Automatic driving automobile obstacle avoiding method and device and storage medium

Abstract

The invention discloses an automatic driving automobile obstacle avoiding method and device and a storage medium. According to the invention, the front obstacle information is obtained through the laser radar carried by the vehicle, and the first feasible region is delimited by combining the centimeter-level high-precision map and the original trajectory of the vehicle. Judging whether obstacle avoidance is needed or not according to the obstacle information and the first feasible region, if the obstacle avoidance is needed, obtaining a state information set through a sampling method, generating an obstacle avoidance trajectory set based on the state information set by applying a quintic polynomial calculation method, further screening the obstacle avoidance trajectory set, determining an optimal trajectory, and determining the optimal trajectory; and controlling the vehicle to execute an obstacle avoidance action along the optimal trajectory, and returning to the original trajectory after obstacle avoidance is completed. By adopting the technical scheme of the invention, the calculation efficiency can be improved, the smooth obstacle avoidance track conforming to the kinematics constraint is generated, and safer and more efficient obstacle avoidance is realized.

Description

Technical Field

[0001] This application relates to the field of autonomous driving, and particularly to a method, device, and storage medium for an autonomous vehicle to avoid obstacles. Background Art

[0002] Currently, autonomous vehicles highly rely on multi-sensor fusion environmental perception systems such as lidar and cameras during obstacle avoidance. However, there are still significant defects in the existing technologies. On the one hand, path planning lacks in-depth integration of high-precision map information and only relies on real-time perception data to generate local paths, resulting in blurred boundaries of drivable areas and potentially planning reverse trajectories or entering non-drivable areas by mistake. On the other hand, traditional polynomial trajectory planning requires handling collision detection and parameter optimization of a large number of candidate trajectories, consuming a large amount of computing resources and being difficult to meet real-time requirements. Moreover, the generated trajectories have poor smoothness, cannot meet vehicle kinematic constraints, and are difficult to ensure the smoothness and safety of the obstacle avoidance process.

[0003] Therefore, how to improve the computing efficiency and generate smooth and kinematically constrained obstacle avoidance trajectories on the premise of ensuring the safety and reliability of the autonomous vehicle's obstacle avoidance process is a technical problem that needs to be solved currently.

[0004] Application Content

[0005] This application provides a method, device, and storage medium for an autonomous vehicle to avoid obstacles, which can achieve safer and more efficient obstacle avoidance by means of a high-precision map.

[0006] In a first aspect, this application provides a method for an autonomous vehicle to avoid obstacles, including:

[0007] Obtaining obstacle information in front of the vehicle through a lidar mounted on the vehicle, and demarcating a first feasible area according to a centimeter-level high-precision map and the original trajectory of the vehicle;

[0008] Judging whether obstacle avoidance is required according to the obstacle information and the first feasible area;

[0009] If obstacle avoidance is required, obtaining a set of state information through a sampling method, generating a set of obstacle avoidance trajectories using a quintic polynomial calculation method based on the set of state information, and then screening the set of obstacle avoidance trajectories to determine the optimal trajectory;

[0010] Controlling the vehicle to perform an obstacle avoidance action along the optimal trajectory and returning to the original trajectory after completing the obstacle avoidance.

[0011] Compared with the prior art, the embodiments of the present application have the following beneficial effects: By obtaining obstacle information in front of the vehicle through the lidar mounted on the vehicle, and combining the centimeter-level high-precision map and the original vehicle trajectory to delimit the feasible region, the position of the obstacle and the feasible region can be identified, providing a basis for obstacle avoidance judgment. Based on this, it is further determined whether obstacle avoidance is required according to the obstacle information and the feasible region, thus avoiding unnecessary obstacle avoidance actions. By using the sampling method to obtain the state information set, and generating the obstacle avoidance trajectory set by using the fifth-order polynomial calculation method based on the state information set, and further screening out the optimal trajectory, a smooth trajectory that conforms to the vehicle kinematic constraints can be generated quickly. Finally, the vehicle is controlled to execute the obstacle avoidance action along the optimal trajectory, and returns to the original trajectory after the obstacle avoidance is completed, ensuring that the vehicle can smoothly return to the original route after the obstacle avoidance and maintain a normal driving state. Compared with the prior art, the present application can improve the calculation efficiency, generate a smooth obstacle avoidance trajectory that conforms to the kinematic constraints, and thus achieve safer and more efficient obstacle avoidance.

[0012] Further, the obtaining the state information set by the sampling method is specifically as follows:

[0013] Translate the original trajectory in the obstacle avoidance direction to obtain an obstacle avoidance reference trajectory, and sample at multiple preset sampling points on the obstacle avoidance reference trajectory to generate an end state set;

[0014] Combine the starting state of the vehicle and the end state set to generate a state information set.

[0015] Compared with the prior art, the above embodiments have the following beneficial effects: By setting multiple longitudinal sampling points on the obstacle avoidance reference trajectory to generate various end states, the obstacle avoidance requirements under different conditions can be met. This flexible sampling method enables the vehicle to quickly find a suitable obstacle avoidance end point when facing obstacles of different sizes and positions, enhancing the vehicle's adaptability in complex traffic environments.

[0016] Further, the generating the obstacle avoidance trajectory set by using the fifth-order polynomial calculation method based on the state information set is specifically as follows:

[0017] Obtain the trajectory respectively according to the horizontal and vertical dimensions. The vertical trajectory is represented by the fifth-order polynomial s(t)=c0 + c1t + c2t 2 + c3t 3 + c4t 4 + c5t 5 The horizontal trajectory is represented by the fifth-order polynomial d(t)=a0 + a1t + a2t 2 + a3t 3 + a4t 4 + a5t 5It is represented that, where s(t) represents the longitudinal distance of the vehicle at time t, d(t) represents the lateral distance of the vehicle at time t, t is a time variable, and (c0, c1, c2, c3, c4, c5), (a0, a1, a2, a3, a4, a5) are the polynomial coefficients to be solved;

[0018] Set the initial state conditions and termination state conditions of the longitudinal trajectory and the lateral trajectory respectively according to the state information set, solve the polynomial coefficients of the longitudinal trajectory and the lateral trajectory, and generate an obstacle avoidance trajectory set.

[0019] Compared with the prior art, the above embodiments have the following beneficial effects: By using a fifth-degree polynomial to represent the trajectory in both the lateral and longitudinal dimensions respectively, a smooth and continuous obstacle avoidance trajectory can be generated. This representation method ensures the continuity of the trajectory in time and space, avoids sudden acceleration or speed during the obstacle avoidance process of the vehicle, and improves the smoothness of the obstacle avoidance process.

[0020] Further, the screening of the obstacle avoidance trajectory set to determine the optimal trajectory is specifically as follows:

[0021] Screen the obstacle avoidance trajectory set, eliminate the trajectories whose trajectory curvature range, speed range, acceleration range, and acceleration change rate range do not meet the requirements, and obtain a first trajectory set;

[0022] Taking the tangent direction of each trajectory in the first trajectory set as the horizontal axis and the normal direction of each trajectory in the first trajectory set as the vertical axis, establish a corresponding first Frenet coordinate system for each trajectory in the first trajectory set;

[0023] Obtain the lateral distance of the obstacle in each first Frenet coordinate system, and eliminate the trajectories with a lateral distance less than a preset threshold from the first trajectory set to obtain a second trajectory set; Calculate the maximum acceleration change rate of all trajectories in the second trajectory set, and take the trajectory with the minimum maximum acceleration change rate as the optimal trajectory.

[0024] Compared with the prior art, the above embodiments have the following beneficial effects: By performing multi-dimensional screening on the obstacle avoidance trajectory set, eliminating the trajectories that do not meet the requirements of trajectory curvature, speed, acceleration, and acceleration change rate, unsafe trajectories can be eliminated, the collision risk can be avoided, and the obstacle avoidance path with the best smoothness can be selected, improving the safety of autonomous vehicles in complex environments and meeting the real-time requirements at the same time.

[0025] Further, the judgment of whether obstacle avoidance is required according to the obstacle information and the first feasible region is specifically as follows:

[0026] Taking the direction of the road center line as the longitudinal axis and the direction perpendicular to the road center line as the transverse axis, a second Frenet coordinate system is established;

[0027] Calculate a first lateral offset required for the vehicle to bypass the obstacle from the left and a second lateral offset required for the vehicle to bypass the obstacle from the right;

[0028] According to the obstacle information, determine a first distance from the closest point of the obstacle to the vehicle along the original trajectory and a second distance from the farthest point of the obstacle to the vehicle along the original trajectory;

[0029] Based on the first distance, the second distance, the first lateral offset, the second lateral offset, the second Frenet coordinate system, and the first feasible region, determine whether obstacle bypassing is required.

[0030] Compared with the prior art, the above embodiments have the following beneficial effects: By establishing a Frenet coordinate system, calculating the lateral offset and the distance to the obstacle, and making a comprehensive judgment in combination with the feasible region, it is possible to accurately determine whether the vehicle needs to bypass the obstacle.

[0031] Further, the specific method for delimiting the first feasible region according to the centimeter-level high-precision map and the original trajectory is as follows:

[0032] Obtain the lane information on both sides of the original trajectory from the high-precision map and combine it into the first feasible region;

[0033] Obtain the lane direction based on the oncoming lane given by the high-precision map. When the angle between the lane direction and the direction of the original trajectory is greater than a preset angle, determine whether it is possible to cross the oncoming lane according to the middle lane line of two lanes with opposite directions;

[0034] If the middle lane line is a double yellow line, a single solid yellow line, or has a railing, exclude the oncoming lane where crossing is not allowed from the first feasible region to obtain the first feasible region;

[0035] Otherwise, use the second feasible region as the first feasible region.

[0036] Compared with the prior art, the above embodiments have the following beneficial effects: By obtaining lane information from the high-precision map and combining the lane line type to judge the passability of the oncoming lane, excluding non-passable areas, thereby determining the feasible region of the vehicle, effectively improving the safety and calculation efficiency of path planning.

[0037] In a second aspect, the present application provides an obstacle bypass device for an autonomous vehicle, including: a data acquisition module, an obstacle bypass judgment module, an obstacle bypass trajectory generation module, and an obstacle bypass trajectory execution module;

[0038] The data acquisition module is used to obtain obstacle information in front of the vehicle through a lidar mounted on the vehicle, and delimit a feasible area according to a centimeter-level high-precision map and the original trajectory of the vehicle;

[0039] The obstacle avoidance judgment module is used to judge whether obstacle avoidance is needed according to the obstacle information and the first feasible area;

[0040] The obstacle avoidance trajectory generation module is used to, if obstacle avoidance is needed, obtain a set of state information through a sampling method, generate a set of obstacle avoidance trajectories based on the set of state information by using a fifth-degree polynomial calculation method, and then screen the set of obstacle avoidance trajectories to determine the optimal trajectory;

[0041] The obstacle avoidance execution module is used to control the vehicle to perform an obstacle avoidance action along the optimal trajectory and return to the original trajectory after the obstacle avoidance is completed.

[0042] Further, the obstacle avoidance trajectory generation module includes: a state sampling unit and a state information generation unit;

[0043] The state sampling unit is used to translate the original trajectory in the obstacle avoidance direction to obtain an obstacle avoidance reference trajectory, sample at multiple preset sampling points on the obstacle avoidance reference trajectory, and generate an end state set;

[0044] The state information generation unit is used to combine the starting state of the vehicle and the end state set to generate a set of state information.

[0045] Further, the obstacle avoidance trajectory generation module further includes: a trajectory generation unit;

[0046] The trajectory generation unit is used to obtain trajectories respectively according to two dimensions of transverse and longitudinal. The longitudinal trajectory is represented by the fifth-degree polynomial s(t)=c0 + c1t + c2t 2 +c3t 3 +c4t 4 +c5t 5 The transverse trajectory is represented by the fifth-degree polynomial d(t)=a0 + a1t + a2t 2 +a3t 3 +a4t 4 +a5t 5 where s(t) represents the longitudinal distance of the vehicle at time t, d(t) represents the transverse distance of the vehicle at time t, t is a time variable, and (c0, c1, c2, c3, c4, c5), (a0, a1, a2, a3, a4, a5) are polynomial coefficients to be solved;

[0047] The trajectory generation unit is further configured to respectively set the initial state conditions and the termination state conditions of the longitudinal trajectory and the lateral trajectory according to the state information set, solve the polynomial coefficients of the longitudinal trajectory and the lateral trajectory, and generate an obstacle avoidance trajectory set.

[0048] Further, the obstacle avoidance trajectory generation module further includes a trajectory screening unit;

[0049] The trajectory screening unit is configured to screen the obstacle avoidance trajectory set, and eliminate the trajectories whose trajectory curvature range, speed range, acceleration range, and acceleration change rate range do not meet the requirements, so as to obtain a first trajectory set;

[0050] Taking the tangent direction of each trajectory in the first trajectory set as the horizontal axis and the normal direction of each trajectory in the first trajectory set as the vertical axis, a corresponding first Frenet coordinate system is established for each trajectory in the first trajectory set;

[0051] Obtain the lateral distance of the obstacle in each first Frenet coordinate system, and eliminate the trajectories with a lateral distance less than a preset threshold from the first trajectory set to obtain a second trajectory set;

[0052] Calculate the maximum acceleration change rate of all the trajectories in the second trajectory set, and take the trajectory with the minimum maximum acceleration change rate as the optimal trajectory.

[0053] Further, the obstacle avoidance judgment module includes a coordinate system establishment unit, a lateral offset calculation unit, an obstacle distance determination unit, and an obstacle avoidance judgment unit;

[0054] The coordinate system establishment unit is configured to establish a second Frenet coordinate system with the direction of the road center line as the vertical axis and the direction perpendicular to the road center line as the horizontal axis;

[0055] The lateral offset calculation unit is configured to calculate a first lateral offset required for the vehicle to avoid the obstacle from the left and a second lateral offset required for the vehicle to avoid the obstacle from the right;

[0056] The obstacle distance determination unit is configured to determine, according to the obstacle information, a first distance from the closest point of the obstacle to the vehicle along the original trajectory and a second distance from the farthest point of the obstacle to the vehicle along the original trajectory;

[0057] The obstacle avoidance judgment unit is configured to judge whether obstacle avoidance is required according to the first distance, the second distance, the first lateral offset, the second lateral offset, the second Frenet coordinate system, and the first feasible region.

[0058] Further, the data acquisition module includes a lane information acquisition unit, a lane direction determination unit, and a feasible area determination unit; the lane information acquisition unit is configured to acquire lane information on both sides of the original trajectory from the high-precision map and combine it into a second feasible area;

[0059] The lane direction determination unit is configured to obtain the lane direction based on the oncoming lanes given in the high-precision map. When the angle between the lane direction and the direction of the original trajectory is greater than a preset angle, it is determined whether it is possible to cross the oncoming lane according to the middle lane line of two lanes with opposite directions;

[0060] The feasible area determination unit is configured to, if the middle lane line is a double yellow line, a single solid yellow line, or has a railing, exclude the oncoming lane that is not allowed to cross in the second feasible area to obtain the feasible area; otherwise, use the second feasible area as the first feasible area.

[0061] In a third aspect, the present application also provides a computer storage medium, characterized in that a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, it implements an obstacle avoidance method for an autonomous vehicle according to any one of claims 1 to 6. Description of the Drawings

[0062] Figure 1 It is a schematic flow chart of an obstacle avoidance method for an autonomous vehicle provided in some embodiments of the present application;

[0063] Figure 2 It is a schematic diagram of the radar distribution of a vehicle in some embodiments provided by the present application;

[0064] Figure 3 It is a schematic diagram of the drivable area in some embodiments provided by the present application;

[0065] Figure 4 It is a schematic diagram of the obstacle avoidance trajectory planning in some embodiments provided by the present application;

[0066] Figure 5 It is a schematic structural diagram of an obstacle avoidance device for an autonomous vehicle provided in some embodiments of the present application. Detailed Embodiments

[0067] Current autonomous vehicles highly rely on multi-sensor fusion environment perception systems such as lidar and cameras during obstacle avoidance. However, there are still significant defects in the existing technologies. On the one hand, path planning lacks in-depth integration of high-precision map information and only relies on real-time perception data to generate local paths, resulting in blurred boundaries of drivable areas and potentially planning reverse trajectories or entering non-drivable areas by mistake. On the other hand, traditional polynomial trajectory planning requires handling collision detection and parameter optimization of a large number of candidate trajectories, consuming a large amount of computing resources and being difficult to meet real-time requirements. Moreover, the generated trajectories have poor smoothness, cannot meet vehicle kinematic constraints, and are difficult to ensure the smoothness and safety during obstacle avoidance.

[0068] To solve the above problems, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0069] Embodiment 1

[0070] Please refer to Figure 1 , a method for obstacle avoidance of an autonomous vehicle provided in an embodiment of the present application, including S10 to S30, specifically:

[0071] S10: Obtain obstacle information in front of the vehicle through the lidar carried by the vehicle, and delimit a first feasible area according to the centimeter-level high-precision map and the original trajectory of the vehicle.

[0072] Further, in some embodiments of the present application, the delimiting of the first feasible area according to the centimeter-level high-precision map and the original trajectory is specifically:

[0073] Obtain lane information on both sides of the original trajectory from the high-precision map and combine it into a second feasible area; obtain the lane direction according to the oncoming lane given by the high-precision map. When the direction angle between the lane direction and the direction of the original trajectory is greater than a preset angle, judge whether it is possible to cross the oncoming lane according to the middle lane line of the two lanes with opposite directions; if the middle lane line is a double yellow line, a single solid yellow line or has a railing, exclude the oncoming lane that does not allow crossing in the second feasible area to obtain the first feasible area; otherwise, use the second feasible area as the first feasible area. Next, in combination with Figure 2 and Figure 3 , step S10 will be further described in detail. Step S10 can be executed through the following preferred implementation:

[0074] In this embodiment, obstacle information in front of the vehicle is obtained through the lidar carried by the vehicle. Specifically, asFigure 2 As shown, the vehicle is equipped with three lidars, including a main lidar and two auxiliary lidars. The main lidar is installed at the front side near the top of the vehicle and can detect obstacles within 50 meters around the vehicle in 360 degrees. Its function is to provide detailed information about obstacles in the distance ahead of the vehicle, help the vehicle perceive potential collision risks in advance, and provide sufficient time and space for path planning. The two auxiliary single-line lidars are installed near the right front wheel and the left rear wheel of the vehicle respectively. Their main function is to sense low obstacles within the fan-shaped areas around the vehicle and supplement the blind spot detection of the main lidar near the vehicle. These low obstacles may include curbs, small obstacles, etc., which are easily ignored by the main lidar but pose potential threats to the driving safety of the vehicle. The main lidar and the auxiliary lidars complement each other to jointly provide 360-degree omnidirectional obstacle detection for the vehicle. The obstacle information obtained by the lidars includes the position, size, shape of the obstacles, as well as the speed and direction relative to the vehicle, etc. These information provide a basis for subsequent path planning and obstacle avoidance decisions.

[0075] Figure 3 It shows how to obtain lane information on both sides of the original trajectory from the high-precision map and combine them into the first feasible region (the orange part). When the angle between the oncoming lane direction and the original trajectory direction is greater than 120 degrees, it is considered that the oncoming lane is in the opposite direction to the vehicle driving trajectory. For lanes in the opposite direction, further check the type of the middle lane line. If the middle lane line is a double yellow line, a single solid yellow line or has a railing, these oncoming lanes are excluded, and finally the feasible region (the green region) is obtained.

[0076] Obtain lane information through the high-precision map and combine the lane line type to judge the passability of the oncoming lane, exclude the non-passable areas, so as to determine the feasible region of the vehicle, effectively improving the safety and calculation efficiency of path planning.

[0077] S20: Judge whether obstacle avoidance is needed according to the obstacle information and the first feasible region.

[0078] Furthermore, in some embodiments of the present application, the judging whether obstacle avoidance is needed according to the obstacle information and the feasible region is specifically as follows: establish a second Frenet coordinate system with the direction of the road center line as the longitudinal axis and the direction perpendicular to the road center line as the transverse axis; calculate the first lateral offset required for the vehicle to avoid obstacles from the left and the second lateral offset required for the vehicle to avoid obstacles from the right; according to the obstacle information, determine the first distance from the closest point of the obstacle to the vehicle along the original trajectory and the second distance from the farthest point of the obstacle to the vehicle along the original trajectory; judge whether obstacle avoidance is needed according to the first distance, the second distance, the first lateral offset, the second lateral offset, the second Frenet coordinate system and the feasible region.

[0079] Specifically, step S20 can be executed through the following preferred embodiments:

[0080] Calculate the first lateral offset d required for the vehicle to bypass the obstacle from the left left and the second lateral offset d required for the vehicle to bypass the obstacle from the right right , and the specific calculation method is as follows:

[0081] d left = d max + d safe

[0082] d right = d min + d safe

[0083] where d min is the minimum lateral distance of the rightmost point of the obstacle, d max is the maximum lateral distance of the leftmost point of the obstacle, and d safe is the reserved safety distance.

[0084] According to the obstacle information, determine the first distance s from the vehicle to the nearest point of the obstacle along the original trajectory min and the second distance s from the vehicle to the farthest point of the obstacle along the original trajectory max .

[0085] Combining the first distance s min , the second distance s max , the first lateral offset d left , the second lateral offset d right , the Frenet coordinate system, and the second feasible region, determine whether obstacle bypassing is required. The specific judgment logic is as follows:

[0086] If the point (s min , d right ) and the point (s max , d right ) are within the feasible region, then the vehicle can bypass the obstacle from the right; if the point (s min , d left ) and the point (s max , d left ) are within the feasible region, then the vehicle can bypass the obstacle from the left; otherwise, obstacle bypassing is not possible.

[0087] By establishing a Frenet coordinate system, calculating the lateral offset and the distance to the obstacle, and making a comprehensive judgment in combination with the feasible region, it is possible to accurately determine whether the vehicle needs to bypass the obstacle.

[0088] S30: If obstacle avoidance is required, obtain a set of state information through a sampling method. Based on the set of state information, use the quintic polynomial calculation method to generate an obstacle avoidance trajectory set, and screen the obstacle avoidance trajectory set to select the optimal trajectory.

[0089] Further, in some embodiments of the present application, the obtaining of the set of state information through the sampling method is specifically as follows: Translate the original trajectory in the obstacle avoidance direction to obtain an obstacle avoidance reference trajectory, sample at multiple preset sampling points on the obstacle avoidance reference trajectory to generate an end state set; combine the starting state of the vehicle and the end state set to generate a set of state information. By setting multiple longitudinal sampling points on the obstacle avoidance reference trajectory to generate multiple end states, it can meet the obstacle avoidance requirements under different conditions. This flexible sampling method enables the vehicle to quickly find a suitable obstacle avoidance end point when facing obstacles of different sizes and positions, enhancing the vehicle's adaptability in complex traffic environments.

[0090] Specifically, this step can be executed through the following preferred implementation manner:

[0091] As Figure 4 shown, assuming that the vehicle bypasses the obstacle to the left, translate the original reference trajectory to the left to obtain the obstacle avoidance reference trajectory, and the translation amount d trans is:

[0092] d trans = d left -(0.5 * d carWidth )

[0093] where d left is the maximum horizontal distance of the leftmost point of the obstacle, and d carWidth is the vehicle width.

[0094] Then select multiple preset sampling points on the obstacle avoidance reference trajectory. The specific longitudinal distances of the sampling points are 5 meters, 7 meters, 9 meters, 11 meters, 14 meters, and 20 meters. For each sampling point, determine the end state of the vehicle at that point, including position, speed, and acceleration. Through the above method, generate an end state set containing 6 sampling points, and then combine the starting state of the vehicle (including the initial position, speed, and acceleration) with the end state set to form a set of state information. The set of state information contains 6 * 1 = 6 elements.

[0095] Further, in some embodiments of the present application, the generating of the obstacle avoidance trajectory set by using the quintic polynomial calculation method based on the set of state information is specifically as follows: Obtain the trajectory according to two dimensions of horizontal and vertical. The longitudinal trajectory uses the quintic polynomial s(t) = c0 + c1t + c2t 2 + c3t 3 + c4t 4+c5t 5 It is represented that the lateral trajectory is represented by a fifth-degree polynomial d(t)=a0+a1t+a2t 2 +a3t 3 +a4t 4 +a5t 5 where s(t) represents the longitudinal distance of the vehicle at time t, d(t) represents the lateral distance of the vehicle at time t, t is the time variable, (c0, c1, c2, c3, c4, c5), (a0, a1, a2, a3, a4, a5) are the polynomial coefficients to be solved; according to the state information set, the initial state conditions and the termination state conditions of the longitudinal trajectory and the lateral trajectory are respectively set, and the polynomial coefficients of the longitudinal trajectory and the lateral trajectory are solved to generate an obstacle avoidance trajectory set. By using a fifth-degree polynomial to represent the trajectory in both the lateral and longitudinal dimensions respectively, a smooth and continuous obstacle avoidance trajectory can be generated. This representation method ensures the continuity of the trajectory in time and space, avoids sudden acceleration or speed during the obstacle avoidance process of the vehicle, and improves the smoothness of the obstacle avoidance process.

[0096] Further, in some embodiments of the present application, the screening of the obstacle avoidance trajectory set to determine the optimal trajectory is specifically as follows: screening the obstacle avoidance trajectory set, removing the trajectories whose trajectory curvature range, speed range, acceleration range, and acceleration change rate range do not meet the requirements to obtain a first trajectory set; screening the obstacle avoidance trajectory set, removing the trajectories whose trajectory curvature range, speed range, acceleration range, and acceleration change rate range do not meet the requirements to obtain a first trajectory set; taking the tangent direction of each trajectory in the first trajectory set as the horizontal axis and the normal direction of each trajectory in the first trajectory set as the vertical axis, and respectively establishing a corresponding first Frenet coordinate system for each trajectory in the first trajectory set; obtaining the lateral distance of the obstacle in each first Frenet coordinate system, and removing the trajectories with a lateral distance less than a preset threshold from the first trajectory set to obtain a second trajectory set; calculating the maximum acceleration change rate of all trajectories in the second trajectory set, and taking the trajectory with the minimum maximum acceleration change rate as the optimal trajectory. By performing multi-dimensional screening on the obstacle avoidance trajectory set and removing the trajectories that do not meet the requirements of trajectory curvature, speed, acceleration, and acceleration change rate, unsafe trajectories can be removed, collision risks can be avoided, and the obstacle avoidance path with the best smoothness can be selected, improving the safety of autonomous vehicles in complex environments and meeting the real-time requirements at the same time.

[0097] Specifically, removing the trajectories whose trajectory curvature range, speed range, acceleration range, and acceleration change rate range do not meet the requirements can be performed through the following preferred implementation manners:

[0098] The absolute value of the curvature at any point on the trajectory must be less than Where R is the turning radius of the vehicle, the speed at any point on the trajectory is in the range of 0 to 4 m / s, and the acceleration at any point on the trajectory is in the range of -2 to 2 m / s 2 , and the acceleration change rate at any point on the trajectory must be in the range of -0.2 to 0.2 m / s 3 .

[0099] S40: Control the vehicle to perform an obstacle avoidance action along the optimal trajectory and return to the original trajectory after completing the obstacle avoidance.

[0100] In summary, it can be seen that an obstacle avoidance method for an autonomous vehicle provided by this embodiment has the following beneficial effects: By obtaining obstacle information in front of the vehicle through a lidar mounted on the vehicle and combining a centimeter-level high-precision map and the original trajectory of the vehicle to delimit a feasible area, the position of the obstacle and the feasible area can be identified, providing a basis for obstacle avoidance judgment. Based on this, further determine whether obstacle avoidance is required according to the obstacle information and the feasible area, thus avoiding unnecessary obstacle avoidance actions. By using a sampling method to obtain a set of state information and generating an obstacle avoidance trajectory set based on the set of state information using a fifth-degree polynomial calculation method, and further screening out the optimal trajectory, a smooth trajectory that conforms to the vehicle kinematic constraints can be generated quickly. Finally, control the vehicle to perform an obstacle avoidance action along the optimal trajectory and return to the original trajectory after completing the obstacle avoidance, ensuring that the vehicle can smoothly return to the original route and maintain a normal driving state after completing the obstacle avoidance. Compared with the prior art, the present application can improve the calculation efficiency, generate a smooth obstacle avoidance trajectory that conforms to the kinematic constraints, and thus achieve safer and more efficient obstacle avoidance.

[0101] Embodiment 2

[0102] Reference Figure 5 , an obstacle avoidance device for an autonomous vehicle provided by an embodiment of the present application, includes: a data acquisition module 11, an obstacle avoidance judgment module 12, an obstacle avoidance trajectory generation module 13, and an obstacle avoidance trajectory execution module 14.

[0103] Further, in some embodiments of the present application, the data acquisition module 11 is configured to obtain the obstacle information in front of the vehicle through a lidar mounted on the vehicle and delimit a first feasible area according to a centimeter-level high-precision map and the original trajectory of the vehicle; the obstacle avoidance judgment module 12 is configured to judge whether obstacle avoidance is required according to the obstacle information and the first feasible area; the obstacle avoidance trajectory generation module 13 is configured to, if obstacle avoidance is required, obtain a set of state information through a sampling method and generate an obstacle avoidance trajectory set based on the set of state information using a fifth-degree polynomial calculation method, and then screen the obstacle avoidance trajectory set to determine the optimal trajectory; the obstacle avoidance execution module 14 is configured to control the vehicle to perform an obstacle avoidance action along the optimal trajectory and return to the original trajectory after completing the obstacle avoidance.

[0104] Further, in some embodiments of the present application, the obstacle avoidance trajectory generation module 13 obtains a set of state information through a sampling method, specifically including: a state sampling unit and a state information generation unit; the state sampling unit is used to translate the original trajectory in the obstacle avoidance direction to obtain an obstacle avoidance reference trajectory, sample points at multiple preset sampling points on the obstacle avoidance reference trajectory, and generate an end state set; the state information generation unit is used to combine the starting state of the vehicle and the end state set to generate a set of state information.

[0105] Further, in some embodiments of the present application, the obstacle avoidance trajectory generation module 13 generates an obstacle avoidance trajectory set based on the set of state information by using a quintic polynomial calculation method, and further includes: a trajectory generation unit; the trajectory generation unit is used to obtain trajectories respectively according to two dimensions of horizontal and vertical. The longitudinal trajectory is represented by the quintic polynomial s(t)=c0 + c1t + c2t 2 + c3t 3 + c4t 4 + c5t 5 The horizontal trajectory is represented by the quintic polynomial d(t)=a0 + a1t + a2t 2 + a3t 3 + a4t 4 + a5t 5 where s(t) represents the longitudinal distance of the vehicle at time t, d(t) represents the horizontal distance of the vehicle at time t, t is a time variable, (c0, c1, c2, c3, c4, c5), (a0, a1, a2, a3, a4, a5) are polynomial coefficients to be solved; the trajectory generation unit is further used to respectively set the initial state conditions and termination state conditions of the longitudinal trajectory and the horizontal trajectory according to the set of state information, solve the polynomial coefficients of the longitudinal trajectory and the horizontal trajectory, and generate an obstacle avoidance trajectory set.

[0106] Further, in some embodiments of the present application, the obstacle avoidance trajectory generation module 13 further includes: a trajectory screening unit; the trajectory screening unit is used to screen the obstacle avoidance trajectory set, eliminate trajectories whose trajectory curvature range, speed range, acceleration range, and acceleration change rate range do not meet the requirements, and obtain a first trajectory set; taking the tangent direction of each trajectory in the first trajectory set as the horizontal axis and the normal direction of each trajectory in the first trajectory set as the vertical axis, respectively establish a corresponding first Frenet coordinate system for each trajectory in the first trajectory set; obtain the horizontal distance of the obstacle in each first Frenet coordinate system, and eliminate the trajectories with a horizontal distance less than a preset threshold from the first trajectory set to obtain a second trajectory set; calculate the maximum acceleration change rate of all trajectories in the second trajectory set, and take the trajectory with the minimum maximum acceleration change rate as the optimal trajectory.

[0107] Further, in some embodiments of the present application, the obstacle avoidance judgment module 12 includes a coordinate system establishment unit, a lateral offset calculation unit, an obstacle distance determination unit, and an obstacle avoidance judgment unit; the coordinate system establishment unit is configured to establish a second Frenet coordinate system with the direction of the road center line as the longitudinal axis and the direction perpendicular to the road center line as the transverse axis; the lateral offset calculation unit is configured to calculate a first lateral offset required for the vehicle to avoid the obstacle from the left and a second lateral offset required for the vehicle to avoid the obstacle from the right; the obstacle distance determination unit is configured to determine, according to the obstacle information, a first distance from the nearest point of the obstacle to the vehicle along the original trajectory and a second distance from the farthest point of the obstacle to the vehicle along the original trajectory; the obstacle avoidance judgment unit is configured to judge whether obstacle avoidance is required according to the first distance, the second distance, the first lateral offset, the second lateral offset, the second Frenet coordinate system, and the first feasible region.

[0108] Further, in some embodiments of the present application, the data acquisition module includes a lane information acquisition unit, a lane direction judgment unit, and a feasible region determination unit 11; the lane information acquisition unit is configured to acquire the lane information on both sides of the original trajectory from the high-precision map and combine it into a second feasible region; the lane direction judgment unit is configured to obtain the lane direction according to the oncoming lane given by the high-precision map, and when the angle between the lane direction and the direction of the original trajectory is greater than a preset angle, judge whether it is possible to cross the oncoming lane according to the middle lane line of the two lanes with opposite directions; the feasible region determination unit is configured to, if the middle lane line is a double yellow line, a single yellow solid line, or a railing, exclude the oncoming lanes that are not allowed to cross in the second feasible region to obtain the feasible region; otherwise, use the second feasible region as the first feasible region.

[0109] In summary, it can be seen that an obstacle avoidance device for an autonomous vehicle provided in this embodiment has the following beneficial effects: By obtaining obstacle information in front of the vehicle through the lidar carried by the vehicle, and combining with a centimeter-level high-precision map and the original trajectory of the vehicle to delimit a feasible area, it can identify the position of the obstacle and the feasible area, providing a basis for obstacle avoidance judgment. Based on this, it further determines whether obstacle avoidance is required according to the obstacle information and the feasible area, thus avoiding unnecessary obstacle avoidance actions. By using a sampling method to obtain a set of state information, and generating an obstacle avoidance trajectory set based on the set of state information using a fifth-degree polynomial calculation method, and further screening out the optimal trajectory, it can quickly generate a smooth trajectory that conforms to the kinematic constraints of the vehicle. Finally, the vehicle is controlled to perform an obstacle avoidance action along the optimal trajectory, and returns to the original trajectory after the obstacle avoidance is completed, ensuring that the vehicle can smoothly return to the original route and maintain a normal driving state after completing the obstacle avoidance. Compared with the prior art, this application can improve the calculation efficiency, generate a smooth obstacle avoidance trajectory that conforms to the kinematic constraints, and thus achieve a safer and more efficient obstacle avoidance.

[0110] The more detailed step flow and working principle of this embodiment can be but are not limited to referring to the relevant records in Embodiment 1.

[0111] Embodiment 3

[0112] Based on the above embodiment of the method for obtaining the router management server address in the private network environment, another embodiment of this application provides a storage medium, where the storage medium includes a stored computer program, and when the computer program runs, it controls the device where the storage medium is located to execute the obstacle avoidance method for an autonomous vehicle according to any embodiment of this application.

[0113] In this embodiment, the above storage medium is a computer-readable storage medium, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device that can carry the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0114] The specific embodiments described above further elaborate on the objective, technical solution, and beneficial effects of the present application. It should be understood that the above are only specific embodiments of the present application and are not used to limit the protection scope of the present application. In particular, for those skilled in the art, any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included within the protection scope of the present application.

Claims

1. A method for autonomous driving vehicle to avoid obstacles, characterized in that: include: Obtain obstacle information in front of the vehicle through a laser radar carried by the vehicle, and define a first feasible area based on a centimeter-level high-precision map and the original trajectory of the vehicle; Determining whether it is necessary to bypass the obstacle according to the obstacle information and the first feasible area; If obstacle avoidance is required, a state information set is obtained through a sampling method, and based on the state information set, a quintic polynomial calculation method is used to generate an obstacle avoidance trajectory set, and then the obstacle avoidance trajectory set is screened to determine the optimal trajectory; The vehicle is controlled to perform an obstacle avoidance action along the optimal trajectory, and returns to the original trajectory after completing the obstacle avoidance action.

2. The method for avoiding obstacles by an autonomous driving vehicle according to claim 1, characterized in that: The state information set is obtained by sampling method, specifically: The original trajectory is translated in the obstacle avoidance direction to obtain an obstacle avoidance reference trajectory, and sampling is performed at multiple preset sampling points on the obstacle avoidance reference trajectory to generate an end state set; The starting state of the vehicle and the ending state set are combined to generate a state information set.

3. The method for avoiding obstacles by an autonomous driving vehicle according to claim 1, characterized in that: Based on the state information set, a fifth-order polynomial calculation method is used to generate an obstacle avoidance trajectory set, specifically: The trajectories are obtained according to the horizontal and vertical dimensions respectively. The vertical trajectory is expressed by the fifth-order polynomial s(t)=c0+c1t+c2t 2 +c3t 3 +c4t 4 +c5t 5 The lateral trajectory is represented by a fifth-order polynomial d(t) = a0+a1t+a2t 2 +a3t 3 +a4t 4 +a5t 5 Represents, where s(t) represents the longitudinal distance of the vehicle at time t, d(t) represents the lateral distance of the vehicle at time t, t is the time variable, (c0, c1, c2, c3, c4, c5), (a0, a1, a2, a3, a4, a5) are the polynomial coefficients that need to be solved; The initial state conditions and the terminal state conditions of the longitudinal trajectory and the lateral trajectory are respectively set according to the state information set, the polynomial coefficients of the longitudinal trajectory and the lateral trajectory are solved, and an obstacle avoidance trajectory set is generated.

4. The method for avoiding obstacles by an autonomous driving vehicle according to claim 1, characterized in that: The obstacle avoidance trajectory set is screened to determine the optimal trajectory, specifically: The obstacle avoidance trajectory set is screened to eliminate trajectories whose trajectory curvature range, speed range, acceleration range, and acceleration change rate range do not meet the requirements, to obtain a first trajectory set; Taking the tangent direction of each track in the first track set as the horizontal axis and the normal direction of each track in the first track set as the vertical axis, respectively establishing a corresponding first Frenet coordinate system for each track in the first track set; Obtaining a lateral distance of the obstacle in each first Frenet coordinate system, and removing trajectories whose lateral distances are less than a preset threshold from the first trajectory set to obtain a second trajectory set; The maximum acceleration change rate of all trajectories in the second trajectory set is calculated, and the trajectory with the smallest maximum acceleration change rate is taken as the optimal trajectory.

5. The method for avoiding obstacles by an autonomous driving vehicle according to claim 1, characterized in that: The determining whether it is necessary to bypass the obstacle according to the obstacle information and the first feasible area is specifically: A second Frenet coordinate system is established with the direction of the road centerline as the longitudinal axis and the direction perpendicular to the road centerline as the transverse axis; Calculating a first lateral offset required for the vehicle to circumvent the obstacle from the left and a second lateral offset required for the vehicle to circumvent the obstacle from the right; Determine, according to the obstacle information, a first distance of the obstacle from the closest point of the vehicle along the original trajectory and a second distance of the obstacle from the farthest point of the vehicle along the original trajectory; Whether obstacle avoidance is required is determined according to the first distance, the second distance, the first lateral offset, the second lateral offset, the second Frenet coordinate system, and the first feasible area.

6. The method for avoiding obstacles by an autonomous driving vehicle according to claim 1, characterized in that: The first feasible area is delineated according to the centimeter-level high-precision map and the original trajectory, specifically: Acquire lane information on both sides of the original trajectory from the high-precision map and combine them into a second feasible area; The lane direction is obtained according to the opposite lane given by the high-precision map, and when the angle between the lane direction and the direction of the original trajectory is greater than a preset angle, it is determined whether the opposite lane can be crossed according to the middle lane lines of the two opposite lanes; If the middle lane line is a double yellow line, a single yellow solid line or has a guardrail, the opposite lane that is not allowed to cross the domain is eliminated in the second feasible area to obtain the first feasible area; Otherwise, the second feasible region is taken as the first feasible region.

7. An obstacle avoidance device for an autonomous driving vehicle, characterized in that: include: Data acquisition module, obstacle avoidance judgment module, obstacle avoidance trajectory generation module and obstacle avoidance trajectory execution module: The data acquisition module is used to obtain obstacle information in front of the vehicle through a laser radar carried by the vehicle, and to delineate a first feasible area based on a centimeter-level high-precision map and an original trajectory of the vehicle; The obstacle avoidance judgment module is used to judge whether obstacle avoidance is needed according to the obstacle information and the first feasible area; The obstacle avoidance trajectory generation module is used to obtain a state information set by a sampling method if obstacle avoidance is required, and generate an obstacle avoidance trajectory set based on the state information set by using a quintic polynomial calculation method, and then screen the obstacle avoidance trajectory set to determine the optimal trajectory; The obstacle avoidance execution module is used to control the vehicle to execute the obstacle avoidance action along the optimal trajectory and return to the original trajectory after completing the obstacle avoidance.

8. The obstacle circumvention device for an automatic driving vehicle according to claim 7, characterized in that: The obstacle avoidance trajectory generation module includes: a state sampling unit and a state information generation unit; The state sampling unit is used to translate the original trajectory toward the obstacle avoidance direction to obtain an obstacle avoidance reference trajectory, and to sample points at multiple preset sampling points on the obstacle avoidance reference trajectory to generate an end state set; The state information generating unit is used to combine the starting state of the vehicle and the ending state set to generate a state information set.

9. The obstacle circumvention device for an automatic driving vehicle according to claim 7, characterized in that: The obstacle avoidance trajectory generation module further includes: a trajectory generation unit; The trajectory generation unit is used to obtain the trajectory according to the horizontal and vertical dimensions respectively. The vertical trajectory is obtained by the fifth-order polynomial s(t)=c0+c1t+c2t 2 +c3t 3 +c4t 4 +c5t 5 The lateral trajectory is represented by a fifth-order polynomial d(t) = a0+a1t+a2t 2 +a3t 3 +a4t 4 +a5t 5 Represents, where s(t) represents the longitudinal distance of the vehicle at time t, d(t) represents the lateral distance of the vehicle at time t, t is the time variable, (c0, c1, c2, c3, c4, c5), (a0, a1, a2, a3, a4, a5) are the polynomial coefficients that need to be solved; The trajectory generation unit is further used to respectively set initial state conditions and terminal state conditions of the longitudinal trajectory and the lateral trajectory according to the state information set, solve the polynomial coefficients of the longitudinal trajectory and the lateral trajectory, and generate an obstacle avoidance trajectory set.

10. The obstacle circumvention device for an autonomous driving vehicle according to claim 7, characterized in that: The obstacle avoidance trajectory generation module further includes: a trajectory screening unit; The trajectory screening unit is used to screen the obstacle avoidance trajectory set, remove trajectories whose trajectory curvature range, speed range, acceleration range, and acceleration change rate range do not meet the requirements, and obtain a first trajectory set; Taking the tangent direction of each track in the first track set as the horizontal axis and the normal direction of each track in the first track set as the vertical axis, respectively establishing a corresponding first Frenet coordinate system for each track in the first track set; Obtaining a lateral distance of the obstacle in each first Frenet coordinate system, and removing trajectories whose lateral distances are less than a preset threshold from the first trajectory set to obtain a second trajectory set; The maximum acceleration change rate of all trajectories in the second trajectory set is calculated, and the trajectory with the smallest maximum acceleration change rate is taken as the optimal trajectory.

11. The obstacle circumvention device for an automatic driving vehicle according to claim 7, characterized in that: The obstacle avoidance judgment module includes a coordinate system establishment unit, a lateral offset calculation unit, an obstacle distance determination unit and an obstacle avoidance judgment unit; The coordinate system establishing unit is used to establish a second Frenet coordinate system with the road centerline direction as the longitudinal axis and the direction perpendicular to the road centerline as the transverse axis; The lateral offset calculation unit is used to calculate a first lateral offset required for the vehicle to avoid an obstacle from the left and a second lateral offset required for the vehicle to avoid an obstacle from the right; The obstacle distance determination unit is used to determine, according to the obstacle information, a first distance of the obstacle from the closest point of the vehicle along the original trajectory and a second distance of the obstacle from the farthest point of the vehicle along the original trajectory; The obstacle avoidance judgment unit is used to judge whether obstacle avoidance is needed according to the first distance, the second distance, the first lateral offset, the second lateral offset, the second Frenet coordinate system and the first feasible area.

12. The obstacle circumvention device for an automatic driving vehicle according to claim 7, characterized in that: The data acquisition module includes a lane information acquisition unit, a lane direction determination unit and a feasible area determination unit; The lane information acquisition unit is used to acquire lane information on both sides of the original trajectory from the high-precision map and combine them into a second feasible area; The lane direction judgment unit is used to obtain the lane direction according to the opposite lane given by the high-precision map, and when the angle between the lane direction and the direction of the original trajectory is greater than a preset angle, judge whether it is possible to cross the opposite lane according to the middle lane lines of the two opposite lanes; The feasible area determination unit is used to eliminate the opposite lane that is not allowed to cross the second feasible area to obtain the feasible area if the middle lane line is a double yellow line, a single yellow solid line or has a guardrail; otherwise, the second feasible area is used as the first feasible area.

13. A computer storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for avoiding obstacles of an autonomous driving vehicle as described in any one of claims 1 to 6 is implemented.

Images