Multi-degree-of-freedom drive-by-wire unmanned vehicle obstacle avoidance path planning method

Abstract

A multi-degree-of-freedom drive-by-wire unmanned vehicle obstacle avoidance path planning method mainly comprises the following steps that based on environment perception information, surrounding obstacles are analyzed and processed, and an environment safety boundary is established; on the basis of an environment safety boundary, speed sampling is carried out based on dynamic window trajectory planning of hierarchical response, and an optimal local trajectory is calculated and selected; in the driving and obstacle avoidance process, according to obstacle distribution and road condition fluctuation real-time information, the posture of a vehicle body is dynamically adjusted, speed sampling is updated, and path planning is updated. According to the method, the optimal obstacle avoidance path can be planned in real time, so that the unmanned vehicle can quickly respond in a complex scene, and potential collision risks can be avoided in time; the characteristics that the multi-degree-of-freedom drive-by-wire unmanned vehicle is suspended and the vehicle body posture is adjustable are fully utilized, vehicle posture adjustment is fused into an obstacle avoidance strategy, and the feasibility of an obstacle avoidance planning track and the stability of vehicle execution are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of driverless technology, and particularly relates to a method for obstacle avoidance path planning of a multi-degree-of-freedom by-wire unmanned vehicle. Background Art

[0002] At present, most of the mainstream means for obstacle avoidance path planning of unmanned vehicles rely on traditional path search algorithms, such as the A* algorithm for example. After constructing a map, starting from the starting point, it gradually explores a feasible path to the target point on the map by expanding nodes. In a relatively ideal scenario with regular obstacle distribution and static state, it can find a route for the vehicle to bypass known obstacles. However, the disadvantages are also obvious: Firstly, it has strict requirements on the map accuracy. If there are slight deviations in the map, the planned path may hit a wall during actual driving and cannot pass smoothly; Secondly, in the face of dynamic obstacles, such as suddenly breaking-in pedestrians, vehicles, or the unique kinematic constraint scenarios of multi-degree-of-freedom vehicles, the A* algorithm lacks a mechanism for real-time and flexible path adjustment and is difficult to quickly adapt to the dynamic changes of the environment; Thirdly, the prior art does not consider the additional behavioral constraints brought by the complex mechanical transmission structure of the multi-degree-of-freedom unmanned vehicle itself to its kinematics, dynamics, etc., and cannot guarantee the motion efficiency and safety of the unmanned vehicle in a dynamically changing environment, and it is difficult to adjust the path in time to cope with the emergence of sudden obstacles or environmental changes.

[0003] Some other studies focus on real-time obstacle avoidance strategies based on sensors, using sensing devices such as lidar and cameras to capture the information of obstacles around the vehicle in real time, and making operations such as steering, accelerating, and decelerating instantly according to preset rules to avoid obstacles. This method is likely to make the vehicle fall into a local optimal dilemma, resulting in chaotic obstacle avoidance actions. Summary of the Invention

[0004] The present disclosure provides a method for obstacle avoidance path planning of a multi-degree-of-freedom by-wire unmanned vehicle. Under the environmental perception technology, this method analyzes and processes the existing obstacles around, and establishes an environmental safety boundary; on this basis, it designs a multi-objective safety trajectory planning method for environment-vehicle coupling, combines the mechanical performance and dynamic constraints of the vehicle, constructs a multi-objective optimization function and constraint function, transforms the vehicle trajectory planning problem into a dynamic optimization problem, and solves the vehicle motion planning instruction in real time, so as to realize the emergency obstacle avoidance function of the vehicle;

[0005] Secondly, fully exploit the unique advantages of the multi-degree-of-freedom steer-by-wire unmanned vehicle. Given its adjustable suspension and controllable vehicle body posture, innovatively incorporate vehicle posture adjustment into the scope of obstacle avoidance safety constraints. Through multi-degree-of-freedom coordinated control, focus on enhancing the feasibility of the obstacle avoidance planning trajectory and the stability of vehicle execution. When encountering unexpected situations in a dynamically changing scenario, enable the unmanned vehicle to dynamically optimize the vehicle posture and driving trajectory based on real-time road conditions, ensuring safe driving and fully meeting the urgent needs of the stable and safe operation of the unmanned vehicle under complex real-world working conditions.

[0006] The obstacle avoidance path planning method for the multi-degree-of-freedom steer-by-wire unmanned vehicle provided by the present disclosure includes the following steps:

[0007] S1. Based on the environmental perception information, extract and analyze the surrounding obstacles, judge the collision risk of the obstacles, and establish an environmental safety boundary;

[0008] S2. When there is a collision risk, adopt a dynamic window trajectory planning method based on hierarchical response, perform speed sampling, and calculate and select the optimal local trajectory;

[0009] S3. During driving and obstacle avoidance, dynamically adjust the vehicle body posture according to the real-time information of the obstacle distribution and road conditions, and optimize and update the path planning by updating the speed sampling at the next moment.

[0010] Further, the step S1 includes:

[0011] Initialize the system performance parameters and the target point position;

[0012] Obtain the environmental perception information during the driving of the unmanned vehicle, and perform the recognition and feature extraction of the ground obstacles;

[0013] Based on the obstacle avoidance function, judge the collision risk of the obstacles, and construct an environmental safety boundary including the non-passable area, the characteristics of static obstacles, and the characteristics of dynamic obstacles.

[0014] Further, in the step S1, an obstacle avoidance function in the following form is adopted:

[0015]

[0016] Among them, J obs,i is used to quantify the degree of influence of the obstacle on the vehicle safety. The larger the value of J obs,i , the higher the potential collision risk between the vehicle and the obstacle, indicating the higher the current urgency of obstacle avoidance for the vehicle;

[0017] S obs is the weight coefficient; v i is the vehicle driving speed, d iis the distance between the vehicle and the obstacle at the i-th step, (x i ,y i ) is the position of the vehicle at the i-th step, (x0, y0) is the position coordinate of the obstacle point in the vehicle body coordinate system, α is the minimum distance at which the obstacle does not affect the vehicle; k is a set positive number used to prevent the denominator from being 0.

[0018] Further, the step S2 specifically includes:

[0019] S21. In the velocity space, according to the multi-constraint model for obstacle avoidance of the vehicle, set a velocity sampling dynamic window based on hierarchical response;

[0020] S22. If there is a collision risk with the obstacle, perform multiple sets of motion state samplings and correspondingly generate hierarchical prediction trajectories;

[0021] S23. Use an evaluation function to comprehensively evaluate each trajectory and select the optimal local trajectory.

[0022] Further, the velocity space in the step S21 is a two-dimensional space composed of the vehicle's linear velocity v and angular velocity ω, where each sampling point (v, ω) represents a possible velocity combination.

[0023] Further, the multi-constraint model for obstacle avoidance in the step S21 includes: the performance constraints of the vehicle, the geometric kinematic constraints of the vehicle, the obstacle avoidance safety constraints of the vehicle, and the yaw angular velocity safety boundary based on vehicle kinematics. The final velocity window is the intersection of the above constraint conditions; where:

[0024] The performance constraints of the vehicle include the range limit boundaries of the maximum speed, acceleration, yaw angular velocity, and yaw angular velocity change rate;

[0025] The geometric kinematic constraints of the vehicle include the minimum turning radius boundary, the maximum front wheel and rear wheel steering angles;

[0026] The obstacle avoidance safety constraint of the vehicle means that in the established environmental safety boundary, under the condition of the maximum deceleration, the speed should meet the following range. When the sampling speed of the driverless vehicle is within this range, it can achieve safe deceleration with the constraint of the maximum deceleration until it avoids the obstacle:

[0027]

[0028] In the formula, dist(v, ω) represents the closest distance between the corresponding simulated trajectory and the obstacle at the current speed, a v max 、a ωmax are the maximum linear acceleration and maximum angular acceleration of the vehicle respectively;

[0029] Yaw rate safety boundary:

[0030]

[0031] In the above formula, L is the wheelbase and v is the vehicle speed.

[0032] Furthermore, in the step S21, the method for setting a dynamic window for speed sampling based on hierarchical response includes:

[0033] Set a three-level sampling window {V1, V2, V d}:

[0034] V1 = {(v, ω)|v ∈ [v c -a f Δt, v c , ω = ω c};

[0035] V2 = {(v, ω)|v ∈ [v c -a f Δt, v c +a max Δt], ω = ω c};

[0036]

[0037] where a f is the deceleration, Δt is the time step, a max is the maximum acceleration, represents the maximum angular velocity change, v c is the current speed, and ω c is the current angular velocity;

[0038] V1 represents adjusting only the speed, V2 represents the need to expand the speed adjustment range, and V d represents adjusting both the speed and the direction simultaneously;

[0039] The design logic of the three-level sampling window according to the priority: Give priority to using V1, avoid obstacles by decelerating, and maintain vehicle stability; if V1 cannot meet the obstacle avoidance requirements, then use V2, avoid obstacles by accelerating or decelerating; if V2 still cannot meet the obstacle avoidance requirements, then use V d , avoid obstacles by adjusting both the speed and the direction simultaneously.

[0040] Furthermore, the specific method of the step S23 includes:

[0041] For each sampling point (v, ω) in the dynamic window, there corresponds an expected trajectory point coordinate P, and calculate its evaluation value G(v, ω). The calculation formula is as follows:

[0042]

[0043] Where: α, β, γ are the distribution coefficients of the evaluation function, is the dynamic distance evaluation function, obs(B, P) is the ground obstacle avoidance evaluation function, heading(G, P) is the target distance evaluation function, B is the ground obstacle coordinate, and G is the target point coordinate;

[0044] Dynamic distance evaluation function: Used to evaluate the proximity between the trajectory point P and the dynamic obstacle velocity vector The closer the value is, the higher the risk of collision. The formula is as follows:

[0045]

[0046] Ground obstacle avoidance evaluation function: obs(B, P) is used to evaluate the closest distance between the current trajectory point P and the ground static obstacle coordinate B. When the distance is greater than the critical safety distance, there is no risk of collision. The smaller the distance value, the higher the risk of collision.

[0047]

[0048] Target distance evaluation function: heading(G, P) is used to evaluate the distance between the coordinate of the trajectory point P generated at the current sampling speed and the target point G. The greater the distance, the higher the degree of deviation.

[0049]

[0050] Among them, L1 is the default distance value;

[0051] Combining the above evaluation functions, after weighted calculation by the distribution coefficient, the trajectory point with the lowest risk and the highest evaluation value is obtained, and the optimal sampling point (v best , ω best ) is obtained to achieve local optimal path planning.

[0052] Furthermore, the specific method of the step S3 includes:

[0053] During the obstacle avoidance path planning process, according to the current state of the vehicle, road conditions, and obstacle information, the body attitude is adjusted in real time through the active suspension, and the control of the steering, braking, and driving systems is coordinated to achieve the optimal dynamic response of the vehicle during obstacle avoidance;

[0054] Specific adjustments include the following:

[0055] When avoiding obstacles, the vehicle is performing a high-speed steering maneuver, and the body attitude tilts towards the inner side of the curve; the body tilt angle is dynamically adjusted according to the curve curvature and vehicle speed to reduce the path deviation caused by understeering or oversteering;

[0056] When the vehicle is traveling at high speed, to reduce wind resistance, the suspension system lowers the body height, making the vehicle chassis closer to the road surface;

[0057] When the vehicle brakes emergently, the vertical loads on the front and rear axles are changed by the body tilting backward, reducing the risk of sideslip and loss of control during braking.

[0058] Furthermore, the environmental perception information includes one or more of video, millimeter-wave radar, and lidar detection data.

[0059] Compared with the prior art, the beneficial effects of the present disclosure are as follows: ① It can plan the optimal obstacle avoidance path in real time, enabling the unmanned vehicle to respond quickly in complex scenarios and avoid potential collision risks in a timely manner; ② It fully considers the kinematic and dynamic constraints of the vehicle to ensure the feasibility of the planned path; ③ It makes full use of the characteristics of the multi-degree-of-freedom steer-by-wire unmanned vehicle's suspension and adjustable body attitude, integrates the vehicle attitude adjustment into the obstacle avoidance strategy, and executes obstacle avoidance attitude actions that can ensure driving safety and stability as much as possible; ④ During the obstacle avoidance process, the body attitude is dynamically optimized according to real-time information such as the distribution of obstacles and road conditions; ⑤ It enables the vehicle to pass through narrow channels and cross small obstacles at a more reasonable angle and height, expanding the passable area; ⑥ It reduces unnecessary steering and braking operations, enhances the vehicle's response ability under extreme conditions, and improves the obstacle avoidance effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] By describing the exemplary embodiments of the present disclosure in more detail in conjunction with the drawings, the above and other objects, features, and advantages of the present disclosure will become more obvious. Among them, in the exemplary embodiment mode of the present disclosure, the same reference numerals generally represent the same components.

[0061] Figure 1 It is a flow chart for solving the trajectory of obstacle avoidance path planning under an exemplary embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0062] The preferred embodiments of the present disclosure will be described in more detail below with reference to the drawings. Although the preferred embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to make the present disclosure more thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art.

[0063] The present disclosure provides a method for obstacle avoidance path planning of a multi-degree-of-freedom by-wire unmanned vehicle. First, under the environmental perception technology, the surrounding obstacles are analyzed and processed to establish an environmental safety boundary. On this basis, a multi-objective safety trajectory planning method for environment-vehicle coupling is designed. Combining the mechanical performance and dynamic constraints of the vehicle, a multi-objective optimization function and constraint function are constructed to transform the vehicle trajectory planning problem into a dynamic optimization problem, and the vehicle motion planning command is solved in real time, so as to realize the emergency obstacle avoidance function of the vehicle.

[0064] In an exemplary embodiment, the hardware platform of the multi-degree-of-freedom by-wire unmanned vehicle has the following characteristics:

[0065] The by-wire unmanned vehicle has twelve degrees of freedom, namely four-wheel independent drive, four-wheel independent steering, and four-wheel suspension independently adjustable. The multi-degree-of-freedom by-wire vehicle realizes the vehicle's ability to present states such as left tilt, right tilt, forward tilt, or backward tilt through active adjustment of the suspension on the basis of moving forward and backward, turning left and right. Thus, the vehicle has the following 7 types of motion postures and can perform rapid obstacle avoidance: 1. Extreme acceleration; 2. Extreme deceleration; 3. Extreme steering; 4. Body squatting; 5. Body forward tilt; 6. Body backward tilt; 7. Body side tilt.

[0066] The four-wheel independent steering system enables each wheel to independently control the steering angle. When making a small-radius turn, by reversing the steering of the rear wheels and the front wheels, the turning radius of the vehicle can be significantly reduced, improving the vehicle's passability in scenarios such as narrow curves and complex intersections; when changing lanes at high speed, the front and rear wheels cooperate in steering to ensure that the vehicle can complete the lane change operation smoothly and quickly, reducing body roll and yaw and enhancing driving safety. Facing right-angle curves, U-shaped curves, etc., the vehicle flexibly adjusts the steering angle of the wheels to complete the steering action with the smallest space and reduce the collision risk; the active suspension adjustment system accurately regulates the body height based on the driving state information, effectively enhancing the driving stability and maneuverability of the vehicle.

[0067] The obstacle avoidance path planning of such multi-degree-of-freedom by-wire unmanned vehicles includes the following steps:

[0068] 1. Construct a safety boundary

[0069] Obtain the environmental perception information during the driving process of the unmanned vehicle, including video, millimeter-wave radar, and lidar detection data, and perform the identification and feature extraction of ground obstacles. Then, based on the obstacle avoidance function, construct an environmental safety boundary including infeasible regions, static obstacle features, and dynamic obstacle features.

[0070] Among them, the design of the obstacle avoidance function is used to describe the proximity of the local path to environmental obstacles. The basic idea of the obstacle avoidance function is that the closer the target point on the local path is to the obstacle point, the larger the value of the obstacle avoidance function should be; the greater the current speed of the vehicle, the higher the difficulty of avoidance, and the larger the value of the obstacle avoidance function should be. When the distance between the vehicle and the obstacle exceeds a certain value, the vehicle will not be affected by it.

[0071] To enable the vehicle to dynamically adjust its obstacle avoidance behavior according to speed and distance when approaching an obstacle, considering the distance between the vehicle and the obstacle and the impact of the vehicle's current speed on obstacle avoidance, in this embodiment, an obstacle avoidance function of the following form is selected, as shown in the following formula:

[0072]

[0073] Where J obs,i is used to quantify the degree of influence of the obstacle on the vehicle's safety. The larger the value of J obs,i , the higher the potential collision risk between the vehicle and the obstacle, which means the higher the current urgency of obstacle avoidance for the vehicle, and the system needs to take more active obstacle avoidance measures to ensure safety;

[0074] S obs is the weight coefficient, v i is the vehicle's driving speed, d i is the distance between the vehicle and the obstacle at the i-th step, (x i , y i ) is the position of the vehicle at the i-th step, (x0, y0) is the position coordinates of the obstacle point in the vehicle body coordinate system, α is the minimum distance at which the obstacle has no effect on the vehicle, and k is a small positive number used to prevent the denominator from being zero.

[0075] When d i ≤α, the function value J obs,i is directly proportional to the reciprocal of the vehicle speed ν i and the distance d i , which means that the faster the vehicle speed or the closer the distance to the obstacle, the larger the function value, indicating that the obstacle avoidance behavior is more important;

[0076] When d i >α, the function value is zero, indicating that the obstacle has no effect on the vehicle.

[0077] The environmental safety boundary is constructed based on the perception information and the obstacle avoidance function. The position and geometric features of static obstacles are extracted, and the position, speed, and movement direction of dynamic obstacles are extracted. According to the obstacle avoidance function, the influence range of the obstacles is determined. Combining the non-passable area, the boundaries of static obstacles, and the boundaries of dynamic obstacles forms a complete environmental safety boundary.

[0078] The environmental safety boundary is dynamic and will be updated in real time as the vehicle moves and obstacles change. For example, when a dynamic obstacle approaches the vehicle, the safety boundary will shrink; when the obstacle moves away, the safety boundary will expand.

[0079] 2. Dynamic Window Trajectory Planning Based on Hierarchical Response

[0080] To achieve the most efficient and safe evasive planning results, a dynamic window trajectory planning method based on hierarchical response is adopted. This algorithm means:

[0081] (1) In the velocity space, according to the multi-constraint model of vehicle obstacle avoidance, calculate the hierarchical velocity sampling space, and sample multiple groups of velocities within the velocity sampling space;

[0082] (2) Analyze and predict the vehicle's movement trajectories under each group of velocities within a certain period of time;

[0083] (3) Use an evaluation function to comprehensively evaluate each trajectory and select the optimal local trajectory. Among the sampled velocity groups, through three influence weights, weighted solutions are carried out for various weights to realize the evaluation of each state point within the trajectory planning interval, so that the vehicle can reach the expected target point at high speed and stably under the constraints of the safety boundary.

[0084] The further description is as follows:

[0085] 2.1 Multi-constraint Model of Obstacle Avoidance

[0086] Solving the optimal local trajectory based on the dynamic window mechanism of hierarchical response, the core lies in the sampling of the velocity space and the establishment of the evaluation function.

[0087] To improve the feasibility of the planned trajectory and the stability of vehicle execution, a multi-constraint model of obstacle avoidance is integrated for velocity space sampling. The velocity space in this embodiment is a two-dimensional space composed of two main motion parameters of the vehicle, linear velocity v and angular velocity ω, where each sampling point (v, ω) represents a possible velocity combination.

[0088] The multi-constraint model of obstacle avoidance mainly includes: the performance constraints of the vehicle, the obstacle avoidance safety constraints of the vehicle, the geometric kinematic constraints of the vehicle, and the yaw angular velocity safety boundary based on vehicle kinematics. The final velocity of the velocity window is the intersection of the above constraint conditions. Among them:

[0089] 1) The performance constraints of the vehicle mainly include the range limit boundaries in terms of maximum speed, acceleration, yaw angular velocity, and yaw angular velocity change rate.

[0090] 2) The geometric kinematic constraints of the vehicle include the minimum turning radius boundary, the maximum front and rear wheel steering angles.

[0091] 3) The obstacle avoidance safety constraint of the vehicle means that in the established environmental safety boundary, to achieve safe obstacle avoidance for the vehicle, it is necessary to ensure that it can stop or change its trajectory and attitude before reaching the obstacle. Therefore, under the condition of maximum deceleration, there is a speed range:

[0092]

[0093] In the formula, dist(v, ω) represents the closest distance between the simulated trajectory corresponding to the current speed and the obstacle, a vmax 、a ωmax are the maximum linear acceleration and maximum angular acceleration of the vehicle respectively; in the case of no obstacle, dist(v, ω) will be a very large constant value. When the running sampling speed of the driverless vehicle is within the range of the above formula, it can achieve safe deceleration until avoiding the obstacle under the constraint of maximum deceleration.

[0094] 4) Yaw rate safety boundary:

[0095]

[0096] In the above formula, L is the wheelbase and v is the vehicle speed.

[0097] 2.2 Hierarchical sampling space

[0098] To achieve the rapid solution of the optimal solution and sub-optimal solution, a three-level response mechanism is designed to dynamically adjust the window range. For the characteristics of some dynamic obstacles with high speed and short reaction time, in this case, due to the constraints of vehicle dynamics, rapid and sharp steering operations are carried out. Especially during the high-speed driving of the vehicle, it is extremely easy to cause rollover, which is not conducive to maintaining the stability of the vehicle. Therefore, in the face of an emergency obstacle avoidance scenario, the feasibility of non-steering avoidance methods of acceleration and deceleration is preferentially evaluated to avoid rollover caused by sharp steering. If not, then consider the steering avoidance method.

[0099] Based on the range of the dynamic window, it is divided into three-level sampling windows {V1, V2, V d}, and according to the speed, distance of the obstacle and the dynamic constraints of the vehicle, the speed and angular velocity range of the driverless vehicle are dynamically adjusted. When facing obstacle avoidance scenarios with different emergency levels, the optimal or sub-optimal solution can be quickly solved. As follows:

[0100] V1 = {(v, ω)|v ∈ [v c -a f Δt, v c , ω = ω c}

[0101] V2 = {(v, ω)|v ∈ [v c -a f Δt, v c +amax Δt], ω = ω c}

[0102]

[0103] where a f is the deceleration, Δt is the time step, and a max is the maximum acceleration, represents the maximum angular velocity change. V1 only adjusts the speed and is suitable for gentle scenarios. V2 expands the speed adjustment range and is suitable for scenarios that require flexible speed adjustment. V d adjusts both the speed and direction simultaneously and is suitable for emergency obstacle avoidance scenarios.

[0104] The three-level sampling window follows the design logic of priority: V1 is preferentially used to avoid obstacles by decelerating and maintain vehicle stability; if V1 cannot meet the obstacle avoidance requirements, then V2 is used to avoid obstacles by accelerating or decelerating; if V2 still cannot meet the obstacle avoidance requirements, then V d is used to avoid obstacles by adjusting both the speed and direction simultaneously.

[0105] Through this hierarchical mechanism, efficient obstacle avoidance decisions can be made while ensuring vehicle stability.

[0106] 2.3 Evaluation of the predicted trajectory

[0107] Among the sampled speed groups, an evaluation function is used to comprehensively evaluate each trajectory. By weighting and solving various weights through three influence weights, the evaluation of each state point within the trajectory planning interval is achieved, enabling the vehicle to reach the desired target point at high speed and stably under the constraints of the safety boundary.

[0108] For each sampling point (v, ω) in the dynamic window, there corresponds a trajectory point coordinate P. Calculate its evaluation value G(v, ω), and the point with the highest evaluation value is taken as the optimal trajectory point. The evaluation function formula is as follows:

[0109]

[0110] In the formula: α, β, γ are the evaluation function distribution coefficients, is the dynamic distance evaluation function, obs(B, P) is the ground obstacle avoidance evaluation function, heading(G, P) is the target distance evaluation function, B is the ground obstacle coordinate, and G is the target point coordinate.

[0111] (1) Dynamic distance evaluation function: is used to evaluate the predicted trajectory point P and the dynamic obstacle velocity vector The degree of proximity, this evaluation directly reflects the possibility of the trajectory colliding with obstacles during the dynamic process. The closer it is, the higher the collision risk. The formula is as follows:

[0112]

[0113] (3) Ground obstacle avoidance evaluation function: obs(B, P) is used to evaluate the closest distance between the current trajectory point P and the coordinates B of the ground static obstacle. When the distance is greater than the critical safety distance, there is no danger of collision. The smaller the distance value, the higher the collision risk.

[0114]

[0115] (4) Target distance evaluation function: heading(G, P) is used to evaluate the distance between the coordinates of the trajectory end point P generated at the current sampling speed and the target point G. The greater the distance, the higher the degree of deviation, and the higher the cost value.

[0116]

[0117] Among them, L1 is the default distance value;

[0118] Combining the above evaluation functions, after weighted calculation by the distribution coefficient, the trajectory point with the lowest cost and the highest evaluation value is obtained, and the optimal sampling point (v best , ω best ) is obtained to achieve local optimal path planning.

[0119] 3. Regulation of vehicle posture

[0120] During the obstacle avoidance path planning process, according to the current state of the vehicle, road conditions, and obstacle information, the body posture is adjusted in real time through the active suspension, and combined with the precise control of the steering, braking, and drive systems, the optimal dynamic response of the vehicle during obstacle avoidance is achieved.

[0121] When the vehicle is performing a high-speed steering operation during obstacle avoidance, the body posture tilts moderately towards the inner side of the curve, increasing the ground contact force of the outer tires of the vehicle and improving the lateral adhesion limit of the tires, thereby allowing the vehicle to turn safely at a higher speed. The body tilt angle can be dynamically adjusted according to factors such as the curve curvature and vehicle speed, reducing the path deviation caused by understeering or oversteering.

[0122] When the vehicle is driving at high speed, to reduce wind resistance, the suspension system will lower the body height, making the vehicle chassis closer to the road surface.

[0123] When the vehicle brakes emergently, the center of gravity of the vehicle will shift forward. At this time, the body can be tilted backward to change the vertical load on the front and rear axles, reducing the risk of side slip and loss of control of the vehicle during braking.

[0124] When adjusting the vehicle body attitude by suspension, such as tilting the vehicle body to the left or right by an angle θ, the ground contact force distribution of the vehicle's tires will change. At this time, the lateral dynamic equation of the vehicle needs to additionally consider the gravitational component caused by the vehicle body tilt, that is:

[0125]

[0126] where m is the vehicle mass, v x is the vehicle's forward speed, v y is the lateral speed, ω is the yaw angular velocity, F yf and F yr are the lateral forces of the front and rear tires respectively, and g is the acceleration due to gravity.

[0127] In this embodiment, the vehicle attitude adjustment is incorporated into the scope of the obstacle avoidance safety constraint conditions. Through multi-degree-of-freedom collaborative control, the feasibility of the obstacle avoidance planning trajectory and the stability of vehicle execution are mainly improved.

[0128] The overall process of the trajectory planning in this embodiment is as follows: trajectory prediction is carried out through speed sampling to obtain a better path planning; during the process of driving along the path, the vehicle attitude is adjusted in real time; the adjustment of the vehicle attitude is reflected in the vehicle speed, and the sampling is continuously updated based on the speed, and then the path planning is updated.

[0129] The overall process of this embodiment is as shown in the appendix Figure 1 as follows:

[0130] First, the system initializes the performance parameters and the target point position, obtains the vehicle state and environmental information, and constructs the environmental safety boundary. Then, the system checks whether there is a collision risk. If there is a risk, the system will sample the motion state and generate a simulated trajectory with three levels of priorities. It will first check whether the level 1 simulated trajectory will collide. If there is no feasible trajectory, the system will judge that there is no feasible trajectory at present, and then check the level 2 simulated trajectory. If level 2 still cannot meet the obstacle avoidance requirements, level 3 will be used; otherwise, the optimal trajectory will be selected and the dynamic window trajectory will be traversed, and finally the process will end.

[0131] Among them, the three-level trajectories are generated simultaneously. In each time window, first evaluate the feasibility of applying the non-steering avoidance strategy to reduce the risk of vehicle instability. If a feasible non-steering solution cannot be generated, then consider the steering avoidance strategy.

[0132] The above technical solutions are only exemplary embodiments of the present invention. For those skilled in the art, based on the application methods and principles disclosed in the present invention, it is very easy to make various types of improvements or deformations, not limited to the methods described in the above specific embodiments of the present invention. Therefore, the above-described manner is only preferred and does not have a restrictive meaning.

Claims

1. A method for obstacle avoidance path planning of a multi-degree-of-freedom wire-controlled unmanned vehicle, comprising the following steps: S1. Based on the environmental perception information, extract and analyze the surrounding obstacles, judge the collision risk of the obstacles, and establish an environmental safety boundary; S2. When there is a collision risk, adopt a dynamic window trajectory planning method based on hierarchical response, perform speed sampling, and calculate and select the optimal local trajectory; S3. During driving and obstacle avoidance, dynamically adjust the vehicle body posture according to the real-time information of the obstacle distribution and road conditions undulation, and optimize and update the path planning by updating the speed sampling at the next moment.

2. The method according to claim 1, characterized in that, The step S1 includes: Initializing the system performance parameters and the target point position; Obtaining the environmental perception information during the driving process of the unmanned vehicle, and identifying and extracting the features of the ground obstacles; Based on the obstacle avoidance function, judge the collision risk of the obstacles, and construct an environmental safety boundary including non-passable areas, static obstacle features and dynamic obstacle features.

3. The method according to claim 2, wherein In the step S1, an obstacle avoidance function in the following form is adopted: Among them, J obs,i is used to quantify the degree of influence of obstacles on vehicle safety. The larger the value of J obs,i , the higher the potential collision risk between the vehicle and the obstacle, indicating a higher urgency for the vehicle to avoid obstacles currently; S obs is the weight coefficient; v i is the vehicle driving speed, d i is the distance between the vehicle and the obstacle at the i-th step, (x i , y i ) is the position of the vehicle at the i-th step, (x0, y0) is the position coordinate of the obstacle point in the vehicle body coordinate system, α is the minimum distance at which the obstacle does not affect the vehicle; k is a set positive number used to prevent the denominator from being 0.

4. The method according to claim 1, characterized in that, The step S2 specifically includes: S21. In the speed space, according to the multi-constraint model of vehicle obstacle avoidance, set a dynamic window for speed sampling based on hierarchical response; S22. If there is a collision risk for the obstacle, perform multiple groups of motion state samplings and correspondingly generate hierarchical prediction trajectories; S23. Use an evaluation function to comprehensively evaluate each trajectory and select the optimal local trajectory.

5. The method according to claim 4, wherein The speed space in the step S21 is a two-dimensional space composed of the vehicle linear speed v and the angular speed ω, and each sampling point (v, ω) represents a possible speed combination.

6. The method according to claim 4, characterized in that The multi-constraint model of vehicle obstacle avoidance in the step S21 includes: the performance constraints of the vehicle, the geometric kinematic constraints of the vehicle, the obstacle avoidance safety constraints of the vehicle, and the yaw angular speed safety boundary based on vehicle kinematics. The final speed window is the intersection of the above constraint conditions; where: The performance constraints of the vehicle include the range limit boundaries of the maximum speed, acceleration, yaw angular speed, and yaw angular speed change rate; The geometric kinematic constraints of the vehicle include the minimum turning radius boundary, the maximum values of the front and rear wheel angles; The obstacle avoidance safety constraint of the vehicle means that in the established environmental safety boundary, under the condition of the maximum deceleration, the speed should meet the following range. When the running sampling speed of the unmanned vehicle is within this range, it can achieve safe deceleration until avoiding the obstacle under the constraint of the maximum deceleration: where dist(v, ω) represents the closest distance between the corresponding simulated trajectory at the current speed and the obstacle, a vmax and a ωmax are the maximum linear acceleration and the maximum angular acceleration of the vehicle, respectively; Yaw angular speed safety boundary: In the above formula, L is the wheelbase and v is the vehicle speed.

7. The method according to claim 4, wherein In the step S21, the method for setting the dynamic window for speed sampling based on hierarchical response includes: Set a three-level sampling window {V1, V2, V d}: V1 = {(v, ω)|v ∈ [v c -a f Δt, v c , ω = ω c}; V2 = {(v, ω)|v ∈ [v c -a f Δt, v c +a max Δt], ω = ω c}; where a f is the deceleration, Δt is the time step, and a max is the maximum acceleration, represents the maximum angular velocity change, v c is the current velocity, and ω c is the current angular velocity; V1 represents adjusting only the speed, V2 represents the need to expand the speed adjustment range, and V d represents adjusting both the speed and the direction simultaneously; Design logic of the three - level sampling window according to priority: First, use V1. Avoid obstacles by decelerating to maintain vehicle stability. If V1 cannot meet the obstacle - avoidance requirements, then use V2. Avoid obstacles by accelerating or decelerating. If V2 still cannot meet the obstacle - avoidance requirements, then use V d , avoid obstacles by adjusting both speed and direction simultaneously.

8. The method according to claim 4, characterized in that, The specific method of the step S23 includes: For each sampling point (v, ω) in the dynamic window, there corresponds an expected trajectory point coordinate P, and calculate its evaluation value G(v, ω). The calculation formula is as follows: Where: α, β, γ are the distribution coefficients of the evaluation function, is the dynamic distance evaluation function, obs(B, P) is the ground obstacle avoidance evaluation function, heading(G, P) is the target distance evaluation function, B is the ground obstacle coordinate, and G is the target point coordinate; Dynamic distance evaluation function: Used to evaluate the proximity between the trajectory point P and the velocity vector of the dynamic obstacle The evaluation directly reflects the possibility of collision between the trajectory and the obstacle during the dynamic process. The closer they are, the higher the collision risk. The formula is as follows: The ground obstacle avoidance evaluation function: obs(B, P) is used to evaluate the closest distance between the current trajectory point P and the ground static obstacle coordinate B. When the distance is greater than the critical safety distance, there is no risk of collision. The smaller this distance value, the higher the collision risk. Target distance evaluation function: heading(G, P) is used to evaluate the distance between the coordinates of the trajectory point P generated at the current sampling speed and the target point G. The greater the distance, the higher the degree of deviation. Wherein, L1 is the default distance value; Based on the above evaluation functions, after weighted calculation by the distribution coefficient, the trajectory point with the lowest risk and the highest evaluation value is obtained, and the optimal sampling point (v best , ω best ) is obtained to achieve local optimal path planning.

9. The method according to any one of claims 1-8, characterized in that, The specific method of step S3 includes: During the obstacle avoidance path planning process, according to the current vehicle state, road conditions and obstacle information, the body attitude is adjusted in real time through the active suspension, and the control of the steering, braking and driving systems is coordinated to achieve the optimal dynamic response of the vehicle during obstacle avoidance. Specifically, the following adjustments are included: When the vehicle is performing a high-speed steering operation during obstacle avoidance, the body attitude tilts towards the inner side of the curve; the body tilt angle is dynamically adjusted according to the curve curvature and vehicle speed to reduce the path deviation caused by understeering or oversteering. When the vehicle is driving at high speed, to reduce wind resistance, the suspension system lowers the body height, making the vehicle chassis closer to the road surface. When the vehicle brakes emergently, the vertical loads on the front and rear axles are changed by the backward tilt of the vehicle body, reducing the risk of side slip and loss of control during braking.

10. The method according to claim 1, wherein The environmental perception information includes one or more of video, millimeter wave radar and lidar detection data.

Images