Vehicle obstacle avoidance control method, device and equipment, vehicle and storage medium

Abstract

The application provides a vehicle obstacle avoidance control method, device, equipment, vehicle and storage medium, and relates to the technical field of intelligent driving control of vehicles. The method comprises the following steps: acquiring vehicle state information and obstacle information in the direction of vehicle travel; based on a vehicle kinematics model considering Ackerman steering characteristics, predicting a vehicle body motion envelope of the vehicle in a future time period according to the vehicle state information; wherein the vehicle body motion envelope is used to indicate a space region passed by the vehicle when the vehicle travels along a predicted trajectory; if it is determined that the ego vehicle has a collision risk according to the obstacle information and the vehicle body motion envelope, then an active intervention control instruction for avoiding the collision risk is generated according to the collision risk intensity. By adopting the scheme of the application, the potential interference risk in scenes such as low-speed turning, warehouse entry and narrow passage can be more accurately identified, the timeliness of early warning and the adaptability of intervention are improved, and the anti-collision effect, the passage safety and the driving smoothness are enhanced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent driving control of vehicles, and in particular to a vehicle obstacle avoidance control method, device, equipment, vehicle and storage medium. BACKGROUND

[0002] In low-speed moving scenarios such as underground garages, community alleys and roadside parking spaces, vehicles usually use technologies such as reversing radar, panoramic image and automatic emergency braking for obstacle perception and anti-collision control.

[0003] However, existing solutions focus more on real-time perception or single braking intervention, and have limited identification capability for low static obstacles, and insufficient prediction of the actual swept trajectory of the vehicle body when turning at low speed, which can easily cause early warning lag; at the same time, the control strategy has weak adaptability to different vehicle power architectures, and it is difficult to balance the effectiveness of intervention and driving smoothness.

[0004] Therefore, how to improve the prediction accuracy and intervention adaptability of the risk of collision with static obstacles during the low-speed movement of the vehicle to achieve effective protection has become a technical problem to be solved. SUMMARY

[0005] Embodiments of the present application provide a vehicle obstacle avoidance control method, device, equipment, vehicle and storage medium to solve the above technical problems. The method faces the obstacle avoidance control needs in the low-speed movement of the vehicle, comprehensively considers the running state of the vehicle and the obstacle situation in the direction of travel, makes an early decision on the collision risk in the future driving process of the vehicle, and implements corresponding active intervention according to the risk level, thereby improving the collision risk prediction capability and control adaptability of the vehicle in low-speed complex scenarios.

[0006] In a first aspect, the present application provides a vehicle obstacle avoidance control method, which comprises:

[0007] obtaining vehicle state information and obstacle information in the direction of travel of the vehicle;

[0008] predicting a vehicle body motion envelope of the vehicle in a future time period according to the vehicle state information based on a vehicle kinematics model considering Ackerman steering characteristics; wherein the vehicle body motion envelope is used to indicate a spatial region passed by the vehicle along a predicted trajectory;

[0009] if it is determined that there is a collision risk of the ego vehicle according to the obstacle information and the vehicle body motion envelope, generating an active intervention control instruction for avoiding the collision risk according to the collision risk intensity.

[0010] In a possible implementation, the vehicle body motion envelope of the vehicle in the future time period is predicted according to the vehicle state information based on the vehicle kinematics model considering Ackerman steering characteristics, which comprises:

[0011] based on the vehicle state information, predicting a driving trajectory of the vehicle in a future time period by using a vehicle kinematics model considering Ackermann steering characteristics;

[0012] calculating, according to geometric parameters of the vehicle and in combination with the Ackermann steering characteristics, an inner wheel difference offset between front wheel trajectories and rear wheel trajectories of the vehicle when driving along the driving trajectory, wherein the geometric parameters at least include a wheelbase and a track;

[0013] determining, according to the driving trajectory and the inner wheel difference offset, swept paths of each part of a vehicle body in a driving process, and forming a vehicle body motion envelope by the swept paths of each part.

[0014] In a possible implementation, the determining that the ego vehicle has a collision risk according to the obstacle information and the vehicle body motion envelope comprises:

[0015] performing geometric overlap detection on a two-dimensional plane between a space region covered by the vehicle body motion envelope and an obstacle profile in the obstacle information;

[0016] if an intersection region between the vehicle body motion envelope and the obstacle profile on the two-dimensional plane is detected, determining that the collision risk exists.

[0017] In a possible implementation, the generating an active intervention control instruction for avoiding the collision risk according to a collision risk intensity comprises:

[0018] calculating an intrusion depth and / or a predicted collision time of the intersection region between the vehicle body motion envelope and the obstacle profile on the two-dimensional plane;

[0019] determining a current collision risk intensity and a corresponding initial intervention level according to the intrusion depth and / or the predicted collision time;

[0020] starting from an intervention measure corresponding to the initial intervention level, sequentially evaluating intervention measures corresponding to each intervention level from low to high in terms of intervention intensity, and determining a first intervention measure capable of avoiding the collision risk as a current active obstacle avoidance measure;

[0021] generating the active intervention control instruction according to the current active obstacle avoidance measure.

[0022] In a possible implementation, the starting from an intervention measure corresponding to the initial intervention level, sequentially evaluating intervention measures corresponding to each intervention level from low to high in terms of intervention intensity, and determining a first intervention measure capable of avoiding the collision risk as a current active obstacle avoidance measure comprises:

[0023] obtaining an intervention measure corresponding to the initial intervention level; wherein different intervention levels include corresponding intervention measures in order from low to high intervention intensity;

[0024] predicting a vehicle body motion envelope of the ego vehicle after performing the intervention measure corresponding to the currently evaluated intervention level based on a vehicle kinematic model considering Ackerman steering characteristics;

[0025] if it is determined that the vehicle body motion envelope and the obstacle profile do not have an intersection region in a two-dimensional plane, determining the intervention measure corresponding to the currently evaluated intervention level as the current active obstacle avoidance measure;

[0026] if it is determined that the vehicle body motion envelope and the obstacle profile still have an intersection region in the two-dimensional plane, continuing to evaluate intervention measures corresponding to next higher intervention levels until an intervention level that can avoid the collision risk is found;

[0027] if the collision risk cannot be avoided after traversing all intervention levels, triggering automatic emergency braking to stop the vehicle.

[0028] In a possible implementation, according to the drive system configuration of the vehicle, the different intervention levels include corresponding intervention measures in order from low to high intervention intensity, including at least two of the following:

[0029] a first intervention level, the corresponding intervention measure including: controlling an electric power steering system to apply a steering correction torque;

[0030] a second intervention level, the corresponding intervention measure including: controlling the electric power steering system to perform steering correction and controlling a braking system to apply a braking force to the wheels;

[0031] a third intervention level, the corresponding intervention measure including: controlling a torque difference between left and right drive wheels to generate a yaw moment;

[0032] a fourth intervention level, the corresponding intervention measure including: triggering automatic emergency braking to stop the vehicle.

[0033] In a possible implementation, in the process of evaluating intervention measures corresponding to each intervention level, further comprising:

[0034] determining a desired yaw rate change amount required to eliminate the intersection region according to the spatial position and the intrusion depth of the intersection region between the vehicle body motion envelope and the obstacle profile;

[0035] determining a target actuator corresponding to the currently evaluated intervention level according to the intervention level;

[0036] Based on the expected change in yaw rate, determine the control quantity that the target actuator should execute.

[0037] In one possible implementation, the method further includes:

[0038] Obtain the vehicle's location information;

[0039] The location information is compared with the stored collision risk geographic area data. If it is determined that the vehicle is in or is approaching the collision risk geographic area, a warning message is generated and output to the driver.

[0040] Secondly, embodiments of this application provide a vehicle obstacle avoidance control device, the device comprising:

[0041] The first processing unit is used to acquire vehicle status information and obstacle information in the direction of vehicle travel;

[0042] The second processing unit is used to predict the vehicle's motion envelope in a future time period based on the vehicle kinematics model that takes into account Ackermann steering characteristics and the vehicle state information; wherein the vehicle motion envelope is used to indicate the spatial area traversed by the vehicle along the predicted trajectory.

[0043] The third processing unit is used to generate an active intervention control command to avoid the collision risk if it is determined that the vehicle has a collision risk based on the obstacle information and the vehicle motion envelope.

[0044] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0045] The memory stores computer-executed instructions;

[0046] The processor executes computer execution instructions stored in the memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0047] Fourthly, embodiments of this application provide a vehicle for implementing the first aspect and / or various possible implementations of the first aspect.

[0048] Fifthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0049] In a sixth aspect, an embodiment of the present application provides a computer program product, comprising a computer program, which, when executed by a processor, is used to implement the first aspect and / or various possible implementation manners of the first aspect.

[0050] Advantages of the present application:

[0051] The vehicle obstacle avoidance control method, device, equipment, vehicle and storage medium provided by the embodiments of the present application can more accurately reflect the actual swept space of the vehicle during low-speed turning and driving, improve the early identification ability of the collision risk of obstacles, and then generate an active intervention control instruction for avoiding the collision risk according to the collision risk intensity when it is determined that there is a collision risk, so as to effectively protect against the risk of collision with static obstacles in a low-speed scenario, and take into account the effectiveness of intervention and driving smoothness. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 A flowchart of a vehicle obstacle avoidance control method provided by an embodiment of the present application is shown in the figure;

[0053] Figure 2 A flowchart of another vehicle obstacle avoidance control method provided by an embodiment of the present application is shown in the figure;

[0054] Figure 3 A flowchart based on positioning early warning provided by an embodiment of the present application is shown in the figure;

[0055] Figure 4 An architectural diagram of a vehicle obstacle avoidance control system provided by an embodiment of the present application is shown in the figure;

[0056] Figure 5 A structural diagram of a vehicle obstacle avoidance control device provided by an embodiment of the present application is shown in the figure;

[0057] Figure 6 A structural diagram of an electronic device provided by an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0058] The embodiments of the present application will be described below with reference to the accompanying drawings and preferred embodiments, and those skilled in the art can easily understand other advantages and effects of the present application from the content disclosed in the specification. The present application can also be implemented or applied by means of other different specific embodiments, and the details in the specification can be modified or changed in various ways based on different views and applications without departing from the spirit of the present application. It should be understood that the preferred embodiments are only for the purpose of illustrating the present application, and are not intended to limit the protection scope of the present application.

[0059] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of this application. Therefore, the drawings only show the components related to this application and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0060] When vehicles move at low speeds in narrow spaces (such as underground garages, alleyways in residential areas, and roadside parking spaces), there is a very high risk of lateral and undercarriage collisions. For example, when reversing into a parking space, the A-pillar obstructs the view and there is a blind spot at the rear of the vehicle, making it difficult for the driver to directly observe the side pillars or protrusions on the ground. When making a right-angle turn, due to the Ackermann steering geometry, there is an inner wheel difference between the rear wheel trajectory and the front wheel trajectory, making it easy for the rear or undercarriage of the vehicle to scrape against curbs or walls. In addition, static obstacles such as mechanical parking spaces and low guardrails are small in size and low in height, making it difficult for traditional sensors (such as reversing radar) to accurately identify them. When drivers rely on rearview mirrors and reversing cameras, they may scrape against these obstacles due to operational errors or delayed reactions.

[0061] Existing vehicle collision avoidance systems are mostly designed for high-speed dynamic collisions. They typically combine technologies such as reversing radar, panoramic imaging, and automatic emergency braking to detect obstacles and provide risk warnings. Their basic principle is to determine the presence of obstacles near the vehicle based on distance or image information collected by sensors, and to issue an alarm or brake intervention when preset conditions are met. This type of solution is effective in general reversing or close-range warning scenarios. However, when turning at low speeds, entering parking spaces, or exiting narrow areas, the actual sweep range of the vehicle's sensors does not equate to the wheel or front-end trajectory directly observed by the driver. Especially under the influence of Ackermann steering characteristics, the rear wheel trajectory and the movement path of the rear side of the vehicle are more prone to interference with curbs, corners, pillars, or low obstacles.

[0062] In view of this, how to more accurately predict the potential interference relationship between the vehicle and obstacles during low-speed vehicle movement, and generate appropriate avoidance control according to the risk level, has become an urgent technical problem to be solved. To solve the above problem, embodiments of this application provide a vehicle obstacle avoidance control method. First, vehicle state information and obstacle information in the vehicle's direction of travel are obtained. Then, based on a vehicle kinematic model considering Ackermann steering characteristics, the vehicle motion envelope in the future time period is predicted. When a collision risk is determined based on obstacle information and vehicle motion envelope, an active intervention control command for avoiding the collision risk is generated according to the collision risk intensity, so as to improve the anti-collision effect in low-speed scenarios.

[0063] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0064] It should be noted that the vehicle obstacle avoidance control method provided in this application can be executed by a vehicle obstacle avoidance control device. This device can be deployed in electronic devices such as the vehicle controller, domain controller, or dedicated processor that communicates with the vehicle wireless communication equipment. This application does not impose any limitations on this. The vehicle obstacle avoidance control device can be implemented through software, hardware, or a combination of both.

[0065] Figure 1 This is a flowchart illustrating a vehicle obstacle avoidance control method provided in an embodiment of this application. Figure 1 As shown in the embodiments of this application, the vehicle obstacle avoidance control method may include:

[0066] S101. Obtain vehicle status information and obstacle information in the vehicle's direction of travel.

[0067] For example, in this step, the vehicle obstacle avoidance control device acquires two types of information: the vehicle's own state information and obstacle information in the vehicle's direction of travel. The vehicle state information is used to characterize the vehicle's current operating state and serves as the basic input for subsequent trajectory prediction and collision risk assessment; the obstacle information is used to characterize the distribution of obstacles in the vehicle's direction of travel and serves as the basis for determining whether there is a collision risk in the spatial area in the predicted direction of travel.

[0068] Vehicle status information may include at least a portion of the following: vehicle speed, steering wheel angle, front wheel angle, yaw rate, vehicle longitudinal acceleration, gear status, brake pedal status, drive torque request, vehicle current position, and vehicle body geometry parameters. This vehicle status information can be provided by the vehicle's existing status acquisition modules and processed within the coordinate system and time reference corresponding to the vehicle's current control. For example, vehicle speed can be acquired through wheel speed sensors or an onboard inertial measurement unit; steering wheel angle can be acquired through a steering wheel angle sensor; yaw rate can be acquired through an onboard inertial measurement unit; vehicle geometry parameters such as wheelbase and track width can be pre-stored in the vehicle's electronic control unit. Furthermore, for vehicles equipped with rear-wheel steering, the vehicle status information may also include the rear wheel steering angle. The vehicle status information may further include suspension height to determine the vehicle's ground clearance, thereby enabling prediction of bottom impact risk in subsequent steps.

[0069] Obstacle information can be acquired through the environmental perception module installed on the vehicle. This module can fuse data from at least two sensors, including visual cameras, LiDAR, and millimeter-wave radar, to construct a real-time model of the environment surrounding the vehicle and extract the location, size, and outline information of obstacles from this model. Obstacles can be stationary, such as curbs, walls, pillars, low guardrails, stopped vehicles, and road structures; or they can be moving, such as other vehicles or pedestrians. By using multi-sensor fusion, the limitations of single sensors in terms of sensing range, accuracy, or anti-interference capability can be overcome, improving the accuracy and robustness of obstacle detection.

[0070] This step, by acquiring vehicle status information and obstacle information, provides a complete data foundation for subsequent trajectory prediction and collision risk assessment.

[0071] S102. Based on the vehicle kinematics model that considers Ackermann steering characteristics, predict the vehicle's motion envelope in the future time period according to the vehicle state information; wherein, the vehicle motion envelope is used to indicate the spatial area traversed by the vehicle along the predicted trajectory.

[0072] For example, in this step, the vehicle obstacle avoidance control device predicts the vehicle's motion envelope within a future preset time period (e.g., 1 second, 2 seconds, or 3 seconds) based on a vehicle kinematics model that takes into account Ackermann steering characteristics and the vehicle state information obtained in step S101.

[0073] It's important to note that Ackermann steering characteristics are a fundamental property of vehicle steering kinematics. In Ackermann steering geometry, when a vehicle is turning, the trajectories of the front and rear wheels do not coincide; the rear wheel trajectory deviates inwards from the front wheel trajectory, a phenomenon known as "inner wheel difference." When turning or entering / exiting a parking space in confined spaces (such as garages, alleyways, or roadside parking spaces), the presence of this inner wheel difference makes it easy for the rear wheels or sides of the vehicle to scrape or collide with obstacles in blind spots (such as curbs, walls, or pillars). Most existing collision avoidance solutions predict the trajectory of a point mass or longitudinal safety distance, failing to accurately reflect the actual sweep paths of different parts of the vehicle body. Therefore, they struggle to effectively predict and avoid the risk of lateral collisions caused by the inner wheel difference.

[0074] To address the aforementioned issues, this step employs a vehicle kinematics model considering Ackermann steering characteristics for trajectory prediction. Specifically, the vehicle obstacle avoidance control device can use the current moment as the prediction starting point, and based on the previously acquired vehicle state information and vehicle body geometric parameters, predict the vehicle's motion state over a future time period using the vehicle kinematics model considering Ackermann steering characteristics, and determine the corresponding vehicle motion envelope accordingly. In a concrete implementation, the spatial area traversed by the vehicle along the predicted trajectory can be obtained based on the vehicle's position changes, attitude changes, and vehicle body outline range over the future time period. When the vehicle is in a continuous steering state, the control device repeatedly calls the model and re-predicts the vehicle motion envelope over the future time period in each control cycle, ensuring that the vehicle motion envelope is updated in real time with changes in the vehicle state. Optionally, the vehicle motion envelope can be expressed as a polygonal region or an occupied grid, etc., and this embodiment does not impose any limitations.

[0075] This application predicts the vehicle body motion envelope by using a vehicle kinematic model that considers Ackermann steering characteristics, rather than simply predicting the trajectory of a mass or longitudinal safety distance. This accurately reflects the actual sweep paths of various parts of the vehicle body (especially the rear wheels and the sides of the vehicle body) during steering, thus providing a precise spatial basis for subsequent collision risk assessment and effectively solving the problem of predicting lateral and bottom impact risks caused by inner wheel difference.

[0076] Optionally, in one possible embodiment, predicting the vehicle's motion envelope over a future time period based on a vehicle kinematic model that considers Ackermann steering characteristics and vehicle state information may include:

[0077] S11. Based on vehicle state information, use a vehicle kinematics model that considers Ackermann steering characteristics to predict the vehicle's trajectory in the future time period.

[0078] S12. Based on the vehicle's geometric parameters and Ackermann steering characteristics, calculate the inner wheel difference offset between the front wheel track and the rear wheel track when the vehicle is traveling along the driving trajectory; wherein, the geometric parameters include at least the wheelbase and track width.

[0079] S13. Based on the driving trajectory and the inner wheel difference offset, determine the sweep path of each part of the vehicle body during the driving process, and the sweep paths of each part together constitute the motion envelope of the vehicle body.

[0080] For example, when predicting the vehicle's motion envelope over a future time period, the vehicle obstacle avoidance control device can first input the acquired vehicle state information (including at least vehicle speed, steering wheel angle, and yaw rate) into a pre-built vehicle kinematics model. This model considers Ackermann steering geometry and can recursively calculate the vehicle's trajectory over a future time period (e.g., 1 second, 2 seconds, or 3 seconds) based on the vehicle's current motion state. This trajectory can refer to the baseline trajectory of the vehicle's previous reference point (e.g., the vehicle's center of gravity or rear axle center). This baseline trajectory forms the basis for subsequent calculations of the sweep paths of various parts of the vehicle body. Since Ackermann steering geometry describes the inherent relationship between the front and rear wheel trajectories during steering, the trajectory predicted based on this model accurately reflects the vehicle's steering kinematics characteristics, providing an accurate baseline for subsequent inner wheel difference calculations.

[0081] Furthermore, the vehicle obstacle avoidance control device can calculate the inner wheel difference offset between the front and rear wheel trajectories when the vehicle travels along a predicted trajectory, based on the vehicle's geometric parameters (at least wheelbase and track width) and Ackerman steering characteristics. Specifically, in Ackerman steering geometry, when the vehicle turns, the turning radius of the front wheels (especially the outer front wheels) is greater than that of the rear wheels (especially the inner rear wheels), causing the rear wheel trajectories to deviate inward relative to the front wheels; this offset is the inner wheel difference offset. The magnitude of the inner wheel difference offset is related to the vehicle's wheelbase, track width, and the current steering wheel angle (i.e., the front wheel angle). Generally, the longer the wheelbase and the larger the steering wheel angle, the larger the inner wheel difference offset. By quantitatively calculating the inner wheel difference offset, key offset parameters can be provided for subsequently determining the sweep paths of various parts of the vehicle body, ensuring that the predicted motion envelope accurately reflects the influence of Ackerman steering characteristics on the vehicle trajectory.

[0082] Furthermore, the vehicle obstacle avoidance control device can determine the sweep paths of various parts of the vehicle body during driving based on the driving trajectory predicted in S11 (reference trajectory) and the inner wheel difference offset calculated in S12. These sweep paths collectively constitute the vehicle body motion envelope. Specifically, the sweep paths of various parts of the vehicle body can be determined as follows: using the reference trajectory as a reference, and combining the vehicle's geometric dimensions (including length, width, wheelbase, track width, etc.), the offsets and sweep ranges of key collision-prone points such as the two corner points at the front, two corner points at the rear, the middle of both sides of the vehicle body, and points at the bottom of the vehicle are calculated as they follow the reference trajectory. Based on this, the inner wheel difference offset calculated in S12 is superimposed to obtain the actual sweep path of each part during turning. Combining the sweep paths of each part constitutes the vehicle body motion envelope, which is a three-dimensional spatial region (projected onto a two-dimensional plane as a two-dimensional spatial region), used to indicate the entire spatial range traversed by the vehicle while traveling along the predicted trajectory.

[0083] Optionally, when calculating the sweep path of each point on the vehicle's underside, the actual ground clearance of the vehicle's underside is also determined based on the suspension height. As each point on the vehicle's underside moves along a reference driving trajectory, the spatial height of its sweep path is determined by this ground clearance—the smaller the ground clearance, the closer the spatial position of each point on the underside is to the ground, and the higher the risk of spatial interference with ground protrusions or low obstacles. By incorporating the suspension height into the calculation of the underside sweep path, the predicted vehicle motion envelope can more accurately reflect the actual space occupied by the vehicle's underside in the vertical direction, thus providing a precise spatial basis for predicting the risk of underside collisions.

[0084] The above-mentioned optional implementation method can accurately construct the vehicle body motion envelope during the steering process through a three-step progressive approach of baseline driving trajectory prediction, inner wheel difference offset calculation, and vehicle body sweep path synthesis. In particular, it can accurately reflect the rear wheel inner wheel difference effect caused by Ackermann steering characteristics, thereby providing accurate spatial basis for subsequent collision risk assessment. It effectively solves the technical problem in the prior art that it is difficult to warn of lateral and bottom collision risks due to the inability to accurately predict the vehicle body sweep path.

[0085] S103. If it is determined that there is a collision risk to the vehicle based on obstacle information and vehicle motion envelope, then an active intervention control command for avoiding the collision risk is generated based on the intensity of the collision risk.

[0086] For example, in this step, the vehicle obstacle avoidance control device determines whether there is a collision risk based on the obstacle information obtained in step S101 and the predicted vehicle motion envelope in step S102. If a collision risk exists, a corresponding active intervention control command is generated based on the intensity of the collision risk. The collision risk intensity is used to characterize the risk level, and can be determined based on information such as distance, time, or spatial relationships. The active intervention control command is sent directly to the target actuator and may include braking control commands, drive torque limiting commands, steering correction commands, or combinations thereof.

[0087] Specifically, the process of determining whether a collision risk exists may include: performing geometric overlap detection on a two-dimensional plane between the spatial area covered by the vehicle's motion envelope and the obstacle outline in the obstacle information. If an intersection area is detected between the vehicle's motion envelope and the obstacle outline on the two-dimensional plane, a collision risk is determined to exist. Unlike existing risk assessment methods based on longitudinal distance threshold comparisons, the geometric overlap detection method can accurately determine whether different parts of the vehicle body will spatially interfere with obstacles, and is particularly suitable for assessing non-longitudinal collision risks such as lateral scrapes and bottom impacts.

[0088] Once a collision risk is identified, the vehicle obstacle avoidance control device can generate corresponding active intervention control commands based on the intensity of the collision risk. The intensity of the collision risk can be quantified by calculating the intrusion depth of the intersection area between the vehicle's motion envelope and the obstacle's outline and / or the predicted collision time. A greater intrusion depth and / or a shorter predicted collision time indicates a higher collision risk intensity.

[0089] Depending on the intensity of the collision risk, the vehicle obstacle avoidance control device can generate active intervention control commands with varying levels of intervention. For example, for low-intensity collision risks, a command can be generated to control the electric power steering system to apply steering correction torque, assisting the driver in avoiding obstacles through minor steering corrections. For medium-intensity collision risks, a command can be generated to control the electric power steering system to make steering corrections while simultaneously controlling the braking system to apply braking force to one side of the wheels, causing the vehicle to decelerate and yaw, thus changing the vehicle's trajectory more significantly. For high-intensity collision risks, if the vehicle has independent drive capability for the left and right wheels (e.g., dual-motor drive or vehicles equipped with an active differential), a command can be generated to control the torque difference between the left and right drive wheels to generate yaw torque, changing the vehicle's trajectory through precise torque vector control. For the highest-intensity collision risk, when none of the above active intervention methods can safely avoid a collision, a command can be generated to trigger automatic emergency braking, bringing the vehicle to a stop and minimizing collision damage.

[0090] Furthermore, when generating active intervention control commands, a step-by-step evaluation approach can be adopted: starting with the initial intervention level corresponding to the current collision risk intensity, the intervention measures corresponding to each intervention level are evaluated step-by-step in order of increasing intervention intensity to determine whether they can avoid the collision risk; the first intervention measure that can avoid the collision risk is identified as the current active obstacle avoidance measure, and the corresponding control command is generated. This approach can avoid excessive intervention affecting the driving experience, while also allowing for timely escalation of intervention intensity when necessary, thus balancing safety and comfort.

[0091] Understandably, after the command is issued, the vehicle obstacle avoidance control device needs to continue to perform rolling prediction and risk reassessment. When the collision risk intensity decreases and falls below the release condition, the intervention can be gradually lifted; when the collision risk intensity continues to rise, the degree of intervention needs to be increased, thus forming a closed-loop control process based on the vehicle motion envelope and the collision risk intensity.

[0092] By detecting the geometric overlap between the vehicle's motion envelope and the obstacle's contour, collision risk can be accurately identified, including lateral and bottom impact risks. By generating progressively graded active intervention control commands based on the intensity of collision risk, appropriate intervention measures can be taken under different risk levels, avoiding unnecessary over-intervention while ensuring effective risk avoidance in high-risk scenarios.

[0093] The vehicle obstacle avoidance control method provided in this application acquires vehicle state information and obstacle information, predicts the vehicle body motion envelope based on a vehicle kinematic model considering Ackermann steering characteristics, and judges the collision risk based on the spatial relationship between the vehicle body motion envelope and the obstacle, generating graded active intervention commands. This achieves active prediction and graded intervention of lateral and bottom collision risks in scenarios such as low-speed turning and entering / exiting parking spaces, improving the safety and reliability of the vehicle when moving in narrow spaces. By combining the vehicle state, obstacle information, and prediction of the vehicle's space occupancy range considering Ackermann steering characteristics in low-speed scenarios, this application can output corresponding execution commands before the vehicle actually contacts the obstacle, enabling the early identification and control of potential collision risks between the vehicle body and obstacles during turning, entering parking spaces, exiting narrow areas, and driving close to the edge.

[0094] Figure 2 This is a flowchart illustrating another vehicle obstacle avoidance control method provided in an embodiment of this application. Figure 2 As shown in the embodiments of this application, the vehicle obstacle avoidance control method may include:

[0095] S201. Obtain vehicle status information and obstacle information in the vehicle's direction of travel.

[0096] It should be noted that the specific implementation of step S201 can be referred to the description of step S101, and will not be repeated here. In short, the vehicle status information includes at least vehicle speed, steering wheel angle, yaw rate, and vehicle geometric parameters (at least wheelbase and track width), and may also include rear wheel steering angle and suspension height, etc.; obstacle information can be obtained by fusing perception data from at least two of the following sensors: visual camera, lidar, and millimeter-wave radar.

[0097] S202. Based on a vehicle kinematics model that considers Ackermann steering characteristics, predict the vehicle's motion envelope in a future time period according to vehicle state information; wherein, the vehicle motion envelope is used to indicate the spatial region traversed by the vehicle along the predicted trajectory.

[0098] It should be noted that the specific implementation of step S202 can be referred to the description of step S102, and will not be repeated here. In the embodiments of this application, the vehicle motion envelope is a spatial region composed of the sweep paths of various parts of the vehicle body (such as the two corner points of the front of the vehicle, the two corner points of the rear of the vehicle, the middle of the two sides of the vehicle body, and various points on the bottom of the vehicle, etc., which are key locations where collisions are likely to occur) in a future time period. It can accurately reflect the actual space encroachment range of the vehicle during the turning process due to the Ackermann steering characteristics (especially the inner wheel difference effect).

[0099] S203. Perform geometric overlap detection on a two-dimensional plane on the spatial region covered by the vehicle motion envelope and the obstacle contour in the obstacle information to determine whether there is an intersection area between the vehicle motion envelope and the obstacle contour on the two-dimensional plane.

[0100] If yes, proceed to step S204; otherwise, return to step S201.

[0101] For example, after predicting the vehicle motion envelope in step S202, the vehicle obstacle avoidance control device needs to determine whether the vehicle motion envelope will spatially interfere with obstacles in the vehicle's direction of travel. Specifically, the vehicle obstacle avoidance control device can project the spatial area covered by the vehicle motion envelope onto a two-dimensional plane to obtain the coverage area of ​​the vehicle motion envelope on the two-dimensional plane; simultaneously, it extracts the contour information of the obstacles (including the position, shape, and size of the obstacles) from the obstacle information obtained in step S201, and maps the obstacle contours onto the two-dimensional plane as well. Then, it detects whether there is an intersection area between the coverage area of ​​the vehicle motion envelope and the obstacle contours on the two-dimensional plane.

[0102] It should be noted that, unlike risk assessment methods based on longitudinal distance threshold comparisons (e.g., determining whether the actual distance between the vehicle and the obstacle is less than a preset safe distance), this application employs a geometric overlap detection method on a two-dimensional plane. Distance threshold comparisons can only determine whether the distance between the vehicle and the obstacle is "far enough," and are generally applicable to warnings in longitudinal rear-end collision scenarios. However, they cannot accurately determine whether the side or rear of the vehicle will scrape against the obstacle, because lateral scraping depends not only on distance but also on whether the sweep paths of various parts of the vehicle overlap with the spatial position of the obstacle during turning. Geometric overlap detection, on the other hand, directly determines whether two spatial regions intersect, accurately reflecting whether actual spatial interference will occur during vehicle movement. It is particularly suitable for assessing non-longitudinal collision risks such as lateral scraping and bottom impacts.

[0103] If the detection result is that there is no overlapping area (i.e., the vehicle motion envelope and the obstacle outline do not overlap on the two-dimensional plane), it indicates that the vehicle will not collide with the obstacle if it follows the currently predicted driving trajectory. The vehicle obstacle avoidance control device can return to step S201 to continue monitoring and continuously obtain new vehicle status information and obstacle information for the next round of prediction.

[0104] If the detection result shows an intersection area, it indicates that if the vehicle travels along the currently predicted trajectory, a certain part of the vehicle body will spatially interfere with the obstacle, which means there is a risk of collision. At this time, step S204 and subsequent steps are executed to quantify and classify the risk.

[0105] This embodiment uses spatial overlap as the basis for collision risk, which can directly reflect the interference state between the vehicle's future occupied area and the obstacle boundary, and provide clear input for the subsequent generation of active intervention control commands based on collision risk intensity. Through geometric overlap detection on a two-dimensional plane, the risk assessment result can correspond to the actual vehicle sweep range, thereby improving the recognition accuracy and response consistency of low-speed obstacle avoidance control.

[0106] S204. Calculate the intrusion depth of the intersection region of the vehicle motion envelope and the obstacle contour in the obstacle information on the two-dimensional plane and / or predict the collision time.

[0107] For example, after determining in step S203 that there is an intersection between the vehicle motion envelope and the obstacle outline, the vehicle obstacle avoidance control device can perform quantitative analysis on the intersection area to assess the severity of the collision risk.

[0108] Intrusion depth refers to the maximum overlap distance between the boundary of the vehicle's motion envelope and the boundary of the obstacle's outline within the intersection area. A greater intrusion depth indicates a more severe spatial overlap between the vehicle and the obstacle; that is, under the current predicted trajectory, the vehicle will enter the space occupied by the obstacle with a greater intrusion, resulting in a higher collision risk. Intrusion depth can be determined by calculating the maximum penetration distance and / or area between the boundary of the vehicle's motion envelope and the boundary of the obstacle's outline within the intersection area.

[0109] The predicted time to collision (TTC) refers to the remaining time from the current moment until the vehicle collides with the obstacle, based on the current vehicle motion state and predicted trajectory. A shorter predicted TTC indicates a higher degree of collision imminence, a shorter time window for system intervention, and a higher collision risk. The predicted TTC can be calculated by dividing the relative distance between the vehicle's motion envelope and the obstacle's outline by the vehicle's current approach speed relative to the obstacle, or it can be directly obtained based on the recursive prediction results of the vehicle's kinematics model; this application does not impose any limitations on this.

[0110] It should be noted that in practical applications, vehicle obstacle avoidance control devices can choose to calculate intrusion depth, predict collision time, or both, depending on specific needs; this application's embodiments do not impose any limitations. For example, when the obstacle is a stationary object (such as a curb, wall, or pillar), intrusion depth can more intuitively reflect the degree of spatial overlap; when the obstacle is a moving object (such as other vehicles or pedestrians), predicted collision time can more accurately reflect the urgency of the risk. Using both simultaneously can yield a more comprehensive risk assessment result.

[0111] S205. Determine the current collision risk intensity and the corresponding initial intervention level based on the intrusion depth and / or predicted collision time.

[0112] For example, after calculating the intrusion depth and / or predicted collision time in step S204, the vehicle obstacle avoidance control device can determine the current collision risk intensity based on these quantitative indicators and determine the initial intervention level corresponding to the risk intensity.

[0113] Specifically, the vehicle obstacle avoidance control device can pre-store the mapping relationship between collision risk intensity and intrusion depth and / or predicted collision time. For example, multiple risk intensity threshold ranges can be set: when the intrusion depth is less than a first preset threshold (or the predicted collision time is greater than a first preset time threshold), it is determined to be low risk intensity; when the intrusion depth is greater than or equal to the first preset threshold and less than a second preset threshold (or the predicted collision time is between the first preset time threshold and the second preset time threshold), it is determined to be medium risk intensity; when the intrusion depth is greater than or equal to the second preset threshold (or the predicted collision time is less than the second preset time threshold), it is determined to be high risk intensity. The above thresholds can be calibrated according to vehicle parameters and actual application scenarios, and this application embodiment does not specifically limit them.

[0114] Meanwhile, the vehicle obstacle avoidance control device pre-stores the correspondence between risk intensity and intervention level. Intervention levels are pre-set in ascending order of intensity, with each level corresponding to one or more intervention measures. For example, a low-risk level might correspond to a lower initial intervention level (e.g., only steering correction), a medium-risk level to an intermediate level (e.g., steering correction combined with unilateral braking), and a high-risk level to a higher level (e.g., torque vectoring control or emergency braking). By associating risk intensity with the initial intervention level, the system can directly attempt interventions at the matching level based on the severity of the current risk, avoiding over-intervention at low risks or under-intervention at high risks.

[0115] S206. Starting with the intervention measures corresponding to the initial intervention level, evaluate the intervention measures corresponding to each intervention level in order of increasing intervention intensity, and determine the first intervention measure that can avoid collision risk as the current active obstacle avoidance measure.

[0116] For example, after determining the initial intervention level, the vehicle obstacle avoidance control device begins to evaluate the intervention measures corresponding to the initial intervention level to determine which level of intervention measures can effectively avoid the collision risk in the current situation.

[0117] It should be noted that the evaluation method used in this step is "computational screening" (or "feedforward evaluation"). This means that before issuing any control commands to the vehicle's actuators, the vehicle obstacle avoidance control device first evaluates the effectiveness of each intervention measure through computational simulation. Specifically, for a given intervention level, the vehicle obstacle avoidance control device assumes the implementation of the corresponding intervention measure. Then, based on the expected vehicle state after implementing the intervention measure (e.g., expected vehicle speed, steering wheel angle, yaw rate change), it uses a vehicle kinematic model considering Ackermann steering characteristics to re-predict the new vehicle motion envelope after implementing the intervention measure. This new motion envelope is then geometrically overlapped with the obstacle contour. If the detection results show that the new motion envelope and the obstacle contour have no intersection, the currently evaluated intervention measure is determined to effectively avoid collision risks. This intervention measure is then identified as the current active obstacle avoidance measure, and the evaluation of higher intervention levels is stopped. If the detection results show that the new vehicle motion envelope and the obstacle outline still have an intersection area, then the current assessment of the intervention measure is deemed insufficient to avoid the collision risk, and the next level of intervention with a higher intensity is assessed. This process is repeated step by step upwards until an intervention level that can avoid the collision risk is found or all preset intervention levels are exhausted.

[0118] If, after going through all intervention levels, no intervention measure can avoid the risk of collision (for example, due to extremely narrow space or the sudden appearance of obstacles, no active intervention can avoid a collision), then the measure with the highest intervention intensity (such as triggering emergency braking to stop) can be selected as the final active obstacle avoidance measure to minimize collision damage as much as possible.

[0119] The aforementioned step-by-step assessment decision-making logic, on the one hand, improves decision-making efficiency and shortens calculation time by starting the assessment from a level that matches the current risk intensity, rather than starting from the lowest level each time. On the other hand, by using "calculation-based screening" rather than "post-execution feedback adjustments," it avoids time delays and increased collision risks caused by insufficient intervention measures during actual execution, while ensuring that the final output is a single optimal intervention instruction, rather than multiple superimposed instructions, thereby avoiding control conflicts between the implementing agencies. Furthermore, adopting a step-by-step upward assessment approach can also avoid excessive intervention while meeting avoidance requirements, and improve obstacle avoidance accuracy and control smoothness in low-speed, narrow scenarios.

[0120] Optionally, in one possible embodiment, starting with the intervention measures corresponding to the initial intervention level, the intervention measures corresponding to each intervention level are evaluated progressively upwards in order of intervention intensity from low to high. The first intervention measure that can avoid the collision risk is determined as the current active obstacle avoidance measure, which may include:

[0121] S21. Obtain the intervention measures corresponding to the initial intervention level; wherein, the different intervention levels include the corresponding intervention measures in order of intervention intensity from low to high.

[0122] S22. Based on a vehicle kinematics model that considers Ackermann steering characteristics, predict the vehicle body motion envelope after the vehicle performs the intervention measures corresponding to the current assessment intervention level.

[0123] S23. If it is determined that there is no intersection between the vehicle motion envelope and the obstacle outline on the two-dimensional plane, then the intervention measures corresponding to the current assessment intervention level shall be determined as the current active obstacle avoidance measures.

[0124] S24. If it is determined that there is still an intersection between the vehicle motion envelope and the obstacle outline on the two-dimensional plane, continue to evaluate the intervention measures corresponding to the next higher intervention level until an intervention level that can avoid the risk of collision is found.

[0125] S25. If the risk of collision cannot be avoided after going through all intervention levels, automatic emergency braking is triggered to bring the vehicle to a stop.

[0126] For example, the vehicle obstacle avoidance control device may pre-store multiple intervention levels, which are arranged in order of increasing intervention intensity. Each intervention level corresponds to one or more intervention measures. The lower the intervention intensity, the less the corresponding intervention measure interferes with the vehicle's motion (e.g., only providing assistance prompts or slight steering corrections); the higher the intervention intensity, the more the corresponding intervention measure interferes with the vehicle's motion (e.g., combined braking or emergency braking).

[0127] Optionally, in one possible embodiment, depending on the vehicle's drive system configuration, different intervention levels include corresponding intervention measures in ascending order of intervention intensity, and may include at least two of the following:

[0128] The first level of intervention corresponds to intervention measures including: controlling the electric power steering system to apply steering correction torque;

[0129] The second level of intervention includes: controlling the electric power steering system to make steering corrections and controlling the braking system to apply braking force to the wheels.

[0130] The third level of intervention includes: controlling the torque difference between the left and right drive wheels to generate yaw moment;

[0131] The fourth level of intervention includes triggering automatic emergency braking to bring the vehicle to a stop.

[0132] For example, in a specific implementation, a command mapping relationship corresponding to the intervention level can be established based on whether the vehicle currently possesses electric power steering, braking control, and independent torque control capabilities for the left and right drive wheels. When the collision risk is low, only a steering correction torque command is output to the electric power steering system, causing a slight deviation in the vehicle's trajectory. When the collision risk increases, a braking control command is superimposed on the steering correction to further shorten the motion path through wheel braking force. When the vehicle has independent drive control capabilities, the torque difference between the left and right drive wheels can also be controlled to generate a yaw moment for more significant attitude correction. When the collision risk continues to increase and a collision cannot be avoided, an automatic emergency braking signal is triggered, and the braking actuator implements braking control to bring the vehicle to a stop.

[0133] The aforementioned intervention level configuration allows different control measures to be adapted to the actual drive system capabilities of the vehicle, with the intervention intensity progressively increasing from low to high. During operation, the control device combines the relative position of the obstacle with the vehicle's motion state to select the appropriate level of active intervention measures and output them to the steering, braking, or drive actuators. This enables the coordinated completion of trajectory correction, attitude adjustment, and emergency deceleration in narrow traffic environments. Using this method, the vehicle can select matching avoidance control based on the collision risk intensity, improving the feasibility and continuity of intervention levels, and allowing vehicles with different drive configurations to adopt obstacle avoidance control strategies adapted to their hardware capabilities.

[0134] It should be noted that the specific classification of the above-mentioned intervention levels and the corresponding intervention measures can be set according to the actual configuration and application requirements of the vehicle, and this application embodiment does not impose specific limitations on this.

[0135] After obtaining the intervention measures corresponding to the current assessment intervention level, the vehicle obstacle avoidance control device further predicts, through simulation calculation, how the vehicle's motion envelope will change after the intervention measures corresponding to the current assessment intervention level are implemented.

[0136] Specifically, for a given intervention level, the vehicle obstacle avoidance control device first assumes the implementation of the corresponding intervention measure. Based on the type and parameters of the intervention measure (e.g., direction and magnitude of steering correction torque, object and magnitude of braking force, magnitude and direction of torque difference), it calculates the expected motion state of the vehicle after implementing the intervention measure (e.g., expected vehicle speed, steering wheel angle, yaw rate change, etc.). Then, based on these expected vehicle motion states, it again uses a vehicle kinematics model considering Ackermann steering characteristics to predict the vehicle's trajectory in the future time period. Following the same method as in step S202, it determines the sweep paths of various parts of the vehicle body after implementing the intervention measure, thus obtaining a new vehicle motion envelope. It should be noted that the prediction in this step is based on simulation calculations of the vehicle kinematics model, not on actually sending control commands to the vehicle's actuators. In other words, before actually controlling the vehicle, the vehicle obstacle avoidance control device "pre-simulates" the effects of each intervention measure through calculation and simulation, thereby selecting the optimal intervention scheme. This "calculate first, then execute" decision-making approach, compared to the "execute first, then provide feedback" real-time adjustment approach, can avoid time delays caused by insufficient initial intervention, while also avoiding the impact on the driving experience caused by excessive intervention.

[0137] After calculating and predicting the new vehicle motion envelope after implementing the current intervention level, the vehicle obstacle avoidance control device performs a two-dimensional geometric overlap detection between the new vehicle motion envelope and the obstacle contour. If the detection result shows that there is no intersection between the new vehicle motion envelope and the obstacle contour, it indicates that the vehicle can effectively avoid collisions with obstacles after implementing the currently evaluated intervention measures. At this time, the vehicle obstacle avoidance control device determines the intervention measure corresponding to the current evaluated intervention level as the current active obstacle avoidance measure and stops evaluating higher intervention intensity levels. This active obstacle avoidance measure will serve as the basis for generating active intervention control commands in step S207. If the geometric overlap detection result shows that there is still an intersection between the new vehicle motion envelope and the obstacle contour in the two-dimensional plane after implementing the currently evaluated intervention measures, it indicates that the currently evaluated intervention measure is insufficient to completely avoid collision risks. In this case, the vehicle obstacle avoidance control device shifts to evaluating the intervention measures corresponding to the next higher intervention level, and repeats the evaluation process of "predicting the new vehicle motion envelope after implementing the intervention level → determining whether there is an intersection between the new vehicle motion envelope and the obstacle contour" for the next higher intervention level. This process is repeated iteratively, evaluating higher levels of intervention intensity step by step, until an intervention level that can avoid the risk of collision is found.

[0138] This application starts the assessment from the initial intervention level, rather than starting from the lowest level each time, which can reduce unnecessary calculations and improve decision-making efficiency. At the same time, by assessing the intervention level in order of increasing intensity, it can prioritize measures with milder intervention levels, and minimize the interference with the driver's normal driving operation while meeting the needs of safe obstacle avoidance, thereby improving driving comfort.

[0139] In certain extreme situations, such as when space is extremely confined and no active intervention can safely avoid an obstacle, or when an obstacle appears suddenly and reaction time is extremely insufficient, it is possible that even after traversing all preset intervention levels, no intervention can completely eliminate the overlap between the vehicle's motion envelope and the obstacle's outline. In such cases, the vehicle's obstacle avoidance control system can trigger Automatic Emergency Braking (AEB), bringing the vehicle to a stop with maximum braking force. Automatic Emergency Braking, as a last line of defense, aims to reduce the collision speed between the vehicle and the obstacle as much as possible when a collision cannot be completely avoided, thereby mitigating the damage and injury caused by the collision.

[0140] The step-by-step evaluation process described in S21 to S25 above, through a closed-loop evaluation mechanism of "simulation execution → re-prediction → re-detection → judgment / iteration", realizes the pre-calculation and screening of the effectiveness of intervention measures at each level. It can ensure that the final output active obstacle avoidance measures are the optimal (i.e., the least intense and effective) solutions that can avoid collision risks under the current conditions, taking into account both safety and comfort. At the same time, by setting a fallback logic of "braking if no avoidance can be achieved after traversing all levels", it ensures that the system can still take effective measures in the most extreme cases, minimizing collision losses to the greatest extent.

[0141] Optionally, in one possible embodiment, the process of evaluating the intervention measures corresponding to each intervention level further includes:

[0142] S31. Based on the spatial location and intrusion depth of the intersection area between the vehicle motion envelope and the obstacle contour, determine the expected change in yaw rate required to eliminate the intersection area.

[0143] S32. Based on the current assessment of the intervention level, determine the target implementing agency corresponding to that intervention level;

[0144] S33. Determine the control quantity that the target actuator should execute based on the expected change in yaw rate.

[0145] Understandably, different intervention levels correspond to different target actuators (e.g., steering correction corresponds to electric power steering, combined braking corresponds to braking systems, and torque vector control corresponds to drive motors), and the control physical quantities of different actuators are different (e.g., torque, braking force, torque difference). However, regardless of the type of target actuator, the determination of its control quantity follows the same underlying technical concept: based on the spatial overlap relationship between the vehicle motion envelope and the obstacle contour, the desired yaw rate change required to eliminate the spatial overlap is calculated, and then this desired yaw rate change is mapped to a control quantity that matches the target actuator of the current intervention level.

[0146] Specifically, the vehicle obstacle avoidance control device can calculate the desired change in yaw rate required to eliminate the intersection area (i.e., avoid collision risk) based on the spatial location and intrusion depth of the intersection area between the vehicle's motion envelope and the obstacle's outline. The spatial location of the intersection area refers to its position relative to the vehicle in a two-dimensional plane; for example, the intersection area could be located to the left front, right rear, or the middle of the left side of the vehicle. This spatial location determines the direction the vehicle needs to change its trajectory to move away from the obstacle. For instance, if the intersection area is on the left side of the vehicle, a rightward yaw motion is required to avoid the obstacle; if the intersection area is on the right side, a leftward yaw motion is required. The intrusion depth refers to the maximum overlap distance between the boundary of the vehicle's motion envelope and the boundary of the obstacle's outline within the intersection area. The magnitude of the intrusion depth reflects the severity of the spatial overlap; a greater intrusion depth indicates that the vehicle requires a larger trajectory adjustment to eliminate the spatial overlap.

[0147] When calculating the desired change in yaw rate required to eliminate the intersection region, the vehicle obstacle avoidance control device first determines the direction of the desired change in yaw rate (left or right turn) based on the spatial location of the intersection region, causing the vehicle to move away from the obstacle. Then, based on the depth of intrusion, combined with the current vehicle speed and vehicle geometry parameters, the magnitude of the desired change in yaw rate is calculated; the greater the intrusion depth, the greater the required change in yaw rate. By changing the vehicle's yaw rate, the vehicle can generate yaw motion during driving, thereby shifting its motion envelope and ultimately eliminating the spatial overlap with the obstacle's outline.

[0148] It should be noted that the expected change in yaw rate is an intermediate target value, which represents the adjustment range that the vehicle needs to achieve in terms of yaw motion to avoid a collision. This intermediate target value does not depend on the specific type of actuator, but is determined by pure geometry and kinematics, and has universality. Regardless of whether it is subsequently achieved through steering correction, unilateral braking, or torque vector control, its physical target (i.e., how much change in yaw rate is required) is uniform.

[0149] Furthermore, the vehicle obstacle avoidance control device can determine the target actuator required to execute the intervention measures corresponding to the current assessment intervention level. As described above, different intervention levels involve different actuators. If the current assessment intervention level is the first intervention level (steering correction intervention), the target actuator is the electric power steering system (EPS), whose task is to convert the desired change in yaw rate into steering correction torque. If the current assessment intervention level is the second intervention level (steering correction combined with single-sided braking intervention), the target actuator includes the electric power steering system and the braking system (ESP), which work together to output steering correction torque and single-sided wheel braking force, respectively. If the current assessment intervention level is the third intervention level (torque vector control intervention), the target actuator is the drive motor (or active differential), whose task is to convert the desired change in yaw rate into the torque difference between the left and right drive wheels. If the current assessment intervention level is the fourth intervention level (emergency braking intervention), the target actuator is the braking system, whose task is to stop the vehicle with maximum braking force.

[0150] By identifying the target actuator corresponding to the current intervention level, the vehicle obstacle avoidance control device can determine which actuator(s) to send the control input to in subsequent steps, and the types of control input that actuator can accept.

[0151] After determining the desired yaw rate change and the target actuator, the vehicle obstacle avoidance control device can further convert the desired yaw rate change into a specific control quantity that matches the target actuator.

[0152] It should be noted that different types of actuators achieve changes in yaw rate in different ways, and therefore the forms of control quantities are also different, but in essence, they all use the action of the actuator to make the vehicle produce the desired yaw motion.

[0153] Specifically, when the target actuator is an electric power steering system, the vehicle obstacle avoidance control device can calculate the required steering correction torque based on the desired change in yaw rate, combined with the current vehicle speed and the steering system's gear ratio. For example, the desired change in yaw rate is first converted into the desired change in front wheel angle, and then the steering correction torque that the electric power steering system needs to apply is calculated using the torque-angle mapping relationship of the steering system. The direction of the steering correction torque should be consistent with the direction of the desired change in yaw rate.

[0154] When the target actuator is a braking system (performing single-sided braking), the vehicle obstacle avoidance control device can first determine the single-sided wheel that needs to be braked based on the direction of the expected change in yaw rate (the left wheel needs to be braked when turning left, and the right wheel needs to be braked when turning right). Then, based on the magnitude of the expected change in yaw rate, the current vehicle speed, and the vehicle track, it calculates the amount of braking force that needs to be applied to that single-sided wheel.

[0155] When the target actuator is a drive motor (performing torque vector control), the vehicle obstacle avoidance control device can calculate the target torque difference required by the left and right drive wheels based on the expected change in yaw rate, wheel rolling radius, and vehicle track width. Then, while keeping the total change in drive torque less than a preset threshold, the device distributes the torque of the left and right drive wheels according to the target torque difference (for example, if the vehicle needs to yaw to the left, the torque of the right drive wheel is increased and the torque of the left drive wheel is decreased; and vice versa).

[0156] In this way, the vehicle obstacle avoidance control device can adapt the unified spatial overlap elimination requirement (expected yaw rate change) into the control quantity that different actuators can execute, realizing a complete mapping from "geometric space determination" to "actuator action command".

[0157] The aforementioned schemes S31 to S33 establish a unified mapping framework of "spatial overlap → desired yaw rate change → actuator control quantity," making the system's algorithm architecture simpler and clearer. This also facilitates flexible expansion or tailoring of intervention levels for different vehicle configurations (such as whether they possess torque vectoring capabilities) without requiring redesign of the control quantity calculation logic. Furthermore, by adopting this specific implementation method, control outputs matching the spatial location and intrusion depth of the intersection area can be generated under different intervention levels. This makes the vehicle's attitude correction in low-speed, confined environments more targeted, and the correspondence between intervention intensity and collision risk intensity clearer, thereby improving the adaptability and execution accuracy of active obstacle avoidance control.

[0158] S207. Generate active intervention control instructions based on the current active obstacle avoidance measures.

[0159] For example, after determining the current active obstacle avoidance measure, the vehicle obstacle avoidance control device can generate a corresponding active intervention control command based on the measure and output the command to the vehicle's actuator.

[0160] Specifically, the vehicle's actuators may include, but are not limited to, an electric power steering system, an electronic stability control system with active pressure-building capabilities, an engine management system or motor controller, and a braking system. Depending on the active obstacle avoidance measures, the generated active intervention control commands may include at least one of the following types:

[0161] Steering Correction Command: This command controls the electric power steering system to apply a steering correction torque of a specific direction and magnitude, assisting the driver in making minor adjustments to the steering wheel to guide the vehicle away from obstacles. The direction of the steering correction torque depends on the spatial overlap between the vehicle's motion envelope and the obstacle's outline: if the overlap area is on the left side of the vehicle, the steering correction torque should be directed to the right; if the overlap area is on the right side of the vehicle, the steering correction torque should be directed to the left. The magnitude of the steering correction torque depends on the depth of intrusion and / or the predicted collision time; the greater the intrusion depth or the shorter the predicted collision time, the greater the required steering correction torque.

[0162] Joint Correction Command: While controlling the electric power steering system to correct steering, the braking system applies braking force to one wheel. By applying braking force to one wheel, the vehicle can yaw, further altering its trajectory based on the steering correction and achieving greater obstacle avoidance. The wheel requiring braking depends on the direction of the required yaw moment: if the vehicle needs to yaw to the left, braking force is applied to the left wheel; if the vehicle needs to yaw to the right, braking force is applied to the right wheel. The magnitude of the braking force depends on the magnitude of the required yaw moment, which is determined by the depth of intrusion and / or the predicted collision time.

[0163] Torque vectoring control commands: For vehicles with independent left and right wheel drive capabilities (such as dual-motor electric vehicles or vehicles equipped with active differentials), yaw moment can be generated by controlling the torque difference between the left and right drive wheels, thereby changing the vehicle's trajectory. The difference in principle between torque vectoring control and single-sided braking in the aforementioned combined correction commands is that single-sided braking generates yaw moment by consuming energy (applying braking force), while torque vectoring control generates yaw moment by redistributing drive torque between the left and right drive wheels. This achieves precise trajectory correction while maintaining a relatively constant total drive torque, resulting in faster response and higher energy efficiency. The torque distribution method between the left and right drive wheels depends on the direction and magnitude of the required yaw moment: if the vehicle needs to yaw to the left, the torque of the right drive wheel is increased and the torque of the left drive wheel is decreased; conversely, the opposite is also true. The magnitude of the torque difference is determined by the magnitude of the required yaw moment.

[0164] Emergency Braking Command: Activates Automatic Emergency Braking (AEB) to bring the vehicle to a stop with maximum braking force. This command is typically used as a last resort, triggered when all other active intervention methods fail to safely avoid a collision, aiming to minimize collision damage.

[0165] The vehicle obstacle avoidance control device can output the above-mentioned active intervention control commands to the corresponding actuators, and the actuators can work together to complete the corresponding steering correction, braking or torque distribution actions.

[0166] The vehicle obstacle avoidance control method provided in this application acquires vehicle state information and obstacle information, predicts the vehicle motion envelope based on a vehicle kinematic model considering Ackermann steering characteristics, then determines the existence of collision risk through geometric overlap detection, and further calculates the intrusion depth and / or predicts the collision time to quantify the risk intensity when a risk exists. This determines the initial intervention level and evaluates each intervention measure step by step, ultimately identifying the first intervention measure that can avoid the risk as the active obstacle avoidance measure and generating control commands. This method achieves active prediction, accurate quantification, and graded intervention of lateral and bottom impact risks, improving the safety and reliability of vehicles moving at low speeds in confined spaces. Simultaneously, through step-by-step evaluation and adaptive selection of intervention measures, it balances driving comfort and execution efficiency.

[0167] Optionally, based on any of the above embodiments, in one possible embodiment, the vehicle obstacle avoidance control method provided in this application may further include:

[0168] S41. Obtain the vehicle's location information;

[0169] S42. Compare the location information with the stored collision risk geographic area data. If it is determined that the vehicle is in or approaching a collision risk geographic area, generate a warning message and output the warning message to the driver.

[0170] It should be noted that the positioning and early warning scheme provided in this optional embodiment can be executed in parallel or independently with the "perception-prediction-intervention" main process in the aforementioned embodiments, and there is no strict temporal dependency between the two.

[0171] For example, vehicle positioning information can be obtained through an onboard high-precision satellite positioning system. This system can include one or more combinations of systems such as the Global Positioning System (GPS) and the BeiDou Navigation Satellite System (BDS), and can also be combined with an inertial navigation system for assisted positioning to improve accuracy and reliability. Furthermore, in scenarios with weak satellite signals (such as underground parking garages or tunnels), dead reckoning can be performed using information from vehicle wheel speed sensors and steering wheel angle sensors to continuously obtain vehicle positioning information.

[0172] Optionally, the positioning information includes at least the latitude and longitude coordinates of the vehicle's current location, and may also include information such as the vehicle's direction of travel and speed. The positioning accuracy can be set according to the needs of the actual application scenario. For example, when it is necessary to accurately determine whether a vehicle has entered a certain geographical area, differential positioning technology can be used to improve the positioning accuracy to a preset accuracy level (e.g., centimeter level). This application does not limit the specific positioning method and accuracy.

[0173] Furthermore, the vehicle obstacle avoidance control device compares the acquired vehicle location information with the stored collision risk geographic area data, and determines whether to issue a warning to the driver based on the comparison result. The stored collision risk geographic area data (i.e., the "collision black spot" database) is a geographic information system based on historical data. This database stores a large amount of geographic location information where vehicles have been involved in collisions or where there is a high risk, such as narrow right-angle bends, mechanical parking space entrances where side skirts are easily scratched, road sections with low, dense curbs, and underground parking garage entrances with obstructed visibility. Optionally, this collision risk geographic area data is uploaded by multiple vehicles during driving via IoT communication and, after verification, stored on a cloud server or in a local database, forming a dynamically updated database that can be accessed in real time via the vehicle network.

[0174] In its implementation, when the collision risk geographic area data is stored on a cloud server, the vehicle obstacle avoidance control device can establish a communication connection with the cloud server through an onboard communication module (such as a 4G / 5G / 6G cellular communication module, a V2X communication module, etc.) to obtain the collision risk geographic area data. Then, it compares the real-time latitude and longitude coordinates of the vehicle's current location with the boundary coordinates of the collision risk geographic area. If the comparison determines that the vehicle is currently within the collision risk geographic area, or is approaching it (i.e., the distance between the vehicle's current location and the boundary of the collision risk geographic area is less than or equal to a preset distance threshold, and the vehicle's driving direction is pointing towards that area), it is determined that the vehicle may enter a high-risk area. At this time, the vehicle obstacle avoidance control device generates a warning message and outputs this warning message to the driver through the onboard human-machine interface.

[0175] Optionally, the output of warning information may include one or more combinations of the following: projecting visual cues in front of the driver's field of vision via a head-up display system; displaying warning icons or text information on the instrument panel; displaying the specific location and risk type of the collision risk area on the central control screen; and issuing voice prompts through the vehicle's audio system to inform the driver of the presence of a collision black spot area ahead and to drive with caution. By combining multiple output methods, multiple reminders can be generated under different driving conditions (such as in front of the driver's line of sight, within peripheral vision, and through the auditory channel), ensuring that the warning information can be effectively perceived by the driver.

[0176] For example, Figure 3 This is a schematic diagram illustrating a location-based early warning process provided in an embodiment of this application. Figure 3 As shown, a high-precision positioning system can continuously acquire positioning information, and then compare the positioning information with data in the IoT collision black spot database. If the location is in the area of ​​the collision black spot, a collision risk warning sound will be output.

[0177] It should be noted that the positioning warning scheme provided in this optional embodiment has different technical principles and complementary technical effects compared with the perception-prediction-intervention schemes based on local sensors (cameras, LiDAR, millimeter-wave radar, etc.) in the aforementioned embodiments: the aforementioned schemes based on local sensors rely on the vehicle's own onboard sensing equipment, and their sensing range is limited by the sensor's detection distance (usually tens to hundreds of meters), belonging to line-of-sight perception, capable of accurate modeling and real-time response of the immediate environment. The cloud-based black spot warning scheme adopted in this optional embodiment relies on high-precision positioning and a cloud-based collision risk geographic area database, not on the detection capability of local sensors, and can predict historical risk areas within a range of several kilometers or even further in front of the vehicle, belonging to beyond-line-of-sight perception, and can issue a warning to the driver several seconds to several minutes before the vehicle actually enters the danger zone. The above two schemes can work together: the cloud-based warning scheme provides the driver with more attention time, enabling the driver to slow down or increase vigilance before entering a high-risk area; while the real-time perception and active intervention scheme based on local sensors provides accurate trajectory prediction and graded intervention when the vehicle actually approaches an obstacle. By combining "beyond visual range early warning" with "active intervention within visual range", the risk of collisions can be reduced to a greater extent.

[0178] Furthermore, in an optional implementation, after outputting cloud-based early warning information, the vehicle obstacle avoidance control device can also increase the data acquisition frequency and / or prediction frequency of the environmental perception module and trajectory prediction module. Specifically, when the vehicle is in or approaching a collision-risk geographical area, the system can proactively increase the sampling rate of local sensors (such as cameras, LiDAR, and millimeter-wave radar) while shortening the trajectory prediction cycle. This allows for faster and more frequent perception of environmental changes and updates to prediction results after entering a high-risk area, further improving the response speed and accuracy of proactive intervention. This "early warning-triggered enhanced monitoring" mechanism further demonstrates the synergistic effect between cloud-based early warning and local proactive intervention.

[0179] The above solution compares vehicle location information with stored collision risk geographic area data, which can issue early warnings to the driver when the vehicle enters or approaches a known high-risk area, giving the driver more time to adjust driving operations. It makes up for the limitations of local sensors in terms of line of sight and is especially suitable for risk warning in unfamiliar and complex environments.

[0180] For example, Figure 4 This is a schematic diagram of the architecture of a vehicle obstacle avoidance control system provided in an embodiment of this application. Figure 4 As shown, the vehicle obstacle avoidance control system provided in this application embodiment may include three levels: an information perception layer, a central decision-making layer, and an execution control layer.

[0181] The information perception layer includes an environmental perception module, a vehicle status perception module, and a positioning and historical data module. The environmental perception module integrates perception data from onboard sensors such as visual cameras, LiDAR, and millimeter-wave radar to construct a real-time model of the environment surrounding the vehicle, identifying the position, size, and outline of obstacles. The vehicle status perception module acquires real-time information such as vehicle speed, steering wheel angle, yaw rate, and vehicle geometric parameters (at least wheelbase and track width); for vehicles equipped with rear-wheel steering or active suspension, it can also acquire rear-wheel steering angle and suspension height. The positioning and historical data module obtains the vehicle's location through high-precision satellite positioning and communicates with a cloud server via the vehicle network to obtain geographic area data of "collision black spots" stored in the cloud.

[0182] The central decision-making layer comprises a trajectory prediction unit, a collision risk assessment unit, and a hierarchical decision-making unit. The trajectory prediction unit, based on vehicle state information and utilizing a vehicle kinematic model considering Ackermann steering characteristics, predicts the vehicle's motion envelope over a future time period—that is, the entire spatial area traversed by the vehicle along the predicted trajectory. This prediction fully considers the inner wheel difference effect caused by the non-coincidence of the front and rear wheel trajectories in Ackermann steering geometry, accurately reflecting the sweep paths of various parts of the vehicle body during steering. The collision risk assessment unit performs geometric overlap detection on a two-dimensional plane between the vehicle's motion envelope and the obstacle contour: if an intersection area exists, a collision risk is determined, and the intrusion depth of the intersection area and / or the predicted collision time are further calculated to quantify the risk intensity. The hierarchical decision-making unit determines the initial intervention level based on the risk intensity and evaluates each intervention measure in ascending order of intervention intensity to determine whether it can avoid the risk. The first intervention measure that can avoid the risk is identified as the current active obstacle avoidance measure, and corresponding control commands are generated.

[0183] The execution control layer includes the steering system (such as EPS), braking system (such as ESP), and torque vectoring control system. This layer receives control commands from the central decision-making layer and controls the corresponding actuators to perform steering corrections, braking, or torque distribution actions. The torque vectoring control system is only available when the vehicle has independent drive capability for the left and right wheels.

[0184] Take, for example, a smart electric vehicle equipped with EPS, ESP, and a torque vectoring differential making a narrow right-angle turn. When the vehicle enters the collision black spot area, the positioning and historical data module detects that the vehicle is approaching the collision black spot area marked in the cloud. The hierarchical decision-making unit generates a warning message and outputs an audible warning to the driver through the vehicle's audio system.

[0185] Subsequently, the environmental perception module detects the presence of a curb on the right side using lidar and extracts its contour information; the vehicle state perception module acquires the current vehicle speed, steering wheel angle, and vehicle geometric parameters. Based on this information, the trajectory prediction unit predicts the vehicle's motion envelope using a vehicle kinematics model that considers Ackermann steering characteristics. The collision risk assessment unit performs geometric overlap detection between the vehicle's motion envelope and the curb contour, calculating that there is an intersection between the right rear wheel sweep path and the curb contour, predicting a collision time of 1.5 seconds.

[0186] Furthermore, the tiered decision-making units initiate tiered assessments:

[0187] First, assess the first intervention level (Level 2a), assuming a slight steering correction torque is applied to the left via EPS, and re-predict the vehicle motion envelope. If the calculation shows that the new vehicle motion envelope does not intersect with the curb contour, an EPS correction command is generated and executed by the steering system; obstacle avoidance is completed without driver resistance.

[0188] If the correction space at level 2a is insufficient and the re-prediction results show that there is still an overlap area, then the second intervention level (level 2b) is evaluated. This assumes that while the EPS corrects to the left, the ESP applies a slight braking force to the left front wheel, causing the car to drift to the left earlier. If the re-prediction shows that this can be avoided, then EPS steering correction commands and ESP single-sided braking commands are generated and executed collaboratively by the steering and braking systems.

[0189] If, due to the extremely confined space, a collision is still possible even with a Level 2b combined correction, the assessment will continue to the Level 3 intervention (Level 2c). Since this vehicle is equipped with a torque vectoring differential, the assessment assumes the torque vectoring function is activated—increasing the torque on the right drive wheel and decreasing the torque on the left drive wheel, generating a significant yaw moment to the left. If the re-prediction indicates that the vehicle can escape the danger zone, a torque distribution command is generated and executed by the vector torque control system.

[0190] If all intervention levels indicate that a collision is unavoidable (e.g., an obstacle suddenly appears), then the fourth intervention level (level 2d) is executed, triggering automatic emergency braking to bring the vehicle to a stop.

[0191] The above process embodies the closed-loop mechanism of "prediction-evaluation-tiered intervention" in this application: predicting the vehicle motion envelope by using a kinematic model that considers Ackermann steering characteristics, judging collision risk by using geometric overlap detection, and selecting the optimal intervention measures through tiered evaluation, thereby achieving proactive prediction and precise intervention of lateral collision risk in low-speed steering scenarios.

[0192] Taking a traditional gasoline-powered vehicle equipped only with an electric power steering (EPS) system and an electronic stability program (ESP) system as an example, this vehicle does not have independent drive capability for the left and right wheels, and therefore cannot perform torque vectoring control (Level 2c) intervention. The vehicle is reversing into a mechanical parking space. Mechanical parking spaces are typically narrow, with obstacles such as pillars and limit bars on both sides, making them high-risk scenarios for collisions.

[0193] When a conventional gasoline-powered vehicle approaches the mechanical parking space, the positioning and historical data module acquires the vehicle's location information and compares it with collision risk geographic area data stored in the cloud. When the comparison confirms that the mechanical parking space is marked as a "scratch-prone side pillar" location, the system determines that the vehicle is approaching a collision risk geographic area. The hierarchical decision-making unit generates a Level 1 warning message and outputs an audible and visual warning to the driver through the in-vehicle human-machine interface (such as the central control screen and audio system), reminding the driver that there is a scratch-prone side pillar in the parking space and suggesting caution when operating the vehicle.

[0194] Subsequently, during the driver's reverse parking maneuver, the environmental perception module uses a visual camera to perceive the vehicle's surroundings in real time. When the camera detects a pillar on the left side of the vehicle, the environmental perception module extracts the pillar's position, size, and outline information to construct an environmental model. The vehicle status perception module simultaneously acquires the current vehicle speed (for low-speed reverse parking), steering wheel angle, yaw rate, and vehicle geometric parameters.

[0195] Furthermore, based on the above information, the trajectory prediction unit in the central decision-making layer predicts the vehicle's motion envelope using a vehicle kinematics model that considers Ackermann steering characteristics. Because this model fully considers the inner wheel difference characteristics in Ackermann steering geometry, the prediction results accurately reflect the sweep paths of various parts of the vehicle body during reverse steering. The collision risk assessment unit performs geometric overlap detection on a two-dimensional plane between the predicted vehicle motion envelope and the contour of the left pillar. Calculations show that if the vehicle continues along the current reverse trajectory, there is an intersection area between the vehicle motion envelope and the contour of the left pillar, indicating a risk of collision.

[0196] Subsequently, the tiered decision-making unit initiates tiered interventions based on the risk assessment results. First, the assessment begins with the lowest intervention intensity level (Level 2a), assuming that a steering correction torque is applied via EPS to fine-tune the reversing direction, and the vehicle's motion envelope is re-predicted. If the calculation shows that this correction eliminates the overlap between the vehicle's motion envelope and the pillar profile, a steering correction command is generated and executed by the EPS to assist the driver in fine-tuning the direction to complete the parking maneuver without further intervention.

[0197] If the correction space at level 2a is insufficient, and the re-prediction results show that there is still an intersection between the vehicle's motion envelope and the pillar, the hierarchical decision unit continues to evaluate the second intervention level (level 2b). Assuming that while EPS steering correction is being performed, ESP applies braking force to one wheel, causing the vehicle to yaw during reversing, thereby adjusting the vehicle's attitude. The vehicle's motion envelope is re-predicted; if the calculation shows that the pillar can be avoided, EPS steering correction commands and ESP one-sided braking commands are generated and executed collaboratively to complete obstacle avoidance.

[0198] If, after the 2b-level joint correction, the re-prediction results still show that it is impossible to completely avoid the pillar—that is, the space is so narrow that any combination of steering and braking is difficult to pass safely—since this vehicle does not have torque vectoring control capabilities (it is not equipped with dual motors or an active differential), the system skips the 2c-level (torque vectoring intervention) and directly evaluates the fourth intervention level (2d-level), triggering automatic emergency braking to stop the vehicle and avoid collision with the pillar.

[0199] This embodiment demonstrates the adaptive capability of the proposed solution to different vehicle hardware configurations: for low-end vehicles without torque vectoring, the system automatically skips level 2c intervention and directly executes level 2d braking after level 2b combined correction proves ineffective. This architecture design of "dynamically adjusting the hierarchical sequence according to vehicle hardware configuration" allows the same control algorithm framework to adapt to different models from low-end to high-end, ensuring both the basic safety protection capabilities of low-end vehicles—timely braking when steering correction and combined braking cannot avoid risks—and fully leveraging the handling potential of high-end vehicles—still having torque vectoring control as a more precise intervention method after combined braking.

[0200] Figure 5 This is a schematic diagram of a vehicle obstacle avoidance control device provided in an embodiment of this application. Figure 5 As shown, the vehicle obstacle avoidance control device 50 provided in this application embodiment includes: a first processing unit 501, a second processing unit 502, and a third processing unit 503.

[0201] The first processing unit 501 is used to acquire vehicle status information and obstacle information in the direction of vehicle travel;

[0202] The second processing unit 502 is used to predict the vehicle's motion envelope in a future time period based on the vehicle kinematics model that takes into account Ackermann steering characteristics and the vehicle state information; wherein the vehicle motion envelope is used to indicate the spatial area traversed by the vehicle along the predicted trajectory.

[0203] The third processing unit 503 is used to generate active intervention control commands to avoid collision risks based on the intensity of the collision risk if it is determined that the vehicle has a collision risk based on obstacle information and vehicle motion envelope.

[0204] In one possible embodiment, the second processing unit 502 is specifically used for:

[0205] Based on vehicle state information, a vehicle kinematics model that considers Ackermann steering characteristics is used to predict the vehicle's trajectory in the future time period.

[0206] Based on the vehicle's geometric parameters and Ackermann steering characteristics, calculate the inner wheel difference offset between the front wheel track and the rear wheel track when the vehicle is traveling along the driving trajectory; where the geometric parameters include at least the wheelbase and track width.

[0207] Based on the driving trajectory and the inner wheel difference offset, the sweep paths of each part of the vehicle body during the driving process are determined, and the sweep paths of each part together constitute the motion envelope of the vehicle body.

[0208] In one possible embodiment, the third processing unit 503 is specifically used for:

[0209] Perform geometric overlap detection on a two-dimensional plane between the spatial region covered by the vehicle motion envelope and the obstacle contour in the obstacle information;

[0210] If an intersection area is detected between the vehicle's motion envelope and the obstacle's outline on a two-dimensional plane, a collision risk is determined to exist.

[0211] In one possible embodiment, the third processing unit 503 is specifically used for:

[0212] Calculate the intrusion depth and / or predict the collision time of the intersection region of the vehicle motion envelope and the obstacle contour in the obstacle information on a two-dimensional plane;

[0213] Determine the current collision risk intensity and the corresponding initial intervention level based on the intrusion depth and / or predicted collision time;

[0214] Starting with the intervention measures corresponding to the initial intervention level, the intervention measures corresponding to each intervention level are evaluated step by step in order of increasing intervention intensity, and the first intervention measure that can avoid the collision risk is determined as the current active obstacle avoidance measure.

[0215] Based on the current active obstacle avoidance measures, generate active intervention and control instructions.

[0216] In one possible embodiment, the third processing unit 503 is specifically used for:

[0217] Obtain the intervention measures corresponding to the initial intervention level; where different intervention levels include the corresponding intervention measures in order of intervention intensity from low to high.

[0218] Based on a vehicle kinematics model that takes into account Ackermann steering characteristics, the vehicle body motion envelope is predicted after the vehicle performs the intervention corresponding to the currently assessed intervention level.

[0219] If it is determined that the vehicle motion envelope and the obstacle outline do not intersect in a two-dimensional plane, then the intervention measure corresponding to the current assessment intervention level will be determined as the current active obstacle avoidance measure.

[0220] If it is determined that there is still an intersection between the vehicle motion envelope and the obstacle outline on the two-dimensional plane, the next higher level of intervention will be evaluated until an intervention level that can avoid the risk of collision is found.

[0221] If the risk of collision cannot be avoided after going through all intervention levels, automatic emergency braking will be triggered to bring the vehicle to a stop.

[0222] In one possible embodiment, depending on the vehicle's drive system configuration, different intervention levels include corresponding intervention measures in ascending order of intervention intensity, including at least two of the following:

[0223] The first level of intervention corresponds to intervention measures including: controlling the electric power steering system to apply steering correction torque;

[0224] The second level of intervention includes: controlling the electric power steering system to make steering corrections and controlling the braking system to apply braking force to the wheels.

[0225] The third level of intervention includes: controlling the torque difference between the left and right drive wheels to generate yaw moment;

[0226] The fourth level of intervention includes triggering automatic emergency braking to bring the vehicle to a stop.

[0227] In one possible embodiment, during the evaluation of intervention measures corresponding to each intervention level, the third processing unit 503 is further specifically used for:

[0228] Based on the spatial location and intrusion depth of the intersection area between the vehicle motion envelope and the obstacle contour, determine the expected change in yaw rate required to eliminate the intersection area;

[0229] Based on the current assessment of the intervention level, determine the target implementing agency corresponding to that intervention level;

[0230] Based on the expected change in yaw rate, determine the control quantity that the target actuator should execute.

[0231] In one possible embodiment, the third processing unit 503 is further specifically used for:

[0232] Obtain the vehicle's location information;

[0233] The system compares the location information with stored collision risk geographic area data. If it determines that the vehicle is in or approaching a collision risk geographic area, it generates a warning message and outputs the warning message to the driver.

[0234] The vehicle obstacle avoidance control device provided in this embodiment is used to execute the technical solution in any of the aforementioned method embodiments. Its implementation principle and technical effect are similar, and will not be described again here.

[0235] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 6 As shown, the electronic device 60 provided in this embodiment includes at least one processor 601 and a memory 602. Optionally, the device 60 further includes a communication component 603. The processor 601, memory 602, and communication component 603 are connected via a bus 604.

[0236] In a specific implementation, at least one processor 601 executes computer execution instructions stored in memory 602, causing at least one processor 601 to perform the above-described method.

[0237] The specific implementation process of processor 601 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0238] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0239] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0240] 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 as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0241] This application also provides a vehicle that performs the technical solutions in any of the foregoing method embodiments.

[0242] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0243] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0244] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0245] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0246] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0247] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0248] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0249] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0250] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0251] The above embodiments are merely preferred embodiments provided to fully illustrate this application, and the scope of protection of this application is not limited thereto. Equivalent substitutions or modifications made by those skilled in the art based on this application are all within the scope of protection of this application.

Claims

1. A vehicle obstacle avoidance control method, characterized in that, The method includes: Obtain vehicle status information and obstacle information in the vehicle's direction of travel; Based on a vehicle kinematics model that takes into account Ackermann steering characteristics, the vehicle body motion envelope is predicted in a future time period according to the vehicle state information; wherein, the vehicle body motion envelope is used to indicate the spatial region traversed by the vehicle along the predicted trajectory. If a collision risk is determined based on the obstacle information and the vehicle motion envelope, an active intervention control command is generated to avoid the collision risk based on the intensity of the collision risk.

2. The method according to claim 1, characterized in that, The vehicle kinematics model, which considers Ackermann steering characteristics, predicts the vehicle's motion envelope over a future time period based on the vehicle state information, including: Based on the vehicle state information, the vehicle's driving trajectory in the future time period is predicted using a vehicle kinematic model that considers Ackermann steering characteristics. Based on the vehicle's geometric parameters and the Ackermann steering characteristics, the inner wheel difference offset between the front wheel trajectory and the rear wheel trajectory is calculated when the vehicle travels along the driving trajectory; wherein, the geometric parameters include at least the wheelbase and track width; Based on the driving trajectory and the inner wheel difference offset, the sweep paths of each part of the vehicle body during the driving process are determined, and the sweep paths of each part together constitute the motion envelope of the vehicle body.

3. The method according to claim 1, characterized in that, The step of determining a collision risk for the vehicle based on the obstacle information and the vehicle motion envelope includes: Perform geometric overlap detection on a two-dimensional plane between the spatial region covered by the vehicle motion envelope and the obstacle outline in the obstacle information; If an intersection area is detected between the vehicle motion envelope and the obstacle outline on a two-dimensional plane, then the collision risk is determined to exist.

4. The method according to claim 1, characterized in that, The step of generating active intervention control commands to avoid the collision risk based on the collision risk intensity includes: Calculate the intrusion depth and / or predict the collision time of the intersection region of the vehicle motion envelope and the obstacle contour in the obstacle information on a two-dimensional plane; Based on the intrusion depth and / or the predicted collision time, determine the current collision risk intensity and the corresponding initial intervention level; Starting with the intervention measures corresponding to the initial intervention level, the intervention measures corresponding to each intervention level are evaluated step by step in order of intervention intensity from low to high, and the first intervention measure that can avoid the collision risk is determined as the current active obstacle avoidance measure. Based on the current active obstacle avoidance measures, the active intervention control command is generated.

5. The method according to claim 4, characterized in that, Starting with the intervention measures corresponding to the initial intervention level, the intervention measures corresponding to each intervention level are evaluated progressively upwards in order of intervention intensity from low to high. The first intervention measure that can avoid the collision risk is determined as the current active obstacle avoidance measure, including: Obtain the intervention measures corresponding to the initial intervention level; wherein, different intervention levels include corresponding intervention measures in order of increasing intervention intensity. Based on a vehicle kinematics model that takes into account Ackermann steering characteristics, the vehicle body motion envelope is predicted after the vehicle performs the intervention corresponding to the currently assessed intervention level. If it is determined that the vehicle motion envelope and the obstacle outline do not intersect in a two-dimensional plane, then the intervention measure corresponding to the current assessed intervention level is determined as the current active obstacle avoidance measure. If it is determined that the vehicle motion envelope and the obstacle outline still have an intersection area on the two-dimensional plane, the next higher intervention level will be evaluated until an intervention level that can avoid the collision risk is found. If the collision risk cannot be avoided after going through all intervention levels, automatic emergency braking is triggered to bring the vehicle to a stop.

6. The method according to claim 5, characterized in that, Based on the vehicle's drive system configuration, the different intervention levels, in order of increasing intervention intensity, include corresponding intervention measures, including at least two of the following: The first level of intervention corresponds to intervention measures including: controlling the electric power steering system to apply steering correction torque; The second level of intervention includes: controlling the electric power steering system to make steering corrections and controlling the braking system to apply braking force to the wheels. The third level of intervention includes: controlling the torque difference between the left and right drive wheels to generate yaw moment; The fourth level of intervention includes triggering automatic emergency braking to bring the vehicle to a stop.

7. The method according to claim 6, characterized in that, The process of evaluating the intervention measures corresponding to each intervention level also includes: Based on the spatial location and intrusion depth of the intersection region between the vehicle motion envelope and the obstacle contour, determine the desired change in yaw rate required to eliminate the intersection region; Based on the current assessment of the intervention level, determine the target implementing agency corresponding to that intervention level; Based on the expected change in yaw rate, determine the control quantity that the target actuator should execute.

8. The method according to any one of claims 1-6, characterized in that, The method further includes: Obtain the vehicle's location information; The location information is compared with the stored collision risk geographic area data. If it is determined that the vehicle is in or is approaching the collision risk geographic area, a warning message is generated and output to the driver.

9. A vehicle obstacle avoidance control device, characterized in that, The device includes: The first processing unit is used to acquire vehicle status information and obstacle information in the direction of vehicle travel; The second processing unit is used to predict the vehicle's motion envelope in a future time period based on the vehicle kinematics model that takes into account Ackermann steering characteristics and the vehicle state information; wherein the vehicle motion envelope is used to indicate the spatial area traversed by the vehicle along the predicted trajectory. The third processing unit is used to generate an active intervention control command to avoid the collision risk if it is determined that the vehicle has a collision risk based on the obstacle information and the vehicle motion envelope.

10. An electronic device, characterized in that, include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1-8.

11. A vehicle, characterized in that, The vehicle is used to implement the method as described in any one of claims 1-8.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-8.

13. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method described in any one of claims 1-8.

Images