filtering the initial obstacle lateral distance data to obtain the filtered obstacle lateral distance data, the initial obstacle boundary lateral distance in the initial obstacle lateral distance data is filtered to obtain the filtered obstacle boundary lateral distance.

[0039] It is worth mentioning that in the embodiment of the present application, the boundary of the obstacle can be judged in real time through the obstacle lateral distance measurement data, which can reduce the impact of the jitter caused by contour jump on the obstacle avoidance trajectory to a certain extent, and improve the safety and stability of obstacle avoidance.

[0040] In a specific example of the present application, the initial obstacle lateral distance data may be filtered by a weighted moving average filter.

[0041] Further, in the process of filtering the initial obstacle lateral distance data, the filtered obstacle lateral distance data is calculated based on the initial obstacle lateral distance data and the corresponding weight value.

[0042] Specifically, first, a lateral distance array is created, the lateral distance array includes multiple frames, and is used to store the initial obstacle lateral distance data; then, a weight array is created, the weight array includes multiple frames, and is used to store initial weight values; then, the filtered obstacle lateral distance data can be calculated based on the initial obstacle lateral distance data and the initial weight values, or the initial weight values ​​can be normalized first, and then the filtered obstacle lateral distance data can be calculated based on the initial obstacle lateral distance data and the normalized weight values. The multiple frames of the lateral distance array are sorted in sequence according to the generation time. Accordingly, the multiple frames of the weight array are sorted in sequence according to the generation time.

[0043] Accordingly, in a specific example of the present application, step S112 includes the steps of: S1121, creating a lateral distance array, the lateral distance array including multiple frames for storing the initial obstacle lateral distance data; S1122, creating a weight array, the weight array including multiple frames for storing initial weight values; S1123, normalizing the initial weight values ​​to obtain normalized weight values, and using the normalized weight values ​​as the final weight values; and, S1124, calculating the filtered obstacle lateral distance data based on the initial obstacle lateral distance data and the normalized weight values.

[0044] More specifically, the normalized weight value is: Wherein, i represents the order of the frame; N represents the number of frames in the weight array; ψ(i) represents the normalized weight value; represents the initial weight value; the filtered obstacle lateral distance data is: in, represents the filtered obstacle lateral distance data; l(i) represents the initial obstacle lateral distance data.

[0045] The number of frames in the weight array is the same as the number of frames in the horizontal distance array; the length of the weight array is the same as the length of the horizontal distance array. The number of frames in the horizontal distance array is mainly selected according to the actual application, for example, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, etc. In one example of the present application, the number of frames in the horizontal distance array is 10.

[0046] It is worth mentioning that a high weight is assigned to the newer data. At the same time, if the data changes quickly, the weight of the new data is increased; if the data changes slowly, the weight of the new data is reduced. The newer data refers to the data that is later in time sequence, and the data with a larger value of the corresponding frame order. Correspondingly, in the horizontal distance array, the larger the value of the frame order, the larger the initial weight value corresponding to the initial obstacle horizontal distance data; and the larger the frame rate of the frame of the horizontal distance array, the larger the initial weight value corresponding to the initial obstacle horizontal distance data.

[0047] The boundary in the lateral direction of the vehicle can be determined based on the lane type, lane line type, and lane width of the lane where the vehicle is located.

[0048] In step S120, the degree of the obstacle's invasion into the lane where the vehicle is located is determined based on the obstacle's lateral distance measurement data. Specifically, the larger the obstacle's lateral distance measurement data is and the closer it is to the center of the lane, the greater the degree of the obstacle's invasion into the lane where the vehicle is located. Accordingly, when the obstacle's lateral distance measurement data exceeds a preset distance threshold, it is determined that the obstacle's invasion into the lane where the vehicle is located exceeds a preset invasion range, and an obstacle avoidance request is generated. The preset distance threshold can be set according to the lane type, lane line type, and lane width of the lane where the vehicle is located.

[0049] Furthermore, the degree of the obstacle's intrusion into the lane where the vehicle is located can be determined based on the filtered obstacle boundary lateral distance in the obstacle lateral distance measurement data. The larger the filtered obstacle boundary lateral distance is and the closer it is to the center of the lane, the greater the degree of the obstacle's intrusion into the lane where the vehicle is located. Accordingly, when the filtered obstacle boundary lateral distance exceeds a preset distance threshold, it is determined that the obstacle's intrusion into the lane where the vehicle is located exceeds a preset intrusion range, and an obstacle avoidance request is generated.

[0050] In step S130, the total number of frames corresponding to the generated obstacle avoidance requests is recorded, and when the total number of frames corresponding to the generated obstacle avoidance requests exceeds the preset frame number threshold, the lateral obstacle avoidance trajectory planning and the longitudinal obstacle avoidance trajectory planning are activated. Specifically, the total number of frames corresponding to the generated obstacle avoidance requests can be recorded by a counter, and when the total number of frames corresponding to the generated obstacle avoidance requests exceeds the preset frame number threshold, the obstacle avoidance decision is output; the total generation time of the frames corresponding to the generated obstacle avoidance requests can also be recorded by a timer, and when the time recorded by the timer exceeds the preset duration, it means that the total number of frames corresponding to the generated obstacle avoidance requests exceeds the preset frame number threshold, and the obstacle avoidance decision is output. The obstacle avoidance decision includes lateral obstacle avoidance trajectory planning and longitudinal obstacle avoidance trajectory planning.

[0051] If the total number of frames corresponding to the generated obstacle avoidance request does not exceed the preset frame number threshold, the total number of frames corresponding to the generated obstacle avoidance request originally recorded is cleared, and the obstacle avoidance request is regenerated and the total number of frames corresponding to the generated obstacle avoidance request is re-recorded.

[0052] Accordingly, when the total number of frames accumulated by the counter exceeds the preset frame number threshold or the time accumulated by the timer exceeds the preset time threshold, the total number of frames corresponding to the obstacle avoidance request originally recorded is cleared, and the obstacle avoidance request is regenerated and the total number of frames corresponding to the generated obstacle avoidance request is re-recorded.

[0053] It is worth mentioning that in the process of generating obstacle avoidance requests and recording the total number of frames corresponding to the generated obstacle avoidance requests, when an oncoming vehicle is detected, the total number of frames corresponding to the generated obstacle avoidance requests originally recorded is cleared, and the obstacle avoidance requests are regenerated and the total number of frames corresponding to the generated obstacle avoidance requests are re-recorded.

[0054] In the process of activating the lateral obstacle avoidance trajectory planning, starting from the planning starting point, a plurality of reference points are formed on the reference path at intervals of a preset longitudinal distance in the longitudinal direction S. Accordingly, the reference points on each adjacent reference path are separated by a preset longitudinal distance in the longitudinal direction.

[0055] Traverse the entire reference path, calculate the lateral distance of each reference point on the reference path relative to the reference line in the Frenet coordinate system, so as to obtain the lateral distance measurement data of the vehicle; determine the lateral distance upper boundary value lmax and the lateral distance lower boundary value lmin of the drivable area of ​​the vehicle on the entire reference path based on the lateral distance measurement data of the vehicle and the lateral distance measurement data of the obstacle corresponding to each reference point; generate the reference line ref_1 of the obstacle avoidance trajectory based on the lateral distance upper boundary value lmax and the lateral distance lower boundary value lmin of the drivable area of ​​the vehicle, such as Figure 3 As shown, to generate obstacle avoidance trajectories.

[0056] Specifically, the larger the obstacle lateral distance measurement data is, the closer it is to the center of the lane and the vehicle. Accordingly, the maximum value of the obstacle lateral distance measurement data is used as the lower boundary value of the lateral distance of the drivable area of ​​the vehicle on the entire reference path.

[0057] In the embodiment of the present application, half of the sum of the upper boundary value of the lateral distance and the lower boundary value of the lateral distance of the drivable area of ​​the vehicle is used as the reference line of the obstacle avoidance trajectory.

[0058] In the embodiment of the present application, a preliminary obstacle avoidance trajectory is generated based on the lower boundary value of the lateral distance of the drivable area of ​​the vehicle on the reference path, the upper boundary value of the lateral distance of the drivable area of ​​the vehicle on the reference path, and the reference line of the obstacle avoidance trajectory; the preliminary obstacle avoidance trajectory can be optimized to improve the smoothness of the obstacle avoidance trajectory and ensure the safety and stability of obstacle avoidance. Accordingly, the smoothness of the optimized obstacle avoidance trajectory is higher than the smoothness of the preliminary obstacle avoidance trajectory; the optimized obstacle avoidance trajectory is used as the final obstacle avoidance trajectory.

[0059] Specifically, the preliminary obstacle avoidance trajectory may be optimized by a quadratic programming solution or other optimization methods.

[0060] This application not only performs lateral obstacle avoidance trajectory planning to achieve obstacle avoidance in the lateral space, but also performs longitudinal obstacle avoidance trajectory planning to limit the speed of the vehicle, which can improve the obstacle avoidance safety and stability to a certain extent.

[0061] Specifically, during the process of activating the longitudinal obstacle avoidance trajectory planning, the speed limit of the ego vehicle is generated based on the distance between the ego vehicle and the obstacle in the longitudinal direction, the degree of obstacle invasion into the lane where the ego vehicle is located, and the curvature of the obstacle avoidance trajectory to ensure safe passage through the obstacle.

[0062] It is worth mentioning that the obstacle avoidance method does not rely on lidar, millimeter-wave radar, etc., and can achieve stable obstacle avoidance based only on the images taken by the camera module, which can reduce costs and computing resource requirements to a certain extent.

[0063] Below, the obstacle stabilization and avoidance method according to the embodiment of the present application is further explained through specific examples.

[0064] In a specific example of the present application, first, obtain multiple frames of obstacle lateral distance measurement data. Specifically, project the obstacle onto the reference line of the Frenet coordinate system to obtain the arc length of the obstacle from the vehicle and the lateral distance of the obstacle relative to the reference line of the Frenet coordinate system, wherein the reference line in the Frenet coordinate system is the center line of the lane line, and the lateral distance of the obstacle relative to the reference line of the Frenet coordinate system is used as the initial obstacle lateral distance data. The initial obstacle lateral distance data is filtered using a weighted moving average filter to obtain filtered obstacle lateral distance data, wherein the filtered obstacle lateral distance data is used as the obstacle lateral distance measurement data. In the process of filtering the initial obstacle lateral distance data using a weighted moving average filter, a lateral distance array is first created, the lateral distance array includes multiple frames for storing the initial obstacle lateral distance data; then a weight array is created, the weight array includes multiple frames for storing initial weight values; then, the initial weight values ​​are normalized to obtain normalized weight values, and the normalized weight values ​​are used as the final weight values; finally, the filtered obstacle lateral distance data is calculated based on the initial obstacle lateral distance data and the normalized weight values. In the lateral distance array, the larger the value of the order of the frame is, the larger the initial weight value corresponding to the initial obstacle lateral distance data is; and the larger the frame rate of the frame of the lateral distance array is, the larger the initial weight value corresponding to the initial obstacle lateral distance data is.

[0065] After obtaining multiple frames of obstacle lateral distance measurement data, determine the left and right boundaries of the vehicle based on the lane type, lane line type, and lane width of the lane where the vehicle is located; determine the degree of obstacle invasion into the lane where the vehicle is located based on the obstacle lateral distance measurement data; when the obstacle lateral distance measurement data exceeds the preset distance threshold, determine that the degree of obstacle invasion into the lane where the vehicle is located exceeds the preset invasion range, and generate an obstacle avoidance request. Record the total number of frames corresponding to the generated obstacle avoidance request, and activate the lateral obstacle avoidance trajectory planning and the longitudinal obstacle avoidance trajectory planning when the total number of frames corresponding to the generated obstacle avoidance request exceeds the preset frame number threshold. In the process of generating obstacle avoidance requests and recording the total number of frames corresponding to the generated obstacle avoidance requests, when an oncoming vehicle is detected, the total number of frames corresponding to the generated obstacle avoidance requests originally recorded is cleared, and the obstacle avoidance request is regenerated and the total number of frames corresponding to the generated obstacle avoidance request is re-recorded.

[0066] In the process of activating the lateral obstacle avoidance trajectory planning, starting from the planning starting point, a plurality of reference points are formed at predetermined longitudinal distances in the longitudinal direction S on the reference path. The entire reference path is traversed, and the lateral distance of each reference point on the reference path relative to the reference line in the Frenet coordinate system is calculated to obtain the lateral distance measurement data of the vehicle; based on the lateral distance measurement data of the vehicle and the lateral distance measurement data of the obstacle corresponding to each reference point, the lateral distance upper boundary value lmax and the lateral distance lower boundary value lmin of the drivable area of ​​the vehicle on the entire reference path are determined; based on the lateral distance upper boundary value lmax and the lateral distance lower boundary value lmin of the drivable area of ​​the vehicle, a reference line ref_1 of the obstacle avoidance trajectory is generated; wherein, half of the sum of the lateral distance upper boundary value lmax and the lateral distance lower boundary value lmin of the drivable area of ​​the vehicle is used as the reference line ref_1 of the obstacle avoidance trajectory. A preliminary obstacle avoidance trajectory is generated based on the lower boundary value of the lateral distance of the drivable area of ​​the ego vehicle on the reference path, the upper boundary value of the lateral distance of the drivable area of ​​the ego vehicle on the reference path, and the reference line of the obstacle avoidance trajectory; the preliminary obstacle avoidance trajectory is optimized by a quadratic programming solution method so that the smoothness of the optimized obstacle avoidance trajectory is higher than the smoothness of the preliminary obstacle avoidance trajectory; wherein the optimized obstacle avoidance trajectory is used as the final obstacle avoidance trajectory. In the process of activating the longitudinal obstacle avoidance trajectory planning, the speed limit of the ego vehicle is generated based on the distance between the ego vehicle and the obstacle in the longitudinal direction, the degree of the obstacle intrusion into the lane where the ego vehicle is located, and the curvature of the obstacle avoidance trajectory to ensure safe passage through the obstacle.

[0067] The present application and its implementation methods are described above, and such description is not restrictive. The drawings show only one implementation method of the present application, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it and design structural methods and embodiments similar to the technical solution without creative design without departing from the inventive purpose of the present application, they should all fall within the protection scope of the present application.

Claims

1. A stable obstacle avoidance method, characterized in that: Includes steps: Acquire multiple frames of obstacle lateral distance measurement data; wherein the obstacle lateral distance measurement data is used to express the lateral distance of the obstacle relative to a reference line in a Frenet coordinate system; the reference line in the Frenet coordinate system is the center line of the lane line; Determining the degree of intrusion of the obstacle into the lane where the vehicle is located based on the obstacle lateral distance measurement data, and determining that the degree of intrusion of the obstacle into the lane where the vehicle is located exceeds a preset intrusion range when the obstacle lateral distance measurement data exceeds a preset distance threshold, and generating an obstacle avoidance request; Record the total number of frames corresponding to the generated obstacle avoidance requests, and activate lateral obstacle avoidance trajectory planning and longitudinal obstacle avoidance trajectory planning when the total number of frames corresponding to the generated obstacle avoidance requests exceeds a preset frame number threshold; Wherein, in the process of activating the lateral obstacle avoidance trajectory planning, starting from the planning starting point, a plurality of reference points are formed at intervals of a preset longitudinal distance in the longitudinal direction on the reference path, the entire reference path is traversed, and the lateral distance of each reference point on the reference path relative to the reference line in the Frenet coordinate system is calculated to obtain the lateral distance measurement data of the vehicle; based on the lateral distance measurement data of the vehicle and the lateral distance measurement data of the obstacle corresponding to each reference point, the lateral distance upper boundary value and the lateral distance lower boundary value of the drivable area of ​​the vehicle on the entire reference path are determined; based on the lateral distance upper boundary value and the lateral distance lower boundary value of the drivable area of ​​the vehicle, a reference line of the obstacle avoidance trajectory is generated to generate the obstacle avoidance trajectory; During the activation of the longitudinal obstacle avoidance trajectory planning, the speed limit of the ego vehicle is generated based on the distance between the ego vehicle and the obstacle in the longitudinal direction, the degree of the obstacle intrusion into the lane where the ego vehicle is located, and the curvature of the obstacle avoidance trajectory.

2. The obstacle avoidance method according to claim 1, characterized in that: Obtaining multi-frame obstacle boundary data includes the following steps: Project the obstacle onto the reference line of the Frenet coordinate system to obtain the arc length of the obstacle from the ego vehicle and the lateral distance of the obstacle relative to the reference line of the Frenet coordinate system, wherein the lateral distance of the obstacle relative to the reference line of the Frenet coordinate system is used as the initial obstacle lateral distance data; The initial obstacle lateral distance data is filtered to obtain filtered obstacle lateral distance data, wherein the filtered obstacle lateral distance data is used as the obstacle lateral distance measurement data.

3. The obstacle avoidance method according to claim 2, characterized in that: In the process of filtering the initial obstacle lateral distance data, the filtered obstacle lateral distance data is calculated based on the initial obstacle lateral distance data and the corresponding weight value.

4. The obstacle avoidance method according to claim 3, characterized in that: The initial obstacle lateral distance data is filtered by a weighted moving average filter, comprising the steps of: Creating a lateral distance array, wherein the lateral distance array includes a plurality of frames and is used to store the initial obstacle lateral distance data; Creating a weight array, wherein the weight array includes a plurality of frames and is used to store initial weight values; Normalizing the initial weight value to obtain a normalized weight value, and using the normalized weight value as the final weight value; The filtered obstacle lateral distance data is calculated based on the initial obstacle lateral distance data and the normalized weight value.

5. The obstacle avoidance method according to claim 4, characterized in that: The normalized weight value is: Wherein, i represents the order of the frame; N represents the number of frames in the weight array; ψ(i) represents the normalized weight value; represents the initial weight value; The filtered obstacle lateral distance data is: in, represents the filtered obstacle lateral distance data; l(i) represents the initial obstacle lateral distance data.

6. The obstacle avoidance method according to claim 4, characterized in that: In the lateral distance array, the larger the value of the frame order is, the larger the initial weight value corresponding to the initial obstacle lateral distance data is.

7. The obstacle avoidance method according to claim 1, characterized in that: The obstacle stabilization and avoidance method also includes the steps of: determining the left and right boundaries of the vehicle based on the lane type, lane line type, and lane width of the lane where the vehicle is located.

8. The obstacle avoidance method according to claim 1, characterized in that: In the process of generating obstacle avoidance requests and recording the total number of frames corresponding to the generated obstacle avoidance requests, when an oncoming vehicle is detected, the total number of frames corresponding to the generated obstacle avoidance requests originally recorded is cleared, and the obstacle avoidance requests are regenerated and the total number of frames corresponding to the generated obstacle avoidance requests are re-recorded.

9. The obstacle avoidance method according to claim 1, characterized in that: The maximum value of the obstacle lateral distance measurement data is used as the lower boundary value of the lateral distance of the drivable area of ​​the vehicle on the entire reference path, and half of the sum of the upper boundary value and the lower boundary value of the lateral distance of the drivable area of ​​the vehicle is used as the reference line of the obstacle avoidance trajectory.

10. The obstacle avoidance method according to claim 1, characterized in that: In the lateral obstacle avoidance trajectory planning process, a preliminary obstacle avoidance trajectory is generated based on a lower boundary value of a lateral distance of a drivable area of ​​the ego vehicle on the reference path, an upper boundary value of a lateral distance of a drivable area of ​​the ego vehicle on the reference path, and a reference line of the obstacle avoidance trajectory; The preliminary obstacle avoidance trajectory is optimized so that the smoothness of the optimized obstacle avoidance trajectory is higher than the smoothness of the preliminary obstacle avoidance trajectory; wherein the optimized obstacle avoidance trajectory is used as the final obstacle avoidance trajectory.

Images