A control method, system, vehicle, and storage medium of a vehicle

Abstract

The application relates to the technical field of vehicles, in particular to a vehicle control method and system, a vehicle and a storage medium. The method comprises the following steps: acquiring current environment perception information of a vehicle; determining a current collision risk level of the vehicle and calculating a current road friction coefficient according to the current environment perception information, and determining a cost function based on the current road friction coefficient, a lateral position deviation, a heading angle deviation and a control input; determining a target control strategy of the vehicle according to the current collision risk level and the cost function, and controlling the vehicle according to the target control strategy. Thus, the problem that the current sensor scheme has degraded detection performance when facing extreme weather and the existing vehicle control strategy is insufficient to cope with the scene caused by extreme weather is solved, and on the premise of avoiding dependence on a specific sensor configuration, robust automatic driving control under extreme weather conditions is realized, and the stability of the vehicle during driving is improved.

Description

Technical Field

[0001] This application relates to the field of vehicle technology, and in particular to a vehicle control method, system, vehicle, and storage medium. Background Technology

[0002] Existing autonomous driving solutions heavily rely on sensors to perceive the surrounding environment and achieve precise control. However, real-world operating environments are complex and variable, with frequent extreme weather conditions leading to a decline in sensor performance, making them insufficient to cope with extreme weather scenarios and posing a significant challenge to autonomous driving systems.

[0003] Therefore, improving the performance of autonomous driving systems in extreme weather conditions has become a critical issue that urgently needs to be addressed. Summary of the Invention

[0004] This application provides a vehicle control method, system, vehicle, and storage medium to address the problems that current sensor solutions suffer from performance degradation in extreme weather conditions, and that existing vehicle control strategies are insufficient to cope with extreme weather scenarios.

[0005] The first aspect of this application provides a vehicle control method, comprising the following steps: acquiring current environmental perception information of the vehicle; determining the current collision risk level of the vehicle and calculating the current road surface friction coefficient based on the current environmental perception information, and determining a cost function based on the current road surface friction coefficient, lateral position deviation, heading angle deviation and control input; determining a target control strategy of the vehicle based on the current collision risk level and the cost function, and controlling the vehicle according to the target control strategy.

[0006] Optionally, determining the current collision risk level of the vehicle based on the current environmental perception information includes: calculating the risk value of the vehicle based on the vehicle's environmental perception information; if the risk value is less than a first risk threshold, determining the collision risk level of the vehicle as a first collision risk level; if the risk value is greater than or equal to the first risk threshold and less than a second risk threshold, determining the collision risk level of the vehicle as a second collision risk level, wherein the risk level of the second collision risk level is higher than the risk level of the first collision risk level; if the risk value is greater than or equal to the second risk threshold, determining the collision risk level of the vehicle as a third collision risk level, wherein the risk level of the third collision risk level is higher than the risk level of the second collision risk level.

[0007] Optionally, the current collision risk level is the first collision risk level. The step of determining the target control strategy for the vehicle based on the current collision risk level and the cost function, and controlling the vehicle according to the target control strategy, includes: if the risk level of the vehicle is the first collision risk level, then controlling the vehicle to maintain its current state; if the risk level of the vehicle is the second collision risk level, then controlling the vehicle to perform a preset easing braking action while simultaneously performing a first lateral avoidance control action or a first longitudinal adjustment control action; if the risk level of the vehicle is the third collision risk level, then controlling the vehicle to perform a preset emergency braking action while simultaneously performing a second lateral avoidance control action or a second longitudinal adjustment control action.

[0008] Optionally, when controlling the vehicle according to the target control strategy, the method includes: evaluating the risk potential gradient of the vehicle; constructing a virtual repulsive force field based on the risk potential gradient of the vehicle; planning an initial path for the vehicle based on the repulsive force field; and filtering out regions where the magnitude of the risk potential gradient is greater than a preset threshold to obtain a processed path; and adjusting the processed path according to a near-end strategy optimization algorithm to obtain an obstacle avoidance path for the vehicle, so as to control the vehicle to drive according to the obstacle avoidance path.

[0009] Optionally, the risk value is: ; Where Ψ(x,t) represents the risk value of the vehicle at position x and time t, x i Let x be the position of the i-th obstacle, x be the position of the vehicle, and t be the position of the obstacle. i Let ||x – x be the time when the i-th obstacle is first detected. i || 2 e represents the squared Euclidean distance between the vehicle and the obstacle. -β(t-ti) The risk decays exponentially over time, where β is the decay coefficient and α is the risk level. i Let be the risk weight coefficient for the i-th obstacle.

[0010] Optionally, the risk potential gradient is: ; in, Ψ(x,t) represents the risk potential gradient of the vehicle at position x and time t, where x i Let x be the position of the i-th obstacle, and x be the position of the vehicle. ||x – x i || represents the Euclidean distance between the vehicle and the obstacle, e -β(t-ti) The risk decays exponentially with time t, where β is the decay coefficient and α is the risk level. i Let t be the risk weight coefficient for the i-th obstacle. it is the time when the i-th obstacle is first detected.

[0011] Optionally, the cost function is:

[0012]

[0013] Where J is the cost function, Δy k For the lateral position error, Δ k For heading angle error, For the control input, Q is the weight of the lateral position deviation, R is the weight of the heading angle deviation, and ρ is the optimization judgment condition for minimizing the control input weights. Let || be the road surface friction coefficient, || || 2 This represents the square of the vector's magnitude. Q 0 represents the weight matrix of the cost function at the initial time step. Let Q be the weight matrix of the cost function at time k, λ be the friction sensitivity coefficient, and Q be the weight matrix of the cost function at time k. k It is a subarray of the Q array.

[0014] A second aspect of this application provides a vehicle control system, comprising: an acquisition module for acquiring current environmental perception information of the vehicle; a determination module for determining the current collision risk level of the vehicle and calculating the current road surface friction coefficient based on the current environmental perception information, and determining a cost function based on the current road surface friction coefficient, lateral position deviation, heading angle deviation, and control input; and a control module for determining a target control strategy for the vehicle based on the current collision risk level and the cost function, and controlling the vehicle according to the target control strategy.

[0015] Optionally, the determining module is further configured to: calculate the risk value of the vehicle based on the vehicle's environmental perception information; if the risk value is less than a first risk threshold, determine the collision risk level of the vehicle as a first collision risk level; if the risk value is greater than or equal to the first risk threshold and less than a second risk threshold, determine the collision risk level of the vehicle as a second collision risk level, wherein the risk level of the second collision risk level is higher than the risk level of the first collision risk level; if the risk value is greater than or equal to the second risk threshold, determine the collision risk level of the vehicle as a third collision risk level, wherein the risk level of the third collision risk level is higher than the risk level of the second collision risk level.

[0016] Optionally, the control module is specifically configured to: if the risk level of the vehicle is the first collision risk level, control the vehicle to maintain its current state; if the risk level of the vehicle is the second collision risk level, control the vehicle to perform a preset easing braking action while simultaneously performing a first lateral avoidance control action or a first longitudinal adjustment control action; if the risk level of the vehicle is the third collision risk level, control the vehicle to perform a preset emergency braking action while simultaneously performing a second lateral avoidance control action or a second longitudinal adjustment control action.

[0017] Optionally, when controlling the vehicle according to the target control strategy, the control module is further configured to: evaluate the risk potential gradient of the vehicle; construct a virtual repulsive force field based on the risk potential gradient of the vehicle; plan an initial path of the vehicle based on the repulsive force field; and filter out regions where the magnitude of the risk potential gradient is greater than a preset threshold to obtain a processed path; and adjust the processed path according to a near-end strategy optimization algorithm to obtain an obstacle avoidance path for the vehicle, so as to control the vehicle to drive according to the obstacle avoidance path.

[0018] Optionally, the risk value is: ; Where Ψ(x,t) represents the risk value of the vehicle at position x and time t, x i Let x be the position of the i-th obstacle, x be the position of the vehicle, and t be the position of the obstacle. i Let ||x – x be the time when the i-th obstacle is first detected. i || 2 e represents the squared Euclidean distance between the vehicle and the obstacle. -β(t-ti) The risk decays exponentially over time, where β is the decay coefficient and α is the risk level. i Let be the risk weight coefficient for the i-th obstacle.

[0019] Optionally, the risk potential gradient is: ; in, Ψ(x,t) represents the risk potential gradient of the vehicle at position x and time t, where x i Let x be the position of the i-th obstacle, and x be the position of the vehicle. ||x – x i || represents the Euclidean distance between the vehicle and the obstacle, e -β(t-ti) The risk decays exponentially with time t, where β is the decay coefficient and α is the risk level. i Let t be the risk weight coefficient for the i-th obstacle. i t is the time when the i-th obstacle is first detected.

[0020] Optionally, the cost function is:

[0021]

[0022] Where J is the cost function, Δy k For the lateral position error, Δ k For heading angle error, For the control input, Q is the weight of the lateral position deviation, R is the weight of the heading angle deviation, and ρ is the optimization judgment condition for minimizing the control input weights. Let || be the road surface friction coefficient, || || 2 This represents the square of the vector's magnitude. Q 0 represents the weight matrix of the cost function at the initial time step. Let Q be the weight matrix of the cost function at time k, λ be the friction sensitivity coefficient, and Q be the weight matrix of the cost function at time k. k It is a subarray of the Q array.

[0023] A third aspect of this application provides a vehicle, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the vehicle control method as described in the above embodiments.

[0024] A fourth aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement the vehicle control method as described in the above embodiments.

[0025] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0026] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart of a vehicle control method provided according to an embodiment of this application; Figure 2 This is a schematic diagram of the architecture involved in the vehicle control method according to an embodiment of this application; Figure 3 This is a flowchart illustrating the process of solving for the control quantity using a cost function according to an embodiment of this application. Figure 4 This is a schematic diagram illustrating the processing strategies corresponding to different risk levels according to embodiments of this application; Figure 5 This is a schematic diagram illustrating the adjustment of network weights using the PPO algorithm according to an embodiment of this application. Figure 6 A flowchart of a vehicle control method according to an embodiment of this application; Figure 7 This is an example diagram of a vehicle control system according to an embodiment of this application; Figure 8 This is a schematic diagram of a vehicle structure according to an embodiment of this application. Detailed Implementation

[0027] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0028] The following description, with reference to the accompanying drawings, outlines a vehicle control method, system, vehicle, and storage medium according to embodiments of this application. Addressing the issues mentioned in the background section regarding the performance degradation of current sensor solutions in extreme weather conditions and the inadequacy of existing vehicle control strategies to handle such scenarios, this application provides a vehicle control method. This method involves acquiring the vehicle's current environmental perception information; determining the vehicle's current collision risk level and calculating the current road surface friction coefficient based on the current environmental perception information; determining a cost function based on the current road surface friction coefficient, lateral position deviation, heading angle deviation, and control input; determining the vehicle's target control strategy based on the current collision risk level and the cost function; and controlling the vehicle according to the target control strategy. This solves the problems of performance degradation in current sensor solutions in extreme weather conditions and the inadequacy of existing vehicle control strategies to handle such scenarios. By avoiding reliance on specific sensor configurations, robust autonomous driving control under extreme weather conditions is achieved, improving vehicle stability during driving.

[0029] In current autonomous driving solutions, the performance and accuracy of sensors directly impact the final control precision and feasibility. However, current sensor solutions suffer significant performance degradation or deterioration in extreme weather conditions; furthermore, the control strategies employed are insufficient to address the challenges posed by extreme weather scenarios. Existing solutions often focus on single environmental factors (such as dust) and lack multimodal dynamic weight allocation schemes.

[0030] Specifically, Figure 1 This is a schematic flowchart illustrating a vehicle control method provided in an embodiment of this application.

[0031] like Figure 1 As shown, the vehicle control method includes the following steps: In step S101, the vehicle's current environmental perception information is obtained.

[0032] Specifically, such as Figure 2 As shown, the current environmental perception information is collected from multiple sources of sensors, such as... Figure 2 As shown, the data includes three main dimensions: vehicle status, surrounding environment, and meteorological characteristics. Specific data collected includes: (1) Vehicle's own state information: The vehicle's speed, longitudinal / lateral acceleration, yaw angle, wheel speed, angular acceleration and other dynamic parameters provide core data support for the identification of road friction coefficient; (2) Surrounding environment perception information: By fusing multiple sensors such as cameras, lidar, and millimeter-wave radar, the position, speed, type (vehicle / pedestrian / static object), and relative distance of obstacles are obtained, providing a basis for collision risk calculation; (3) Meteorological and road environment information: Obtain meteorological feature vectors such as visibility, precipitation intensity, and road surface type (compacted snow / water accumulation, etc.) from the vehicle-mounted meteorological detection module or external meteorological platform.

[0033] During the data acquisition process, Kalman filtering preprocessing is required for multi-sensor data to complete data normalization, spatiotemporal synchronization, and noise cancellation. This is to prevent false detections and missed detections by sensors under extreme weather conditions from affecting subsequent calculations and to ensure the accuracy and consistency of the perceived information.

[0034] In step S102, the current collision risk level of the vehicle is determined based on the current environmental perception information and the current road surface friction coefficient is calculated. The cost function is then determined based on the current road surface friction coefficient, lateral position deviation, heading angle deviation, and control input.

[0035] Among them, lateral position deviation refers to the difference between the actual lateral position of the vehicle and the lateral position of the desired trajectory (such as the lane center line or the planned path); heading angle deviation refers to the difference between the actual heading angle of the vehicle and the heading angle of the desired trajectory; control input refers to the vehicle's motion control quantities, such as front wheel steering angle, steering angular velocity, brake opening / deceleration, throttle opening, etc.

[0036] Optionally, in some embodiments, determining the current collision risk level of the vehicle based on the current environmental perception information includes: calculating the risk value of the vehicle based on the vehicle's environmental perception information; if the risk value is less than a first risk threshold, then determining the collision risk level of the vehicle to be a first collision risk level; if the risk value is greater than or equal to the first risk threshold and less than a second risk threshold, then determining the collision risk level of the vehicle to be a second collision risk level, wherein the risk level of the second collision risk level is higher than the risk level of the first collision risk level; if the risk value is greater than or equal to the second risk threshold, then determining the collision risk level of the vehicle to be a third collision risk level, wherein the risk level of the third collision risk level is higher than the risk level of the second collision risk level.

[0037] In some embodiments, the risk value is: ; Where Ψ(x,t) represents the risk value of the vehicle at position x and time t, x i Let x be the position of the i-th obstacle, x be the position of the vehicle, and t be the position of the obstacle. i Let ||x – x be the time when the i-th obstacle is first detected. i || 2 e represents the squared Euclidean distance between the vehicle and the obstacle. -β(t-ti) The risk decays exponentially over time, where β is the decay coefficient and α is the risk level. i Let α be the risk weight coefficient for the i-th obstacle. i The value range of t can be [0.1, 5.0], and the value range of β can be [0.05, 0.2]. i The range of values ​​for can be [0, t] max ], t max The threshold for the continuous exposure time of the target is a calibration value.

[0038] The specific process for calculating the risk value is as follows: Collect the current obstacle state X and time t; iterate through historical risk events from the previous time period to calculate the impact value of each individual event. As shown in the function, the spatial component α needs to be calculated separately. i / ||x – x i |||² and time component e -β(t-ti) Based on this, the contribution value of each individual obstacle is calculated; then, the contribution values ​​of related obstacles are calculated sequentially and accumulated, and finally, the total risk value is output. Here, ||x – x i || 2 The spatial decay component reflects the rapid decay caused by increasing distance, e β(t ti) The time decay component controls the rate at which historical events decay over time. The final total risk value is the sum of all obstacle events, and the risk value accumulates with the increase of neighboring events.

[0039] The risk value is the sum of the risk contributions of all obstacles. This application classifies collision risks by calculating the risk value of the vehicle.

[0040] Two risk thresholds are set: the first risk threshold ΨL and the second risk threshold ΨH; If the vehicle's risk value is less than the first risk threshold, i.e., Ψ < ΨL, the vehicle's collision risk level is determined to be low-risk collision level, and there are no significant collision hazards around the vehicle. If the vehicle's risk value is greater than or equal to the first risk threshold and less than the second risk threshold, i.e., Ψ L ≤Ψ<Ψ H The vehicle was determined to be at a medium-risk collision level, indicating that there were potential collision hazards around the vehicle. If the vehicle's risk value is greater than or equal to the second risk threshold, i.e., Ψ≥Ψ H The vehicle was determined to be at a high-risk collision level, meaning it faced a direct collision threat.

[0041] In some embodiments, the cost function is: ; ; Where J is the cost function, Δy k For the lateral position error, Δ k For heading angle error, For the control input, Q is the weight of the lateral position deviation, R is the weight of the heading angle deviation, and ρ is the optimization judgment condition for minimizing the control input weights. Let || be the road surface friction coefficient, || || 2 This represents the square of the vector's magnitude. Q 0 represents the weight matrix of the cost function at the initial time step. Let Q be the weight matrix of the cost function at time k, λ be the friction sensitivity coefficient, and Q be the weight matrix of the cost function at time k. k It is a subarray of the Q array.

[0042] Among them, the lateral position error is the difference in distance between the actual driving position of the vehicle and the expected trajectory (such as the center line of the lane) in the lateral (left and right direction), the heading angle error is the angle difference between the actual heading of the vehicle and the tangent direction of the expected trajectory, and the control inputs include the steering angle, braking deceleration, throttle opening and other action quantities.

[0043] Road surface friction coefficient μ k These are key parameters for vehicle control under extreme conditions. A friction identification algorithm based on dynamic constraints is used to calculate them in real time, combining parameters such as vehicle wheel speed, acceleration, and road surface slope. The calculation method is as follows: Based on Newton's second law and considering the characteristics of rolling and sliding friction, a dynamic equation is constructed. The wheel angular acceleration is calculated using the wheel speed gradient. Substituting parameters such as longitudinal acceleration, road surface slope, and gravitational acceleration, the current road surface friction coefficient μk is calculated. ; Where, μ k Let x¨ be the real-time friction coefficient, x¨ be the longitudinal acceleration, θ be the road surface slope, and J be the friction coefficient. w For the moment of inertia of the wheel, ω is angular acceleration.

[0044] In step S103, the target control strategy for the vehicle is determined based on the current collision risk level and cost function, and the vehicle is controlled according to the target control strategy.

[0045] Optionally, in some embodiments, a target control strategy for the vehicle is determined based on the current collision risk level and cost function, and the vehicle is controlled according to the target control strategy, including: if the vehicle's risk level is a first collision risk level, then the vehicle is controlled to maintain its current state; if the vehicle's risk level is a second collision risk level, then the vehicle is controlled to perform a preset easing braking action while simultaneously performing a first lateral avoidance control action or a first longitudinal adjustment control action; if the vehicle's risk level is a third collision risk level, then the vehicle is controlled to perform a preset emergency braking action while simultaneously performing a second lateral avoidance control action or a second longitudinal adjustment control action.

[0046] like Figure 3 As shown, the cost function dynamically optimizes the optimal control quantity in this scenario based on the real-time road friction coefficient (such as the specific deceleration of medium-risk braking and the specific steering angle of lateral avoidance), while ensuring that the control quantity does not exceed the vehicle dynamic limits (such as preventing sideslip and lock-up on low-friction road surfaces).

[0047] In this embodiment, the optimal control quantity is solved using the cost function as follows: real-time data on vehicle status, environmental perception, and road conditions are collected to calculate lateral position deviation, heading angle deviation, and real-time road friction coefficient; based on the current collision risk level, the weights of each dimension of the cost function are configured, and the real-time control input is substituted into the cost function. With min J as the objective, the optimal control command sequence within the finite time domain is obtained through rolling optimization and iterative solution using the existing quadratic programming algorithm.

[0048] The strategy is implemented in three tiers: first collision risk level (low risk, Ψ<ΨL), second collision risk level (medium risk, ΨL≤Ψ<ΨH), and third collision risk level (high risk, Ψ≥ΨH). For each level, the optimal control quantity is determined through a cost function based on risk assessment, and the strategy instruction is output, such as... Figure 4 As shown.

[0049] (1) For the first collision risk level (low risk), the current driving state is maintained under the current collision risk level, and only the conventional trajectory optimization is performed. There is no active braking or lateral avoidance. Only when the risk value is close to ΨL, a slight longitudinal speed adjustment is made (such as reducing speed by 0~5km / h by releasing the accelerator). When using the cost function to quantize and solve for the optimal control command sequence that minimizes the cost function J, the cost function adopts an initial weight configuration, and the friction change sensitivity coefficient λ takes the minimum value (2.0). ,because Minimal, Qk≈Q0, lateral position deviation and heading angle deviation are optimized with conventional weights; To minimize the lateral position deviation Δyk and the heading angle deviation Δ With k as the core, the control input weight ρ is kept at 0.01 to avoid small adjustments causing vehicle bumps. The control input is taken within a normal range (steering angle within ±1°, no significant change in throttle opening, and brake opening of 0). Only a small trajectory correction amount (such as steering angle ±0.5°) is output to keep the vehicle driving in the center of the original lane without any active intervention.

[0050] (2) For the second collision risk level (medium risk), under the current collision risk level, the control strategy is: warning + mild braking + first lateral avoidance / first longitudinal adjustment. Mild braking is soft braking by the drive-by-wire retarder, first lateral avoidance is obstacle avoidance within the lane, and first longitudinal adjustment is uniform speed reduction to increase following distance.

[0051] The weight of lateral position deviation in the cost function is increased because lane avoidance is required, and small fluctuations in road surface friction are considered. The friction change sensitivity coefficient λ is taken as a conventional value (2.5). Slight increase (e.g., uneven water accumulation on roads during heavy rain), Qk increases by 10% to 20% compared to Q0.

[0052] When solving for the optimal control command sequence that minimizes the cost function J, the control input weight ρ is appropriately reduced (from 0.01 to 0.005), allowing for a slightly larger control input (such as a steering angle of ±5°) to ensure the effectiveness of the avoidance action.

[0053] Balancing "trajectory correction (avoidance)" and "braking deceleration" ensures that the vehicle avoids obstacles within the lane while maintaining smooth deceleration during gentle braking (without sudden braking). The control input is limited to the range of medium-risk actions (gentle braking deceleration -0.5~-1.0m / s², lateral avoidance steering angle ±5~±10°, steering angular velocity ≤10° / s). Output control commands, such as: brake opening of 30% + steering angle + 6° + throttle reduction to 0, to achieve gradual deceleration and obstacle avoidance within the lane, and all control quantities are the optimal values ​​solved by MPC, avoiding vehicle lateral deviation due to uncoordinated actions.

[0054] (3) For the third collision risk level (high risk), the control strategy is: immediate intervention + emergency braking + second lateral avoidance / second longitudinal adjustment. The braking priority is to use the combined braking of cylinder braking and wire-controlled retarder (ABS hard braking is prohibited to prevent sideslip on low friction road surfaces). The avoidance is selected according to the lateral passable space (obstacle avoidance in this lane / obstacle avoidance across lanes). The longitudinal adjustment is to reduce speed to a stop at full speed (when there is no avoidance space). When solving for the optimal control command sequence that minimizes the cost function J, the lateral position deviation weight in the cost function is maximized and the control input is increased to adapt to the dual constraints of emergency obstacle avoidance and low-friction road surface. The friction change sensitivity coefficient λ is set to its maximum value (3.0), regardless of... Size, Q k It improves upon Q0 by 20%~30%, allowing the algorithm to focus on lateral position deviation correction and ensure accurate obstacle avoidance trajectory; it significantly reduces the control input weight ρ (from 0.01 to 0.001), allowing for larger control input amounts (such as steering angle ±15~±25°), prioritizing obstacle avoidance safety and then considering handling smoothness.

[0055] With collision avoidance as the core objective, the system minimizes the distance between the vehicle and obstacles while constraining braking deceleration and steering angle to within the dynamic limits of low-friction road surfaces (e.g., braking deceleration ≤ -1.5 m / s² on compacted snow). Control inputs are limited to high-risk maneuvers and are categorized into two types of constraints: those with and without obstacle avoidance space. When there is lateral space to avoid a collision: braking deceleration 0~-1.0m / s² (to allow time for avoidance), steering angle ±10~±25°, steering angular velocity ≤5° / s (reduce angular velocity in extreme weather to prevent skidding); In the absence of lateral clearance: braking deceleration -1.0~-3.0 m / s², steering angle 0° (lane keeping, avoid erratic steering); The combination of output braking and obstacle avoidance control quantities, such as: 80% cylinder brake opening + 50% drive-by-wire retarder brake opening + steering angle +18°, ensures that skidding and lock-up do not occur while avoiding obstacles.

[0056] Optionally, in some embodiments, when controlling the vehicle according to the target control strategy, the method includes: evaluating the risk potential gradient of the vehicle; constructing a virtual repulsion field based on the risk potential gradient of the vehicle; planning an initial path of the vehicle based on the repulsion field; and filtering out regions where the magnitude of the risk potential gradient is greater than a preset threshold to obtain a processed path; and adjusting the processed path according to a near-end strategy optimization algorithm to obtain an obstacle avoidance path for the vehicle, so as to control the vehicle to drive according to the obstacle avoidance path.

[0057] Optionally, in some embodiments, the risk potential gradient is: ; in, Ψ(x,t) represents the risk potential gradient of a vehicle at position x and time t, where x i Let x be the position of the i-th obstacle, and x be the position of the vehicle. ||x – x i || represents the Euclidean distance between the vehicle and the obstacle, e -β(t-ti) The risk decays exponentially with time t, where β is the decay coefficient and α is the risk level.i Let t be the risk weight coefficient for the i-th obstacle. i t is the time when the i-th obstacle is first detected.

[0058] Specifically, the system calculates the gradient of the overall risk value Ψ(x,t) with respect to the vehicle's current position to obtain the risk potential gradient. Ψ(x,t), in calculating the risk potential gradient After Ψ(x,t), a virtual repulsive force field is constructed based on the vehicle risk potential gradient, specifically: The core direction of the basic repulsive force is based on the negative gradient of the risk potential gradient, ensuring that the repulsive force points in a safe direction away from the obstacle; the magnitude of the basic repulsive force is positively correlated with the gradient magnitude, and the larger the gradient magnitude, the stronger the basic repulsive force.

[0059] Calculate the relative velocity vector sum based on the relative motion states of the vehicle and each obstacle. , The relative velocity vector sum between each obstacle and the vehicle is given, and a relative velocity coefficient k is introduced, which is expressed by the formula F = - A virtual repulsive force field is constructed; if an obstacle approaches the vehicle, Δvi increases and the repulsive force is enhanced accordingly, avoiding the inability to avoid obstacles in time due to relative motion of dynamic obstacles, and adapting to scenarios where sensor detection accuracy is reduced under extreme weather conditions.

[0060] Furthermore, based on extreme weather types and road surface friction coefficients, the attenuation coefficient β of the risk potential function is dynamically adjusted: for low-friction road surfaces (such as snow), the β value is reduced to prolong the risk memory time, expand the range of influence of the repulsion force, and trigger obstacle avoidance in advance; for high-friction road surfaces, the β value is appropriately increased to reduce the range of influence, ensure driving smoothness, and balance obstacle avoidance effect with path smoothness.

[0061] Based on obstacle information (position / speed / type) and vehicle status information (position / speed) collected in real time by multiple sensors, the risk potential gradient is recalculated and the repulsive force field distribution is updated every 50ms. If the obstacle disappears, moves away, or the vehicle's driving position changes, the magnitude and direction of the repulsive force in the corresponding area are adjusted synchronously to ensure that the force field is highly matched with the real-time environment.

[0062] Further, an improved potential field method combined with a global partitioning of the Venn diagram is used to plan the initial path. Simultaneously, high-risk regions are filtered out using a risk potential gradient magnitude threshold to correct and smooth the initial path, resulting in a processed path that meets basic safety constraints. Specific implementation steps are as follows: The Veno diagram is used to divide the driving area in front of the vehicle into spaces, eliminate impassable areas occupied by road boundaries and static obstacles, clarify the effective driving space of the vehicle, and avoid the path from exceeding the road range or entering the physical obstacle area.

[0063] Taking the vehicle's current position as the starting point of the path and the safe point in front of the center of the lane within the effective driving space as the ending point of the path, a target gravity is introduced. The direction of the target gravity is towards the ending point, and its magnitude decreases slowly as the distance between the vehicle and the ending point decreases.

[0064] By vector synthesis of the repulsive force in the repulsive field and the target gravity, the final direction of motion of the vehicle at each position in the driving space is obtained, ensuring that the path extends in a safe direction while adhering to the center of the lane.

[0065] Based on the synthesized motion direction, a global initial path is generated using an improved potential field method. During the generation process, vehicle kinematic constraints (maximum steering angle, steering angular velocity, and lateral acceleration) are followed to ensure that the initial path is a trajectory that the vehicle can actually execute, without any unachievable path features such as sharp turns or corners. The core requirements of the initial path are to minimize the cumulative repulsive force and optimize the target's gravitational guidance, thereby achieving the basic dual objectives of obstacle avoidance and path tracking.

[0066] Furthermore, based on the vehicle's current speed, road friction coefficient, and visibility in extreme weather, a preset threshold for the risk potential gradient amplitude is dynamically set: the higher the vehicle speed, the lower the road friction coefficient, and the worse the visibility, the lower the threshold is set; the threshold is the critical value for a vehicle to enter a high-risk area, and areas exceeding this threshold are judged as high-risk restricted areas (vehicles are prone to collisions or instability after entering).

[0067] The initial global path is traversed point by point, the risk potential gradient magnitude corresponding to each path point is calculated, and the magnitude is compared with the preset threshold. All path points with gradient magnitudes greater than the preset threshold are marked, and the road segments in the initial path that fall into the high-risk restricted area are identified.

[0068] For marked high-risk path points, offset them in a safe direction with the least repulsive force and a gradient amplitude below the threshold. The offset distance is adjusted according to the road surface friction coefficient (the offset distance is appropriately increased on low-friction roads to prevent vehicles from approaching the risk area). During the offset process, the offset direction of adjacent path points is kept continuous to avoid abrupt changes in the path.

[0069] Curvature smoothing optimization is performed on all path points that have completed the safe offset, eliminating sharp corners and abrupt changes in the path to make the curvature change of the path uniform, meeting the driving smoothness requirements of the vehicle in extreme weather conditions, and avoiding vehicle sideslip and loss of control due to abrupt changes in path curvature.

[0070] After filtering, offset correction and smoothing of high-risk areas, the processed path is obtained, which is the processed path that meets the basic safety constraints, and provides a basic reference trajectory for subsequent fine-tuning of the PPO algorithm.

[0071] Finally, the processed path is adjusted based on the Proximal Policy Optimization (PPO) algorithm to obtain the final obstacle avoidance path. The PPO algorithm is used for local fine-tuning and dynamic weight optimization of the processed path. Combined with the dynamic window method and vehicle dynamics constraints, the PPO algorithm output is used as the weights of the path evaluation function, making the path adaptable to complex environments such as extreme weather. Ultimately, a safe, smooth, and optimal obstacle avoidance path that adapts to vehicle dynamics limits is obtained. Specific implementation steps are as follows: The processed path is used as the initial reference trajectory for the PPO algorithm. At the same time, the required environmental and vehicle parameters are input, including: risk potential gradient, virtual repulsive force field distribution, real-time road friction coefficient, current vehicle status (speed / acceleration / heading angle), real-time obstacle information (location / speed / type), and extreme weather characteristics (visibility / precipitation intensity).

[0072] Start the four core networks of the PPO algorithm—current policy network, old policy network, value function network, and target value network—and complete the network parameter initialization, such as... Figure 5 As shown; simultaneously, based on the environmental complexity of extreme weather, the core hyperparameters of the PPO algorithm are initialized, and the specific initialization values ​​and numerical adjustment rules are as follows: Table 1

[0073] Specifically, the output of the model trained based on the PPO algorithm is used as the evaluation function of the dynamic window method, which restricts the vehicle's speed and angular velocity space. The evaluation function is usually based on geometric factors such as distance and speed to ensure that the model output meets the vehicle's dynamic limits. The weights of the path evaluation function are further optimized using the PPO algorithm.

[0074] Using the processed path as a reference, the PPO algorithm interacts with the real-time driving environment, sampling around the reference trajectory to generate multiple local candidate correction trajectories. The state (path point position / gradient magnitude / repulsion force magnitude), action (trajectory correction direction / magnitude), and reward value of each trajectory are collected to form a training sample set.

[0075] Furthermore, the state value V of each sample is calculated using a value function network. π (s) and action value Q π (s,a), obtaining the dominant function Q π (s,a) is the action value function of taking action a in state s under policy π, while V π (s) is the state value function, which represents the expected reward that can be obtained in state s according to policy π. The advantage function helps us understand whether a certain action is better than the average action.

[0076] Further, based on the importance sampling method, the probability ratio of the new and old strategies is calculated, and the PPO objective function is constructed. The objective function can be expressed as: ; in It is the probability ratio of the new strategy to the old strategy. This is a hyperparameter that controls the magnitude of policy updates. The objective function uses a clipping mechanism to ensure that policy updates are not too aggressive.

[0077] The current policy network is optimized using the gradient ascent method, and the weight parameters of the path evaluation function are updated. At the same time, the KL divergence between the old policy network and the current policy network is calculated. If the divergence exceeds the threshold, the update is stopped to ensure that the new policy does not deviate too much from the old policy and to ensure the stability of the weight optimization. The value function network and the target value network work together. The target value network provides a stable value benchmark for the value function network and guides the direction of weight update.

[0078] During the weight update process, the hyperparameters are dynamically adjusted based on the training effect: if the training is slow, the learning rate is increased appropriately; if the training is unstable, the gradient clipping value is increased; if the environmental complexity changes, the policy clipping range is adjusted. The learning rate adopts a gradual decay strategy, which slowly decreases as the number of training steps increases to ensure convergence in the later stages of training.

[0079] By substituting the optimized weight parameters from the PPO algorithm training into the initial evaluation function, an optimized path evaluation function adapted to the real-time environment is obtained. Under extreme weather conditions, this function will automatically increase the weight of obstacle avoidance safety distance.

[0080] Based on the velocity space constraint of the dynamic window method, multiple local candidate trajectories are generated by sampling around the processed path. The optimized evaluation function is used to give a comprehensive score to each candidate trajectory. The higher the score, the more the trajectory meets the requirements of "safety, smoothness and lane fit".

[0081] The candidate trajectory with the highest comprehensive score is subjected to vehicle dynamics limit verification. The steering angle, lateral acceleration, and braking deceleration of each point on the trajectory are verified one by one to see if they are within the dynamic limits of the vehicle under the current road friction coefficient. If they exceed the limits, the trajectory is slightly modified until all parameters meet the constraints.

[0082] The trajectory optimized by the PPO algorithm and verified by dynamics serves as the vehicle's final obstacle avoidance path and can be directly used as a trajectory reference for the vehicle's lateral and longitudinal control.

[0083] To enable those skilled in the art to further understand the vehicle control method of the embodiments of this application, the following detailed description is provided in conjunction with specific embodiments, such as... Figure 6 As shown.

[0084] First, the system reads and updates sensor data to obtain raw information about the vehicle's environment and its current state. Based on this information, state estimation is performed, such as estimating the ground friction coefficient and constructing or updating a dynamic risk model, thereby calculating or estimating the vehicle's current environment and its dynamic risk level. Based on the risk level, model predictive control methods are used to solve for the optimal control parameters for the vehicle's motion. Finally, control commands (throttle, brake, steering) under different risk levels are output to the actuators, completing one control cycle. Under conditions of no abnormalities or malfunctions, the next control cycle is initiated.

[0085] The vehicle control method proposed in this application involves acquiring the vehicle's current environmental perception information; determining the vehicle's current collision risk level and calculating the current road surface friction coefficient based on the current environmental perception information; determining a cost function based on the current road surface friction coefficient, lateral position deviation, heading angle deviation, and control input; determining the vehicle's target control strategy based on the current collision risk level and the cost function; and controlling the vehicle according to the target control strategy. This solves the problems of current sensor solutions experiencing performance degradation in extreme weather conditions and existing vehicle control strategies being insufficient to cope with extreme weather scenarios. By avoiding reliance on specific sensor configurations, robust autonomous driving control under extreme weather conditions is achieved, improving vehicle stability during driving.

[0086] Next, the vehicle control system proposed according to the embodiments of this application is described with reference to the accompanying drawings.

[0087] Figure 7 This is a block diagram of the vehicle control system according to an embodiment of this application.

[0088] like Figure 7 As shown, the vehicle's control system 10 includes: an acquisition module 100, a determination module 200, and a control module 300.

[0089] The acquisition module 100 is used to acquire the vehicle's current environmental perception information; the determination module 200 is used to determine the vehicle's current collision risk level and calculate the current road surface friction coefficient based on the current environmental perception information, and determine the cost function based on the current road surface friction coefficient, lateral position deviation, heading angle deviation and control input; the control module 300 is used to determine the vehicle's target control strategy based on the current collision risk level and cost function, and control the vehicle according to the target control strategy.

[0090] Optionally, in some embodiments, the determining module 200 is specifically configured to: calculate the risk value of the vehicle based on the vehicle's environmental perception information; if the risk value is less than a first risk threshold, determine the vehicle's collision risk level as a first collision risk level; if the risk value is greater than or equal to the first risk threshold and less than a second risk threshold, determine the vehicle's collision risk level as a second collision risk level, wherein the risk level of the second collision risk level is higher than the risk level of the first collision risk level; if the risk value is greater than or equal to the second risk threshold, determine the vehicle's collision risk level as a third collision risk level, wherein the risk level of the third collision risk level is higher than the risk level of the second collision risk level.

[0091] Optionally, in some embodiments, the control module 300 is specifically configured to: if the vehicle's risk level is a first collision risk level, control the vehicle to maintain its current state; if the vehicle's risk level is a second collision risk level, control the vehicle to perform a preset easing braking action while simultaneously performing a first lateral avoidance control action or a first longitudinal adjustment control action; if the vehicle's risk level is a third collision risk level, control the vehicle to perform a preset emergency braking action while simultaneously performing a second lateral avoidance control action or a second longitudinal adjustment control action.

[0092] Optionally, in some embodiments, when controlling the vehicle according to the target control strategy, the control module 300 is further configured to: evaluate the risk potential gradient of the vehicle; construct a virtual repulsion field based on the risk potential gradient of the vehicle, plan the initial path of the vehicle based on the repulsion field, and filter out areas where the magnitude of the risk potential gradient is greater than a preset threshold to obtain a processed path; adjust the processed path according to the near-end strategy optimization algorithm to obtain the obstacle avoidance path of the vehicle, so as to control the vehicle to drive according to the obstacle avoidance path.

[0093] Optionally, in some embodiments, the risk value is: ; Where Ψ(x,t) represents the risk value of the vehicle at position x and time t, x i Let x be the position of the i-th obstacle, x be the position of the vehicle, and t be the position of the obstacle. i Let ||x – x be the time when the i-th obstacle is first detected. i || 2 e represents the squared Euclidean distance between the vehicle and the obstacle. -β(t-ti) The risk decays exponentially over time, where β is the decay coefficient and α is the risk level. i Let be the risk weight coefficient for the i-th obstacle.

[0094] Optionally, in some embodiments, the risk potential gradient is: ; in, Ψ(x,t) represents the risk potential gradient of a vehicle at position x and time t, where x i Let x be the position of the i-th obstacle, and x be the position of the vehicle. ||x – x i || represents the Euclidean distance between the vehicle and the obstacle, e -β(t-ti) The risk decays exponentially with time t, where β is the decay coefficient and α is the risk level. i Let t be the risk weight coefficient for the i-th obstacle. i t is the time when the i-th obstacle is first detected.

[0095] Optionally, in some embodiments, the cost function is:

[0096]

[0097] Where J is the cost function, Δy k For the lateral position error, Δ k For heading angle error, For the control input, Q is the weight of the lateral position deviation, R is the weight of the heading angle deviation, and ρ is the optimization judgment condition for minimizing the control input weights. Let || be the road surface friction coefficient, || || 2 This represents the square of the vector's magnitude. Q 0 represents the weight matrix of the cost function at the initial time step. Let Q be the weight matrix of the cost function at time k, λ be the friction sensitivity coefficient, and Q be the weight matrix of the cost function at time k. k It is a subarray of the Q array.

[0098] It should be noted that the foregoing explanation of the vehicle control method embodiment also applies to the vehicle control system of this embodiment, and will not be repeated here.

[0099] The vehicle control system proposed in this application acquires the vehicle's current environmental perception information; determines the vehicle's current collision risk level and calculates the current road surface friction coefficient based on the current environmental perception information; and determines a cost function based on the current road surface friction coefficient, lateral position deviation, heading angle deviation, and control input; determines the vehicle's target control strategy based on the current collision risk level and the cost function, and controls the vehicle according to the target control strategy. This solves the problems of current sensor solutions experiencing performance degradation in extreme weather conditions and existing vehicle control strategies being insufficient to cope with extreme weather scenarios. By avoiding reliance on specific sensor configurations, robust autonomous driving control under extreme weather conditions is achieved, improving vehicle stability during driving.

[0100] Figure 8A schematic diagram of the structure of a vehicle provided in an embodiment of this application. The vehicle may include: The memory 801, the processor 802, and the computer program stored on the memory 801 and capable of running on the processor 802.

[0101] When processor 802 executes the program, it implements the vehicle control method provided in the above embodiments.

[0102] Furthermore, the vehicle also includes: Communication interface 803 is used for communication between memory 801 and processor 802.

[0103] The memory 801 is used to store computer programs that can run on the processor 802.

[0104] The memory 801 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0105] If the memory 801, processor 802, and communication interface 803 are implemented independently, then the communication interface 803, memory 801, and processor 802 can be interconnected via a bus to complete communication between them. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized into address buses, data buses, control buses, etc. For ease of representation, Figure 8 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0106] Optionally, in a specific implementation, if the memory 801, processor 802, and communication interface 803 are integrated on a single chip, then the memory 801, processor 802, and communication interface 803 can communicate with each other through an internal interface.

[0107] The processor 802 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0108] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the vehicle control method described above.

[0109] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0110] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0111] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0112] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable storage medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable storage medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable storage medium could be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0113] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0114] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0115] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0116] The computer-readable storage medium mentioned above may be a read-only memory, a magnetic disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A method for controlling a vehicle, characterized in that, Includes the following steps: Obtain the vehicle's current environmental perception information; The vehicle's current collision risk level is determined based on the current environmental perception information, and the current road surface friction coefficient is calculated. A cost function is then determined based on the current road surface friction coefficient, lateral position deviation, heading angle deviation, and control input. The target control strategy for the vehicle is determined based on the current collision risk level and the cost function, and the vehicle is controlled according to the target control strategy.

2. The method according to claim 1, characterized in that, Determining the current collision risk level of the vehicle based on the current environmental perception information includes: Calculate the risk value of the vehicle based on the vehicle's environmental perception information; If the risk value is less than the first risk threshold, the collision risk level of the vehicle is determined to be the first collision risk level. If the risk value is greater than or equal to the first risk threshold and less than the second risk threshold, then the collision risk level of the vehicle is determined to be the second collision risk level, wherein the risk level of the second collision risk level is higher than the risk level of the first collision risk level. If the risk value is greater than or equal to the second risk threshold, the collision risk level of the vehicle is determined to be the third collision risk level, wherein the risk level of the third collision risk level is higher than that of the second collision risk level.

3. The method according to claim 2, characterized in that, The current collision risk level is the first collision risk level. The step of determining the target control strategy for the vehicle based on the current collision risk level and the cost function, and controlling the vehicle according to the target control strategy, includes: If the risk level of the vehicle is the first collision risk level, then control the vehicle to maintain its current state; If the risk level of the vehicle is the second collision risk level, then while controlling the vehicle to perform a preset easing braking action, a first lateral avoidance control action or a first longitudinal adjustment control action is performed. If the risk level of the vehicle is the third collision risk level, then while controlling the vehicle to perform a preset emergency braking action, a second lateral avoidance control action or a second longitudinal adjustment control action is performed.

4. The method according to claim 2, characterized in that, When controlling the vehicle according to the target control strategy, the following are included: Assess the risk potential gradient of the vehicle; A virtual repulsive force field is constructed based on the risk potential gradient of the vehicle. An initial path of the vehicle is planned based on the repulsive force field, and regions where the magnitude of the risk potential gradient is greater than a preset threshold are filtered out to obtain the processed path. The processed path is adjusted according to the near-end strategy optimization algorithm to obtain the obstacle avoidance path of the vehicle, so as to control the vehicle to drive according to the obstacle avoidance path.

5. The method according to claim 2, characterized in that, The risk value is: ； Where Ψ(x,t) represents the risk value of the vehicle at position x and time t, x i Let x be the position of the i-th obstacle, x be the position of the vehicle, and t be the position of the obstacle. i Let ||x – x be the time when the i-th obstacle is first detected. i || 2 e represents the squared Euclidean distance between the vehicle and the obstacle. -β(t-ti) The risk decays exponentially over time, where β is the decay coefficient and α is the risk level. i Let be the risk weight coefficient for the i-th obstacle.

6. The method according to claim 4, characterized in that, The risk potential gradient is: ； in, Ψ(x,t) represents the risk potential gradient of the vehicle at position x and time t, where x i Let x be the position of the i-th obstacle, and x be the position of the vehicle. ||x – x i || represents the Euclidean distance between the vehicle and the obstacle, e -β(t-ti) The risk decays exponentially with time t, where β is the decay coefficient and α is the risk level. i Let t be the risk weight coefficient for the i-th obstacle. i t is the time when the i-th obstacle is first detected.

7. The method according to claim 1, characterized in that, The cost function is: Where J is the cost function, Δy k For the lateral position error, Δ k For heading angle error, For the control input, Q is the weight of the lateral position deviation, R is the weight of the heading angle deviation, and ρ is the optimization judgment condition for minimizing the control input weights. Let || be the road surface friction coefficient, || || 2 This represents the square of the vector's magnitude. Q 0 represents the weight matrix of the cost function at the initial time step. Let Q be the weight matrix of the cost function at time k, λ be the friction sensitivity coefficient, and Q be the weight matrix of the cost function at time k. k It is a subarray of the Q array.

8. A vehicle control system, characterized in that, include: The acquisition module is used to acquire the vehicle's current environmental perception information; The determination module is used to determine the current collision risk level of the vehicle and calculate the current road surface friction coefficient based on the current environmental perception information, and to determine the cost function based on the current road surface friction coefficient, lateral position deviation, heading angle deviation and control input; The control module is used to determine the target control strategy of the vehicle based on the current collision risk level and the cost function, and to control the vehicle according to the target control strategy.

9. A vehicle, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the program to implement the vehicle control method as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the vehicle control method as described in any one of claims 1-7.

Images