In-lane avoidance method and system based on avoidance space construction

Abstract

The invention provides an avoidance space construction-based in-lane avoidance method and system, and the method comprises the steps: determining a vehicle-following target of a vehicle in a current lane, and screening an avoidance obstacle according to the vehicle-following target; sampling the boundary of the avoidance obstacle close to the reference line, and generating a first obstacle data set; expanding the first obstacle data set according to a preset expansion value, and generating a second obstacle data set; integrating the second obstacle data set with the lane line, and generating an avoidance space; and generating an avoiding track according to the avoiding space. The technical problems that in the prior art, when obstacles exist on the left side, on the right side and on the two sides, different offset strategies need to be adopted, calculation logic is complex, and execution efficiency is low are solved.

Description

Technical Field

[0001] The present invention relates to the field of electronic information technology, and in particular to a lane avoidance method and avoidance system based on constructing an avoidance space. Background Art

[0002] With the development of autonomous driving technology and the continuous improvement of vehicle intelligence, obstacle avoidance technology, as a key module in autonomous driving systems, has gradually become a research focus. Early autonomous driving technology focused primarily on proof of concept and path planning. For example, technical research in the 1980s achieved preliminary road recognition and navigation capabilities through vision and radar sensors. Current lane avoidance technology calculates the required lateral offset distance for the vehicle based on obstacles on the left and right sides. This offset is then applied to a reference line, causing subsequent vehicles to move toward the reference line to avoid obstacles.

[0003] However, there is at least one of the following problems in the related technology: the existing technology requires different offset strategies for three different situations: obstacles on the left, obstacles on the right, and obstacles on both sides. The calculation logic is relatively complex and the execution efficiency is low. Summary of the Invention

[0004] The technical problem solved by the present invention is that the existing technology requires different offset strategies for three different situations: obstacles on the left, obstacles on the right, and obstacles on both sides. The calculation logic is relatively complex and the execution efficiency is low.

[0005] To solve the above problems, the present invention provides a lane avoidance method based on constructing an avoidance space, comprising: determining a following target of the ego-vehicle in the current lane, and screening out avoidance obstacles based on the following target; sampling the boundaries of the avoidance obstacles close to a reference line, and generating a first obstacle data set; expanding the first obstacle data set by a preset expansion value, and generating a second obstacle data set; integrating the second obstacle data set with the lane line, and generating an avoidance space; and generating an avoidance trajectory based on the avoidance space.

[0006] Compared with the existing technology, the technical effects achieved by adopting this technical solution are as follows: compared with the existing technology, which requires different offset strategies for the three situations of obstacles on the left, obstacles on the right, and obstacles on both sides; this method reduces the processing of unnecessary obstacles by determining the following target of the vehicle in the current lane and screening out avoidable obstacles based on the following target; samples the boundary of the avoidable obstacles near the reference line and generates a first obstacle data set, which is then expanded by a preset expansion value to generate a second obstacle data set. The expansion ensures that the generated second obstacle data set fully considers the avoidable obstacles in the two adjacent lanes; the second obstacle data set is integrated with the lane line to generate an avoidance space, and an avoidance trajectory is generated based on the avoidance space. By accurately generating the avoidance space and a reasonable avoidance trajectory, the vehicle can avoid obstacles in a timely and accurate manner when encountering obstacles, thereby effectively reducing risks and ensuring the safety of passengers and other road users. The execution logic of this solution is simple and clear, and the execution efficiency is high.

[0007] In one example of the present invention, a following target of the ego vehicle in the current lane is determined, and obstacles to be avoided are screened out based on the following target, including: screening out the vehicle closest to the ego vehicle in the current lane and recording it as the following target; screening out vehicles in adjacent lanes whose longitudinal distance is less than the following target and recording them as the first target group; screening out vehicles in the first target group whose lateral distance from the reference line is less than a preset distance and recording them as obstacles to be avoided.

[0008] Compared with the existing technology, the technical effect achieved by adopting this technical solution is: through clear screening steps, the following vehicle targets and obstacle avoidance are screened out, so as to quickly and accurately screen out the most relevant vehicles, thereby clarifying the obstacle avoidance to achieve accurate and effective screening and identification.

[0009] In one embodiment of the present invention, sampling is performed on a boundary of an obstacle to be avoided close to a reference line, and a first obstacle data set is generated, including: calculating a longitudinal distance between a following target and the vehicle, and recording it as a first distance; calculating a sampling distance based on the first distance; taking multiple sampling points within the sampling distance; and sampling the boundary of an obstacle to be avoided close to the reference line at the sampling points to generate the first obstacle data set.

[0010] Compared with the existing technology, the technical effect achieved by adopting this technical solution is: by calculating the first distance between the following target and the vehicle, and calculating the sampling distance based on the first distance, the appropriate sampling distance and sampling points are determined to achieve more comprehensive and accurate sampling data.

[0011] In one example of the present invention, the second obstacle data set is integrated with the lane line and an avoidance space is generated, including: sampling the lane line at a sampling point to generate a lane line data set; within the same sampling point, the second obstacle data set and the lane line data set are compared, and the data close to the reference line are integrated into an integrated data set; performing secondary processing on the integrated data set to obtain a final data set; and generating an avoidance space based on the final data set.

[0012] Compared with the existing technology, the technical effect achieved by adopting this technical solution is as follows: by comparing the lane line data set and the second obstacle data set within the same sampling point to form an integrated data set, and then performing secondary processing on the integrated data set, an avoidance space is generated according to the final data set, thereby improving the accuracy and rationality of the avoidance space generation.

[0013] In one example of the present invention, the integrated data set includes a left boundary avoidance data set and a right boundary avoidance data set; the integrated data set is subjected to secondary processing to obtain a final data set, including: subtracting the left boundary avoidance data set from the right boundary avoidance data set to obtain an error data set; adjusting the data in the second obstacle data set according to the error data set, and regenerating the avoidance space.

[0014] In one example of the present invention, the second obstacle data set includes a second left obstacle avoidance data set and a second right obstacle avoidance data set; the data in the second obstacle data set is adjusted according to the error data set, and the avoidance space is regenerated, including: comparing the error data in the error data set with the vehicle width data; if the error data is less than the vehicle width data, reducing the preset expansion value of the corresponding sampling point in the second left obstacle avoidance data set and the second right obstacle avoidance data set until the error data is greater than or equal to the vehicle width data; recalculating the second obstacle data set, and regenerating the avoidance space.

[0015] Compared with the existing technology, the technical effect achieved by adopting this technical solution is: by adjusting the preset expansion value to regenerate the second obstacle data set and avoidance space, it is ensured that the avoidance space is sufficient for the vehicle to pass safely, and the feasibility of the avoidance strategy is enhanced.

[0016] In one embodiment of the present invention, generating an avoidance trajectory according to the avoidance space includes: calculating a cost function according to the avoidance space; calculating boundary conditions according to the boundary constraints; and inputting the cost function and the boundary conditions into a solver to generate the avoidance trajectory.

[0017] Compared with the existing technology, the technical effect achieved by adopting this technical solution is: by calculating the cost function and boundary conditions and inputting them into the solver to generate the avoidance trajectory, the optimization degree and feasibility of the avoidance trajectory are improved.

[0018] In one embodiment of the present invention, the cost function includes a reference line difference cost, a first-order derivative cost, a second-order derivative cost, a third-order derivative cost, and an avoidance space center cost; the cost function is calculated to satisfy the following formula 1; wherein formula 1 is: in, is the cost function, is the reference line cost weight value, is the weight value of the first-order derivative cost, is the weight value of the second-order derivative cost, is the weight value of the third-order derivative cost, To avoid the weight value of the spatial center cost, is the optimal value of the i-th sampling point, is the optimal value of the i+1th sampling point, is the left boundary value of the i-th sampling point, is the right boundary value of the i-th sampling point.

[0019] Compared with the existing technology, the technical effect achieved by adopting this technical solution is: by comprehensively considering the reference line cost, the costs of various order derivatives and the cost of the avoidance space center, and assigning corresponding weight values, the accuracy and comprehensiveness of the cost function calculation are improved.

[0020] In one embodiment of the present invention, calculating boundary conditions based on boundary constraints includes: calculating four corner points of the ego vehicle and performing approximation processing to generate ego vehicle data; generating boundary conditions based on the ego vehicle data; and generating the ego vehicle data to satisfy the following formulas 2, 3, 4, and 5: Wherein, Formula 2 is: ; Formula 3 is: ; Formula 4 is: ; Formula 5 is: ; in, is the first corner value, is the second corner value, is the third corner value, is the fourth corner value, is the distance from the center of the rear axle of the vehicle to the front of the vehicle, is the distance from the center of the rear axle to the rear of the vehicle, is the width of the vehicle.

[0021] Compared with the existing technology, the technical effect achieved by adopting this technical solution is: by calculating the four corner points of the ego vehicle and performing approximate processing, more accurate ego vehicle data is generated, and the accuracy of boundary condition calculation is improved.

[0022] On the other hand, the present invention also provides an avoidance system that can implement any of the lane avoidance methods mentioned above. The avoidance system includes: a detection module, the detection module is used to determine the following target of the vehicle in the current lane, and screen out avoidance obstacles based on the following target; the detection module is also used to sample the boundary of the avoidance obstacle close to the reference line and generate a first obstacle data set; a processing module, the processing module is used to expand the first obstacle data set with a preset expansion value and generate a second obstacle data set; the processing module is also used to integrate the second obstacle data set with the lane line and generate an avoidance space; and an execution module, the execution module is used to generate an avoidance trajectory based on the avoidance space.

[0023] Compared with the existing technology, the technical effect achieved by adopting this technical solution is: it can achieve the technical effect corresponding to any of the above examples, which will not be repeated here.

[0024] After adopting the technical solution of the present invention, the following technical effects can be achieved: (1) Compared with the prior art, which requires different offset strategies for the three situations of obstacles on the left, obstacles on the right, and obstacles on both sides, this method reduces the processing of unnecessary obstacles by determining the following target of the vehicle in the current lane and screening out the avoidance obstacles based on the following target; samples the boundary of the avoidance obstacles close to the reference line and generates a first obstacle data set, expands the first obstacle data set by a preset expansion value, and generates a second obstacle data set. The expansion makes the generated second obstacle data set fully consider the avoidance obstacles in the two adjacent lanes; integrates the second obstacle data set with the lane line and generates an avoidance space, generates an avoidance trajectory based on the avoidance space, and by accurately generating the avoidance space and reasonable avoidance trajectory, the vehicle can avoid obstacles in a timely and accurate manner when encountering obstacles, thereby effectively reducing risks and ensuring the safety of passengers and other road users. The execution logic of this solution is simple and clear, and the execution efficiency is high. (2) Through clear screening steps, the following targets and avoidance obstacles are screened out, thereby quickly and accurately screening out the most relevant vehicles, thereby clarifying the avoidance obstacles to achieve accurate and effective screening and identification; (3) By comprehensively considering the reference line cost, the costs of various derivatives, and the cost of the avoidance space center and assigning corresponding weight values, the accuracy and comprehensiveness of the cost function calculation are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings to be used in describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive efforts. Figure 1 A schematic flow chart of a lane avoidance method based on constructing an avoidance space provided by an embodiment of the present invention; Figure 2 A schematic diagram of forming an avoidance space provided in an embodiment of the present invention; Figure 3 A schematic diagram of module connections of an avoidance system provided by an embodiment of the present invention.

[0026] Description of reference numerals: 100. Avoidance system; 1. Obstacle avoidance; 2. Avoidance space; 3. Self-vehicle; 4. Reference line; 5. Sampling point; 6. Following target; 10. Detection module; 20. Processing module; 30. Execution module. DETAILED DESCRIPTION

[0027] The embodiments of the present invention will be described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention and are not to be construed as limiting the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0028] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connection, connection, or integral connection; they may refer to mechanical connection or electrical connection; they may refer to direct connection or indirect connection through an intermediate medium; and they may refer to internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0029] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0030] See also Figure 1 , which is a flow chart of a lane avoidance method based on constructing an avoidance space provided by an embodiment of the present invention. Specifically, a lane avoidance method based on constructing an avoidance space includes: S10: Determine the following target of the ego vehicle in the current lane and select obstacles to avoid based on the following target; S20: Sampling the boundary of the obstacle avoidance object close to the reference line and generating a first obstacle data set; S30: dilating the first obstacle data set by a preset dilation value to generate a second obstacle data set; S40: Integrate the second obstacle data set with the lane line and generate an avoidance space; S50: Generate an avoidance trajectory according to the avoidance space.

[0031] Furthermore, compared with the prior art, which requires different offset strategies for the three situations of obstacles on the left, obstacles on the right, and obstacles on both sides, the present method reduces the processing of unnecessary obstacles by determining the following target of the vehicle in the current lane and screening out avoidable obstacles based on the following target. The boundary of the avoidable obstacle near the reference line is sampled to generate a first obstacle data set, which is then expanded by a preset expansion value to generate a second obstacle data set. The expansion ensures that the generated second obstacle data set fully considers the avoidable obstacles in the two adjacent lanes. The second obstacle data set is integrated with the lane line to generate an avoidance space, and an avoidance trajectory is generated based on the avoidance space. By accurately generating the avoidance space and a reasonable avoidance trajectory, the vehicle can avoid obstacles in a timely and accurate manner when encountering obstacles, thereby effectively reducing risks and ensuring the safety of passengers and other road users. The execution logic of this solution is simple and clear, and the execution efficiency is high.

[0032] Preferably, the following target of the ego vehicle in the current lane is determined, and obstacles to be avoided are screened out based on the following target, including: screening out the vehicle closest to the ego vehicle in the current lane and recording it as the following target; screening out vehicles in the adjacent lane whose longitudinal distance is less than the following target and recording them as the first target group; screening out vehicles in the first target group whose lateral distance from the reference line is less than a preset distance and recording them as obstacles to be avoided.

[0033] Furthermore, the lateral distance is the distance between the nearest corner point of the vehicle and the reference line.

[0034] Furthermore, the calculation of the preset distance satisfies the following formula 6: Wherein, Formula 6 is: d set =d safe +w car / 2; Among them, d set is the preset distance, d safe is the safe distance, w caris the width of the vehicle, and the safety distance is a variable positively correlated with the width of the current lane.

[0035] Furthermore, through clear screening steps, the following targets and obstacle avoidance objects are screened out, thereby quickly and accurately screening out the most relevant vehicles, thereby clarifying the obstacle avoidance objects to achieve accurate and effective screening and identification.

[0036] Preferably, sampling is performed on the boundary of the obstacle to be avoided close to the reference line, and a first obstacle data set is generated, including: calculating the longitudinal distance between the following target and the vehicle, and recording it as the first distance; calculating the sampling distance based on the first distance; taking multiple sampling points within the sampling distance; sampling the boundary of the obstacle to be avoided close to the reference line at the sampling points to generate the first obstacle data set.

[0037] Furthermore, the first distance is the longitudinal distance between the closest point of the following target relative to the own vehicle and the own vehicle.

[0038] Furthermore, the sampling points are selected at equal intervals.

[0039] Furthermore, the calculation of the sampling distance satisfies the following formula 7: Wherein, Formula 7 is: S sample =MIN[v*t+1 / 2*(a*t 2 ), S follow ]; Among them, v is the vehicle speed, a is the vehicle acceleration, t is the planning time, S follow is the first distance.

[0040] Furthermore, the first obstacle data set includes a first left-side obstacle avoidance data set and a first right-side obstacle avoidance data set.

[0041] Furthermore, by calculating the first distance between the nearest point of the following target and the vehicle, and calculating the farthest distance of sampling based on the first distance, the appropriate sampling distance and sampling point are determined to achieve more comprehensive and accurate sampling data.

[0042] In one embodiment of the present invention, the k avoidance obstacles in the left lane and the j avoidance obstacles in the right lane selected in step S10 are obtained, and the sampling point set (s a1 ,……,s b1 ),……,(s ak ,……,s bk )、(s c1 ,……,s d1 ),……,(s cj ,……,s dj), and calculate the boundary l of the obstacle avoidance side close to the reference line, and obtain the first left obstacle avoidance data set (l a1 ,……,l b1 ),……,(l ak ,……,l bk ), and the first right obstacle avoidance data set (l c1 ,……,l d1 ),……,(l cj ,……,l dj ).

[0043] Among them, s a1 is the first sampling point of the first obstacle avoidance in the left lane, l a1 is the corresponding sampling value; s b1 is the last sampling point of the first obstacle avoidance in the left lane, l b1 is the corresponding sampling value; s ak is the first sampling point of the kth obstacle avoidance in the left lane, l ak is the corresponding sampling value; s bk is the last sampling point of the kth obstacle avoidance in the left lane, l bk is the corresponding sampling value; s c1 is the first sampling point of the first obstacle avoidance in the right lane, l c1 is the corresponding sampling value; s d1 is the last sampling point of the first obstacle avoidance in the right lane, l d1 is the corresponding sampling value; s cj is the first sampling point of the jth obstacle avoidance in the right lane, l ck is the corresponding sampling value; s dj is the last sampling point of the jth obstacle avoidance in the right lane, l dk The corresponding sampling value.

[0044] Furthermore, the second obstacle data set includes a second left-side obstacle avoidance data set and a second right-side obstacle avoidance data set.

[0045] In one embodiment of the present invention, the first obstacle data set is expanded by a preset expansion value to obtain a second left obstacle avoidance data set (l a1 -e,……,l b1 -e),……,(l ak -e,……,l bk -e), and the second right obstacle avoidance data set (l c1 +e,……,l d1 +e),……,(l ck +e,……,ldk +e).

[0046] Preferably, the second obstacle data set is integrated with the lane line and an avoidance space is generated, including: sampling the lane line at a sampling point to generate a lane line data set; within the same sampling point, the second obstacle data set and the lane line data set are compared, and the data close to the reference line are integrated into an integrated data set; the integrated data set is processed twice to obtain a final data set; and the avoidance space is generated based on the final data set.

[0047] Furthermore, by comparing the lane line data set and the second obstacle data set within the same sampling point to form an integrated data set, and then performing secondary processing on the integrated data set, an avoidance space is generated according to the final data set, thereby improving the accuracy and rationality of the avoidance space generation.

[0048] Furthermore, the avoidance space includes avoiding the left boundary and avoiding the right boundary.

[0049] Preferably, the integrated data set includes a left boundary avoidance data set and a right boundary avoidance data set; the integrated data set is subjected to secondary processing to obtain a final data set, including: subtracting the left boundary avoidance data set from the right boundary avoidance data set to obtain an error data set; adjusting the data in the second obstacle data set according to the error data set, and regenerating the avoidance space.

[0050] In one embodiment of the present invention, within the same sampling point s, the left lane line data set (l l1 ,……,l ln ) and the second left obstacle avoidance data set (l a1 -e,……,l b1 -e),……,(l ak -e,……,l bk -e), take a small value to avoid the left boundary data set ((s 1， l max1 ),……,(s n , l maxn )). Within the same sampling point s, compare the right lane line data set (l r1 ,……,l rn ) and the second right obstacle avoidance data set (l c1 +e,……,l d1 +e),……,(l ck +e,……,l dk +e), take the larger value to get the avoid right boundary data set ((s1, l min1 ),……,(s n , l minn)). Subtract the left boundary avoidance data set from the right boundary avoidance data set to obtain the error data set ((l max1 -l min1 ),……,(l maxn -l minn )).

[0051] Preferably, the data in the second obstacle data set is adjusted according to the error data set, and the avoidance space is regenerated, including: comparing the error data in the error data set with the vehicle width data; if the error data is smaller than the vehicle width data, reducing the preset expansion values of the corresponding sampling points in the second left obstacle avoidance data set and the second right obstacle avoidance data set until the error data is greater than or equal to the vehicle width data; recalculating the second obstacle data set, and regenerating the avoidance space.

[0052] Furthermore, the preset expansion value is greater than or equal to zero, and the vehicle width data = (W car +2*e safe ), where W car is the vehicle width, e safe Vehicle width expansion value.

[0053] Furthermore, by adjusting the preset expansion value, the second obstacle data set and the avoidance space are regenerated, ensuring that the avoidance space is sufficient for the vehicle to pass safely, thereby enhancing the feasibility of the avoidance strategy.

[0054] Preferably, generating an avoidance trajectory according to the avoidance space includes: calculating a cost function according to the avoidance space; calculating boundary conditions according to the boundary constraints; and inputting the cost function and the boundary conditions into a solver to generate the avoidance trajectory.

[0055] Preferably, the cost function includes a reference line difference cost, a first-order derivative cost, a second-order derivative cost, a third-order derivative cost, and an avoidance space center cost; The calculation cost function satisfies the following formula 1; Wherein, Formula 1 is: in, is the cost function, is the reference line cost weight value, is the weight value of the first-order derivative cost, is the weight value of the second-order derivative cost, is the weight value of the third-order derivative cost, To avoid the weight value of the spatial center cost, is the optimal value of the i-th sampling point, is the optimal value of the i+1th sampling point, is the left boundary value of the i-th sampling point, is the right boundary value of the i-th sampling point.

[0056] Furthermore, by comprehensively considering the reference line cost, each order derivative cost and the avoidance space center cost and assigning corresponding weight values, the accuracy and comprehensiveness of the cost function calculation are improved.

[0057] Preferably, calculating the boundary conditions according to the boundary constraints includes: calculating four corner points of the ego vehicle and performing approximation processing to generate ego vehicle data; generating the boundary conditions according to the ego vehicle data; The generated vehicle data satisfies the following formulas 2, 3, 4, and 5: Wherein, Formula 2 is: ; Formula 3 is: ; Formula 4 is: ; Formula 5 is: ; in, is the first corner value, is the second corner value, is the third corner value, is the fourth corner value, is the distance from the center of the rear axle of the vehicle to the front of the vehicle, is the distance from the center of the rear axle to the rear of the vehicle, is the width of the vehicle.

[0058] Furthermore, the boundary conditions are finally obtained as follows: ; ; ; .

[0059] in, for in The minimum value of for in The maximum value of is the longitudinal distance between the i-th sampling point and the vehicle.

[0060] Furthermore, by calculating the four corner points of the ego vehicle and performing approximate processing, more accurate ego vehicle data is generated, which improves the accuracy of boundary condition calculation.

[0061] On the other hand, see Figures 2 to 3An embodiment of the present invention further provides an avoidance system 100 capable of implementing the lane avoidance method of the above embodiment. The avoidance system includes a detection module 10, a processing module 20, and an execution module 30. The detection module 10 is configured to determine a following target 6 of the ego-vehicle 3 in the current lane and identify an avoidance obstacle 1 based on the following target 6. The detection module 10 is further configured to sample the boundary of the avoidance obstacle 1 near the reference line 4 and generate a first obstacle data set. The processing module 20 is configured to expand the first obstacle data set by a preset expansion value and generate a second obstacle data set. The processing module 20 is further configured to integrate the second obstacle data set with the lane line and generate an avoidance space 2. The execution module 30 is configured to generate an avoidance trajectory based on the avoidance space 2.

[0062] Although the present invention is disclosed as above, the present invention is not limited thereto. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention should be based on the scope defined by the claims.

Claims

1. A lane avoidance method based on constructing an avoidance space, characterized in that: include: Determine the following target of the ego vehicle in the current lane, and select obstacles to avoid based on the following target; Sampling a boundary of the obstacle avoidance object close to the reference line and generating a first obstacle data set; Expanding the first obstacle data set by a preset expansion value to generate a second obstacle data set; Integrating the second obstacle data set with the lane line to generate an avoidance space; An avoidance trajectory is generated according to the avoidance space.

2. The lane avoidance method according to claim 1, characterized in that: Determining a following target of the ego vehicle in the current lane and selecting an obstacle to avoid based on the following target includes: Filter out the vehicle closest to the ego vehicle in the current lane and record it as the following target; Filter out the vehicles in the adjacent lanes whose longitudinal distance is smaller than the following target, and record them as the first target group; The vehicles in the first target group whose lateral distances from the vehicle to the reference line are less than a preset distance are screened out and recorded as the avoidance obstacles.

3. The lane avoidance method according to claim 1, characterized in that: The step of sampling the boundary of the obstacle avoidance object close to the reference line and generating a first obstacle data set includes: Calculating the longitudinal distance between the following target and the own vehicle and recording it as a first distance; Calculating a sampling distance according to the first distance; Taking a plurality of sampling points within the sampling distance; The boundary of the obstacle to be avoided close to the reference line is sampled at the sampling point to generate the first obstacle data set.

4. The lane avoidance method according to claim 3, characterized in that: The step of integrating the second obstacle data set with the lane line and generating an avoidance space includes: Sampling the lane line at the sampling point to generate a lane line data set; At the same sampling point, the second obstacle data set and the lane line data set are compared, and the data close to the reference line are integrated into an integrated data set; Performing secondary processing on the integrated data set to obtain a final data set; The avoidance space is generated according to the final data set.

5. The lane avoidance method according to claim 4, characterized in that: The integrated data set includes a left-edge avoidance data set and a right-edge avoidance data set; Performing secondary processing on the integrated data set to obtain a final data set includes: Subtracting the avoidance left boundary data set from the avoidance right boundary data set to obtain an error data set; The data in the second obstacle data set is adjusted according to the error data set, and the avoidance space is regenerated.

6. The lane avoidance method according to claim 5, characterized in that: The second obstacle data set includes a second left obstacle avoidance data set and a second right obstacle avoidance data set; Adjusting the data in the second obstacle data set according to the error data set and regenerating the avoidance space includes: Comparing the error data in the error data set with the vehicle width data; If the error data is smaller than the vehicle width data, reducing the preset expansion values corresponding to the sampling points in the second left obstacle avoidance data set and the second right obstacle avoidance data set until the error data is greater than or equal to the vehicle width data; The second obstacle data set is recalculated, and the avoidance space is regenerated.

7. The lane avoidance method according to claim 1, characterized in that: Generating an avoidance trajectory according to the avoidance space includes: Calculating a cost function based on the avoidance space; calculating the boundary conditions based on the boundary constraints; The cost function and the boundary conditions are input into a solver to generate the avoidance trajectory.

8. The lane avoidance method according to claim 7, characterized in that: The cost function includes a reference line difference cost, a first-order derivative cost, a second-order derivative cost, a third-order derivative cost, and an avoidance space center cost; The cost function is calculated to satisfy the following formula 1: Wherein, the formula 1 is: in, is the cost function, is the reference line cost weight value, is the weight value of the first-order derivative cost, is the weight value of the second-order derivative cost, is the weight value of the third-order derivative cost, is the weight value of the avoidance space center cost, is the optimal value of the i-th sampling point, is the optimal value of the i+1th sampling point, is the left boundary value of the i-th sampling point, is the right boundary value of the i-th sampling point.

9. The lane avoidance method according to claim 7, characterized in that: The boundary conditions are calculated based on the boundary constraints, including: Calculating four corner points of the ego vehicle and performing approximation processing to generate ego vehicle data; generating boundary conditions based on the ego-vehicle data; The generated vehicle data satisfies the following formulas 2, 3, 4, and 5: Wherein, the formula 2 is: ; The formula 3 is: ; The formula 4 is: ; The formula 5 is: ; in, is the first corner value, is the second corner value, is the third corner value, is the fourth corner value, is the distance from the rear axle center of the vehicle to the front of the vehicle, is the distance from the center of the rear axle to the rear of the vehicle, is the width of the vehicle.

10. An avoidance system, characterized in that: The lane avoidance method according to any one of claims 1 to 9 can be implemented, wherein the avoidance system comprises: a detection module, the detection module being configured to determine the following target of the ego vehicle in the current lane and to filter out the avoidable obstacle based on the following target; The detection module is further configured to sample the boundary of the obstacle to be avoided close to the reference line and generate the first obstacle data set; a processing module, configured to expand the first obstacle data set by a preset expansion value and generate the second obstacle data set; The processing module is further configured to integrate the second obstacle data set with the lane line and generate the avoidance space; An execution module is used to generate the avoidance trajectory according to the avoidance space.

Images