Vehicle detour control methods and vehicles

By using the motion status information of target objects behind and in adjacent lanes during vehicle detours to determine the degree of risk, generating and verifying lateral adjustment trajectories, the problem of insufficient perception of the surrounding traffic environment in existing detour planning technologies is solved, and safe and efficient detour control is achieved.

CN123104698APending Publication Date: 2026-11-17GREAT WALL MOTOR CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611592346.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-14
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing mass-produced NOA detour planning and control schemes lack the ability to perceive and adapt to the surrounding traffic environment during detours, posing safety hazards and resulting in low traffic efficiency. In particular, they are unable to effectively avoid lateral collisions when there is traffic flow behind or in adjacent lanes.

Method used

The risk level is determined by the motion status information of the target object in the area of ​​the vehicle's rear and adjacent lanes. An initial lateral adjustment trajectory is generated, and a collision check is performed based on the predicted driving trajectory envelope of the target object. The trajectory is adjusted to ensure a safety margin, and the vehicle is controlled to perform a bypass operation according to the target's lateral adjustment trajectory.

Benefits of technology

It improves the safety and smoothness of vehicles when navigating obstacles, makes up for the shortcomings of traditional solutions that do not pay enough attention to traffic flow behind and in adjacent lanes, and ensures that there is a safety margin for the future movement of target objects behind and in adjacent lanes during the navigation process, thus preventing side collisions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123104698A_ABST
    Figure CN123104698A_ABST
Patent Text Reader

Abstract

This application discloses a vehicle detour control method and a vehicle, belonging to the field of intelligent driving technology. When a vehicle is detouring around an obstacle, this application determines the risk level by using the motion state information of the target object within the area of ​​concern in the rear and adjacent lanes. This allows the vehicle to perceive the lateral collision risk posed by traffic flow in the rear and adjacent lanes, overcoming the shortcomings of traditional solutions that do not adequately consider the traffic flow in the rear and adjacent lanes. An initial lateral adjustment trajectory is generated based on the risk level and information about the obstacle ahead, ensuring the trajectory is adapted to the subsequent risk. Collision verification is performed on the initial lateral adjustment trajectory using the predicted driving trajectory envelope of the target object. If a collision fails, the trajectory is adjusted promptly to ensure that the target lateral adjustment trajectory maintains a safety margin with the future movement of the target object in the rear and adjacent lanes, preventing lateral collisions caused by occupying adjacent lanes. The vehicle is controlled to execute detouring actions according to the target lateral adjustment trajectory, balancing safety and efficiency, and improving the safety and smoothness of the vehicle when detouring around obstacles.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of intelligent driving technology, and more specifically, to a vehicle detour control method and vehicle in the field of visualization technology. Background Technology

[0002] With the rapid development of intelligent driving technology, NOA (Navigate on Autopilot) function has been widely used. As an important component of NOA function, detour planning control is used to avoid obstacles or detour in situations such as obstacles in front of the vehicle, low-speed vehicles, or limited driving space by adjusting the vehicle's lateral trajectory, thus ensuring continuous passage.

[0003] Currently, existing mass-produced NOA detour planning and control schemes typically only use obstacles, low-speed vehicles, or narrow spaces in front of the vehicle as detour triggers. They employ ILQR (Iterative Linear Quadratic Regulator) for lateral control, calculating the detour trajectory and corresponding lateral offset. These schemes lack sufficient perception and adaptation to the surrounding traffic environment during detours, posing certain safety hazards, and exhibiting low traffic efficiency in some scenarios. Summary of the Invention

[0004] This application provides a vehicle detour control method and vehicle to address the problem of lateral collision risk caused by neglecting rear and adjacent lane traffic flow in existing detour methods. By determining the risk level based on the motion state of targets in the rear and adjacent lanes, and generating and verifying the lateral trajectory in conjunction with obstacles in front, safe and efficient detour is achieved. The technical solution is as follows: On the one hand, a vehicle detour control method is provided, the method comprising: When the vehicle is bypassing an obstacle, the risk level of the vehicle is determined based on the motion state information of the target object in the area of ​​interest behind the vehicle and in the adjacent lane. The risk level is used to indicate the magnitude of the lateral collision risk posed by the traffic flow behind and in the adjacent lane to the vehicle's bypassing. Based on the risk level and the obstacle information in front of the vehicle, an initial lateral adjustment trajectory for the vehicle is generated; Based on the predicted driving trajectory envelope region of the target object, the initial lateral adjustment trajectory is collided, and if the collision verification fails, the initial lateral adjustment trajectory is adjusted to obtain the target lateral adjustment trajectory of the vehicle. The vehicle is controlled to perform a lateral adjustment trajectory to the target and perform a bypass maneuver.

[0005] On the one hand, a vehicle detour control device is provided, the device comprising: The determination module is used to determine the risk level of the vehicle based on the motion state information of the target object in the area of ​​interest behind the vehicle and in the adjacent lane when the vehicle is bypassing an obstacle. The risk level is used to represent the magnitude of the lateral collision risk posed by the traffic flow behind and in the adjacent lane to the vehicle's bypassing. The generation module is used to generate the initial lateral adjustment trajectory of the vehicle based on the risk level and the obstacle information in front of the vehicle. The processing module is used to perform collision verification on the initial lateral adjustment trajectory based on the predicted driving trajectory envelope region of the target object, and adjust the initial lateral adjustment trajectory when the collision verification fails, so as to obtain the target lateral adjustment trajectory of the vehicle. The control module is used to control the vehicle to perform a lateral adjustment trajectory according to the target to perform a bypassing operation.

[0006] On one hand, a vehicle is provided, the vehicle including one or more processors and one or more memories, the one or more memories storing at least one piece of program code, the program code being loaded and executed by the one or more processors to implement the vehicle detour control method.

[0007] On one hand, a computer-readable storage medium is provided, wherein at least one piece of program code is stored in the computer-readable storage medium, the program code being loaded and executed by a processor to implement the vehicle detour control method.

[0008] The vehicle detour control method and vehicle provided in this application determine the risk level by using the motion state information of the target object in the rear and adjacent lanes' monitoring area when the vehicle is detouring around an obstacle. This allows the vehicle to perceive the lateral collision risk posed by the traffic flow in the rear and adjacent lanes, thus overcoming the shortcomings of traditional solutions that do not pay enough attention to the traffic flow in the rear and adjacent lanes. An initial lateral adjustment trajectory is generated based on the risk level and the information of the obstacle in front, making the trajectory compatible with the risk behind. The initial lateral adjustment trajectory is used to perform collision verification with the predicted driving trajectory envelope area of ​​the target object. If the verification fails, the trajectory is adjusted in time to ensure that the target lateral adjustment trajectory maintains a safety margin with the future movement of the target object in the rear and adjacent lanes, preventing lateral collisions caused by occupying adjacent lanes. The vehicle is controlled to perform detouring actions according to the target lateral adjustment trajectory, balancing safety and efficiency, and improving the safety and smoothness of the vehicle when detouring around obstacles. Attached Figure Description

[0009] Figure 1 This is a schematic diagram of the implementation environment of a vehicle detour control method provided in an embodiment of this application; Figure 2 This is a flowchart of a vehicle detour control method provided in an embodiment of this application; Figure 3This is a flowchart of another vehicle detour control method provided in the embodiments of this application; Figure 4 This is a schematic diagram of the structure of a vehicle detour control device provided in an embodiment of this application; Figure 5 This is a schematic diagram of the structure of a vehicle provided in an embodiment of this application. Detailed Implementation

[0010] The technical solutions in this application will be clearly and thoroughly described below with reference to the accompanying drawings. In the description of the embodiments of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B. "And / or" in the text is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Furthermore, in the description of the embodiments of this application, "multiple" refers to two or more than two.

[0011] In the following text, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of technical features reflected. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature.

[0012] With the development of intelligent driving technology, NOA (Navigate on Autopilot) navigation assistance functions are gradually being mass-produced. Vehicles need to have the ability to avoid static obstacles, low-speed vehicles, and narrow spaces in scenarios such as highways and urban expressways. Existing detour control strategies typically use the area in front of the vehicle as the sole or primary planning basis. When an obstacle, low-speed vehicle, or narrow space is detected ahead, the lateral detour trajectory is calculated mainly based on the position and size of the target, adjusting the vehicle's lateral offset and driving path, which lacks comprehensive consideration of the surrounding traffic environment. At the same time, existing ILQR-based lateral detour control is independent of longitudinal deceleration and following control, optimizing only the lateral trajectory during detour, resulting in insufficient coordination between lateral and longitudinal control.

[0013] The above-mentioned scheme has the following problems in practical application: First, it is not adaptable to the dynamic traffic conditions around the vehicle when initiating the detour, and it is easy to cause spatial interference with other road users when deviating laterally; Second, in scenarios with relatively dense traffic flow, it is difficult to dynamically adjust the detour range according to the surrounding traffic conditions during the detour, and the safety margin is insufficient; Third, in scenarios with relatively open surrounding environment, the detour operation is not flexible enough and the traffic efficiency is low; Fourth, the ability to perceive and adapt to the surrounding traffic environment is insufficient, the coordination between lateral control and longitudinal control is insufficient, and it is difficult to balance safety and traffic efficiency.

[0014] Based on this, the vehicle detour control method provided in the embodiments of this application is proposed.

[0015] The implementation environment of the embodiments of this application is described below. See also... Figure 1 The implementation environment of the vehicle detour control method provided in this application embodiment includes a controller 101 and an on-board sensing device 102.

[0016] The controller 101 can be installed on the vehicle. The controller 101 can acquire relevant information and process it. In this embodiment, the controller 101 can acquire the motion state information of the target object and determine the risk level accordingly, and generate an initial lateral adjustment trajectory. The controller 101 is communicatively connected to the on-board sensing device 102, which collects information about the vehicle's surrounding environment and sends the collected sensing data to the controller 101.

[0017] In this embodiment, the vehicle-mounted sensing device 102 may include one or more of millimeter-wave radar, surround-view cameras, lidar, and ultrasonic radar, and is at least capable of collecting motion state information of target objects within the area of ​​interest behind the vehicle and in adjacent lanes, such as the relative longitudinal distance, relative lateral distance, relative speed, heading angle, and lane affiliation information of the target objects. Optionally, the vehicle-mounted sensing device 102 includes a rearward millimeter-wave radar and a surround-view camera to provide speed and distance information of rear targets and visual perception information of targets in adjacent lanes, respectively.

[0018] After introducing the implementation environment of the embodiments of this application, the technical solutions provided by the embodiments of this application are described below. (See also...) Figure 2 This paper will use a controller as an example to illustrate the implementation. It should be noted that the controller in this embodiment can be an independent controller in the vehicle, or it can be integrated into the vehicle's domain controller, central computing platform, or autonomous driving controller. This application does not limit the specific form of the controller.

[0019] Figure 2 A flowchart of a vehicle detour control method provided in an embodiment of this application is illustrated by way of example. Figure 2 The method includes the following steps: 201. When a vehicle is bypassing an obstacle, the controller determines the risk level of the vehicle based on the motion state information of the target object in the area of ​​interest behind the vehicle and in the adjacent lane. The risk level is used to indicate the magnitude of the lateral collision risk posed by the traffic flow behind and in the adjacent lane to the vehicle's bypass.

[0020] In some embodiments, the controller can determine in real time or periodically whether the vehicle is circumventing an obstacle. Situations where the vehicle is circumventing an obstacle include, but are not limited to, scenarios where there is an obstacle to be avoided in front of the vehicle, the speed of the vehicle in front is lower than the vehicle's expected speed, or the available space ahead is limited, requiring the vehicle to make lateral movements to avoid the obstacle or improve traffic efficiency. When the vehicle is circumventing an obstacle, the controller activates risk monitoring behind the vehicle and in adjacent lanes, acquiring motion state information of one or more target objects within the area of ​​interest behind the vehicle and in adjacent lanes through onboard sensing devices. The area of ​​interest behind the vehicle and in adjacent lanes can be pre-set or dynamically adjusted according to the vehicle's position, driving direction, and lane structure to cover vehicles in the rear and adjacent lanes that may laterally interfere with the vehicle's circumvention. The motion state information of the target object may include, but is not limited to, the target object's position, speed, acceleration, heading angle, lane location, and lateral movement trend relative to the vehicle.

[0021] Based on the aforementioned motion state information, the controller assesses the potential lateral collision risk caused by rear and adjacent lane traffic flow swerving around the vehicle, determining the risk level. The risk level quantifies the magnitude of the lateral collision risk posed by rear and adjacent lane traffic flow swerving around the vehicle; a higher risk level indicates a greater likelihood or urgency of a lateral collision between vehicles in the rear and adjacent lanes. In one example, the controller can comprehensively consider factors such as the overall density of traffic flow within the area of ​​interest, the relative motion relationship between a single target object and a vehicle, and the lateral approach trend of the target object, determining the risk level through preset assessment rules, lookup table mapping, or risk calculation models. It should be noted that this embodiment does not limit the specific method for determining the risk level, as long as the risk level characterizing the lateral collision risk can be obtained based on the motion state information of the target object within the area of ​​interest in the rear and adjacent lanes.

[0022] 202. The controller generates the vehicle's initial lateral adjustment trajectory based on the risk level and information about obstacles in front of the vehicle.

[0023] In some embodiments, after determining the risk level of the vehicle, the controller acquires obstacle information ahead of the vehicle and generates an initial lateral adjustment trajectory based on the risk level and the obstacle information. The obstacle information can be collected by onboard sensing devices and is used to characterize obstacles that the vehicle needs to avoid, such as the obstacle's location, size, type, lane area occupied, longitudinal distance from the vehicle, and passable space. Based on the obstacle information, the controller can determine the starting point, ending point, desired lateral offset direction, and minimum lateral adjustment required to avoid the obstacle, thereby ensuring that the initial lateral adjustment trajectory enables the vehicle to bypass the obstacle.

[0024] The risk level is used to constrain or guide the generation process of the initial lateral adjustment trajectory. A higher risk level indicates a greater risk of lateral collision from following and adjacent lane traffic, allowing the controller to employ a more conservative lateral adjustment strategy when generating the initial trajectory. This could include reducing the lateral adjustment amplitude, increasing the lateral safety margin, limiting the degree of deviation to adjacent lanes, or increasing trajectory smoothness requirements. Conversely, a lower risk level indicates a smaller risk from following and adjacent lane traffic, allowing the controller to adopt a more aggressive lateral adjustment strategy, such as allowing a larger lateral adjustment amplitude or a trajectory closer to the optimal travel path. It should be noted that this embodiment does not limit the specific way the risk level affects trajectory generation; it only needs to ensure that the generated initial lateral adjustment trajectory is correlated with the risk level, enabling the lateral adjustment amplitude or safety margin of the trajectory to be dynamically adjusted according to subsequent risks.

[0025] The initial lateral adjustment trajectory can be understood as a preliminary candidate detour trajectory. Its form can be a curve showing the lateral offset changing with longitudinal distance or time, or a series of discrete path points, curvature sequences, or steering wheel angle sequences. The controller can generate the initial lateral adjustment trajectory using a preset trajectory planning model, sampling planning method, geometric planning method, or optimization algorithm. During the generation process, information about obstacles ahead is taken as a necessary detour target, and the degree of risk is used as the basis for adjusting planning parameters or constraints. Thus, the controller can initially incorporate the risk level of traffic flow behind and in adjacent lanes while considering the detour requirements ahead.

[0026] 203. The controller performs collision verification on the initial lateral adjustment trajectory based on the predicted driving trajectory envelope region of the target object, and adjusts the initial lateral adjustment trajectory when the collision verification fails, so as to obtain the target lateral adjustment trajectory of the vehicle.

[0027] In some embodiments, after generating an initial lateral adjustment trajectory, the controller further obtains the predicted trajectory envelope region of the target object and performs collision verification on the initial lateral adjustment trajectory based on the predicted trajectory envelope region. The predicted trajectory envelope region can be determined based on the motion state information of the target object and is used to characterize the spatial area that the target object may occupy in the future over a period of time. For example, the controller can predict the longitudinal and lateral motion trajectory of the target object within a preset prediction time domain based on information such as the target object's current position, velocity, acceleration, heading angle, and lateral motion trend, and extend the motion trajectory and its possible position deviation range into an envelope region. The predicted trajectory envelope region can be a deterministic envelope or a probabilistic envelope that considers prediction uncertainty; this embodiment does not limit this.

[0028] The controller performs a collision check between the initial lateral adjustment trajectory and the predicted driving trajectory envelope of the target object to determine whether the vehicle will collide with the target object at some future moment while traveling along the initial lateral adjustment trajectory. The collision check can be performed by checking one or more trajectory points on the initial lateral adjustment trajectory to determine whether each trajectory point falls within the predicted driving trajectory envelope, or whether the minimum distance between each trajectory point and the predicted driving trajectory envelope meets a preset safety margin; or by checking the overlap between trajectory segments of the initial lateral adjustment trajectory and the predicted driving trajectory envelope. If there is no overlap between the initial lateral adjustment trajectory and the predicted driving trajectory envelope, or if the distance between them always meets the preset safety margin, the collision check is considered successful; if a trajectory point falls within the predicted driving trajectory envelope, or if the distance between them is less than the preset safety margin, the collision check is considered unsuccessful.

[0029] If the collision check fails, the controller adjusts the initial lateral adjustment trajectory to obtain the target lateral adjustment trajectory. The adjustment process includes, but is not limited to: reducing the lateral adjustment range of the initial lateral adjustment trajectory to minimize the vehicle's encroachment on the lateral space of adjacent lanes or obstacles; changing the offset direction or rate of the lateral adjustment trajectory; adjusting the curvature or smoothness of the trajectory to move it further away from the predicted driving trajectory envelope of the target object; or regenerating the initial lateral adjustment trajectory and employing a more conservative lateral adjustment strategy during regeneration. The controller can cyclically perform collision check and trajectory adjustment until the adjusted lateral adjustment trajectory meets the collision check conditions, or the lateral adjustment range reaches a preset minimum adjustment range limit. When the adjusted lateral adjustment trajectory meets the collision check conditions, the controller designates this adjusted lateral adjustment trajectory as the target lateral adjustment trajectory. If adjusting to the minimum adjustment range limit still fails to meet the collision check conditions, the controller can use the trajectory corresponding to this minimum adjustment range as the target lateral adjustment trajectory and trigger the corresponding safety fallback strategy.

[0030] In this way, the target's lateral trajectory adjustment can not only meet the detour requirements of obstacles in front, but also maintain a safety margin with the future movement of targets behind and in adjacent lanes. Thus, the risk constraints of traffic flow behind and in adjacent lanes are introduced in the detour planning stage, reducing the risk of lateral collisions caused by vehicles approaching from behind or vehicles in adjacent lanes running parallel during the detour process.

[0031] 204. The controller controls the vehicle to perform a bypass maneuver according to the target's lateral adjustment trajectory.

[0032] In some embodiments, after obtaining the target lateral adjustment trajectory, the controller converts the target lateral adjustment trajectory into control commands executable by the vehicle and sends them to the vehicle actuators to control the vehicle to perform a circumferential movement according to the target lateral adjustment trajectory. The control commands may include, but are not limited to, one or more of steering wheel angle commands, steering torque commands, lateral acceleration commands, trajectory point sequence commands, or curvature commands. The vehicle actuators may include a steering system, such as an electric power steering system, for adjusting the lateral movement of the vehicle according to the control commands, so that the vehicle travels along the path defined by the target lateral adjustment trajectory, thereby bypassing obstacles or avoiding confined spaces ahead.

[0033] During the process of controlling the vehicle to perform a bypass maneuver, the controller can acquire the motion status information of target objects within the area of ​​interest behind the vehicle and in adjacent lanes in real time or periodically, and re-determine the risk level. Then, following the aforementioned steps, it regenerates or updates the target's lateral adjustment trajectory to achieve closed-loop control. If the risk level changes during the bypass, for example, if a change in the motion trend of the target object behind the vehicle leads to an increase or decrease in the risk of a side collision, the controller can adjust the lateral control command accordingly to increase or decrease the lateral adjustment amplitude. Optionally, the controller can also coordinate the control of the vehicle's longitudinal movement, such as adjusting the vehicle speed or deceleration, to cooperate with the lateral bypass maneuver, ensuring that the vehicle maintains a safety margin with the target objects behind and in adjacent lanes while completing the bypass. It should be noted that this embodiment does not limit the specific method of longitudinal coordinated control, as long as it can control the vehicle to complete the bypass based on the target's lateral adjustment trajectory.

[0034] After the vehicle completes the detour maneuver according to the target lateral adjustment trajectory, the controller can keep the vehicle in the new lane or return to the center line of the original lane, and can decide whether to reactivate or deactivate rear risk monitoring based on changes in the risk level. In this way, the controller can actually apply the target lateral adjustment trajectory, after collision verification of the predicted driving trajectory envelope area of ​​the target object behind and in adjacent lanes, to the vehicle, so that the vehicle's detour maneuver meets the detour requirements ahead while taking into account the lateral collision risk of traffic behind and in adjacent lanes.

[0035] The vehicle detour control method provided in this application determines the risk level by using the motion state information of the target object within the area of ​​concern in the rear and adjacent lanes when the vehicle is detouring around an obstacle. This allows the system to perceive the lateral collision risk posed by traffic flow in the rear and adjacent lanes, thus overcoming the shortcomings of traditional solutions that do not adequately consider the traffic flow in the rear and adjacent lanes. An initial lateral adjustment trajectory is generated based on the risk level and the information of the obstacle ahead, ensuring that the trajectory is adapted to the subsequent risk. The initial lateral adjustment trajectory is then used to perform collision verification with the predicted driving trajectory envelope of the target object. If a collision fails, the trajectory is adjusted in a timely manner to ensure that the target lateral adjustment trajectory maintains a safety margin with the future movement of the target object in the rear and adjacent lanes, preventing lateral collisions caused by occupying adjacent lanes. The system controls the vehicle to perform detouring actions according to the target lateral adjustment trajectory, balancing safety and efficiency while improving the safety and smoothness of the vehicle when detouring around obstacles.

[0036] It should be noted that steps 201-203 above are a simplified explanation of the vehicle detour control method provided in the embodiments of this application. The following will provide a more detailed explanation of the vehicle detour control method provided in the embodiments of this application, using some examples. See [link to relevant documentation]. Figure 3 Taking the controller as the executing entity as an example, the method includes the following steps.

[0037] 301. The controller establishes a vehicle coordinate system with a preset reference point on the vehicle as the origin, and marks the area of ​​interest based on the vehicle coordinate system. The area of ​​interest covers the area within the preset longitudinal detection distance of the vehicle in the longitudinal direction, and covers at least the current lane of the vehicle and one adjacent lane of the current lane in the lateral direction.

[0038] In some embodiments, the controller can establish a vehicle coordinate system with a preset reference point on the vehicle as the origin, and calibrate the area of ​​interest based on this vehicle coordinate system. The preset reference point can be selected according to actual calibration requirements, such as the rear axle center, the vehicle's center of gravity, the front axle center, or other suitable vehicle positions as references. The vehicle coordinate system may include a longitudinal axis extending along the vehicle's longitudinal direction and a transverse axis extending along the vehicle's transverse direction to facilitate the description of the target object's position, distance, and motion trend relative to the vehicle.

[0039] The controller calibrates the region of interest based on the vehicle's coordinate system. The region of interest covers the area within a preset longitudinal detection distance of the vehicle. This preset longitudinal detection distance can be pre-set based on vehicle speed, detour conditions, sensing device performance, or scenario requirements, and can also be dynamically adjusted during actual operation. Laterally, the region of interest covers at least the vehicle's current lane and one adjacent lane, to cover the adjacent lane space that the vehicle may encroach on while detouring around an obstacle, as well as the adjacent lane area that may appear when approaching from behind. The shape of the region of interest can be rectangular, trapezoidal, fan-shaped, or other geometric shapes suitable for sensing coverage. Through this calibration, the controller can collect motion state information of target objects within the defined region of interest, ensuring effective coverage of relevant targets behind and in adjacent lanes when the vehicle is detouring around an obstacle, while avoiding excessive monitoring range that would waste computing power.

[0040] 302. The controller acquires motion state information of the target object within the area of ​​interest.

[0041] In some embodiments, after calibrating the area of ​​interest based on the vehicle coordinate system, the controller can acquire motion state information of target objects within the area of ​​interest through onboard sensing devices. These onboard sensing devices may include any one or more sensors for detecting targets around the vehicle, such as radar, cameras, lidar, ultrasonic sensors, etc. This embodiment does not limit the type, number, or installation location of the sensing devices. The controller can scan the area of ​​interest in real-time or periodically according to a preset sampling frequency or preset triggering conditions to acquire sensing data of one or more target objects. The target object may be a dynamic vehicle within the area of ​​interest, or other traffic participants that may affect the vehicle while it is navigating around an obstacle. The motion state information may include one or more of the following: target object position information relative to the vehicle, speed information, acceleration information, heading angle information, lane affiliation information, lateral movement trend information, target size information, or target type information. The position information may include longitudinal relative distance and lateral relative distance; the speed information may include the absolute speed of the target object and / or the relative speed between the target object and the vehicle; the lateral movement trend information can be used to characterize whether the target object is approaching the vehicle's current lane or an adjacent lane that the vehicle intends to enter while navigating around an obstacle.

[0042] After acquiring the sensing data, the controller can preprocess the data to improve the stability and reliability of the motion state information. Preprocessing may include, but is not limited to: consistency verification of multiple consecutive frames of sensing data, correlation tracking of the same target, filtering out false detections or misdetections that do not meet preset validity conditions, and smoothing filtering of the target's position and velocity. Thus, the controller obtains the motion state information of the target object within the area of ​​interest and uses it to determine the risk level. Optionally, the controller can dynamically adjust the acquisition range, acquisition frequency, or preprocessing parameters based on vehicle speed, the scenario of the vehicle navigating around obstacles, or the distribution of targets within the area of ​​interest, to balance sensing coverage and computational resource consumption. It should be noted that this embodiment does not limit the specific method of acquiring the motion state information of the target object within the area of ​​interest, as long as the motion state information of the target object used for risk level assessment can be obtained within the designated area of ​​interest.

[0043] In one possible implementation, an example of obtaining motion state information of a target object within the region of interest is as follows: Using at least one vehicle-mounted sensing device, sensing data of target objects within the area of ​​interest are collected at a preset sampling frequency. The sensing data includes at least one type of data used to determine the number of target objects, the relative longitudinal distance between the target objects and the vehicle, and the relative lateral motion state. Consistency verification is performed on multiple consecutive frames of sensing data, and target data that does not meet the preset consistency conditions are filtered out to obtain preprocessed motion state information of the target objects.

[0044] In some embodiments, the controller can periodically scan or detect the area of ​​interest using an onboard sensing device at a preset sampling frequency. This preset sampling frequency can be pre-set based on the performance of the sensing device, vehicle speed, or the real-time requirements of the vehicle navigating obstacles. In one example, the controller can collect data at a sampling frequency of 10Hz, i.e., one frame of sensing data is collected every 100ms. It is understood that the specific value of the sampling frequency is not a limiting condition of this application. Higher sampling frequencies can capture changes in the motion state of the target object more promptly, but correspondingly increase the overhead of computing and communication resources; lower sampling frequencies, while reducing resource consumption, may lead to a lag in response to fast-moving targets.

[0045] Vehicle-mounted sensing devices may include at least one of millimeter-wave radar, surround-view cameras, lidar, or ultrasonic radar. Different sensing devices can provide different types of sensing data. For example, millimeter-wave radar can provide relatively accurate information on the relative distance and speed between the target object and the vehicle; surround-view cameras can provide relatively accurate information on the target object's category and lateral position; and lidar can provide the target object's three-dimensional position and contour information. The controller can use a single sensing device for data acquisition or fuse data from multiple sensing devices. For example, in one possible implementation, the controller can simultaneously acquire sensing data from rear-facing millimeter-wave radar and surround-view cameras, and then perform spatiotemporal alignment and fusion to obtain more complete and accurate target object sensing data.

[0046] The perception data may include at least one type of data used to determine the number of target objects, the relative longitudinal distance between the target objects and the vehicle, and the relative lateral motion state. Specifically, the data used to determine the number of target objects may include target object identification information or detection box information detected by the perception device within the area of ​​interest. The controller can determine the number of valid target objects within the area of ​​interest by associating the same target object across multiple frames. The data used to determine the relative longitudinal distance may include the distance between the target object and the vehicle along the longitudinal axis of the vehicle's coordinate system, such as the longitudinal distance between the target object and the center of the vehicle's rear axle or center of mass. The data used to determine the relative lateral motion state may include information such as the distance between the target object and the vehicle along the transverse axis of the vehicle's coordinate system, lateral velocity, lateral acceleration, or heading angle. This information can be used to determine whether the target object is approaching the vehicle's current lane or an adjacent lane that the vehicle intends to enter while circling an obstacle.

[0047] After acquiring multiple frames of perception data, the controller can perform consistency checks on consecutive frames of perception data to filter out unstable target data caused by perception noise, false detections, or environmental interference. Consistency checks may include comparing the position, speed, heading angle, or lane assignment information of the same target object across consecutive frames to determine whether its changes conform to physical motion laws or exceed a preset reasonable range. For example, in one possible implementation, if the longitudinal or lateral distance change of a target object in consecutive frames exceeds a preset threshold, or if its heading angle exhibits a sudden change that does not conform to normal vehicle driving patterns, the controller can determine that the target object's data does not meet the preset consistency conditions and filter it out as unstable data. In another possible implementation, the controller can set a three-frame sliding window; if a target object appears only in a single frame and not in adjacent frames, it is determined as a false detection or misdetection target and filtered out.

[0048] After filtering out target data that does not meet preset consistency conditions, the controller correlates and outputs the remaining target data to obtain preprocessed motion state information of the target object. This motion state information can be used to determine the level of risk. Optionally, the controller can also track the preprocessed target object and continuously update its motion state information during the tracking process, so as to perform risk assessment and trajectory planning based on stable and continuous target object motion state information.

[0049] 303. When a vehicle is bypassing an obstacle, the controller determines the risk level of the vehicle based on the motion state information of the target object in the area of ​​interest behind the vehicle and in the adjacent lane. The risk level is used to indicate the magnitude of the lateral collision risk posed by the traffic flow behind and in the adjacent lane to the vehicle's bypass.

[0050] In one possible implementation, the situation in which the vehicle is circumventing the obstacle includes at least one of the following: If a static obstacle is detected in front of the vehicle, the obstacle avoidance and detour conditions are met; if the speed of the target object in front of the vehicle is lower than the vehicle's current expected speed, the efficiency detour conditions are met; if the passable space in front of the vehicle meets the preset space restriction conditions, the passive avoidance and detour conditions are met.

[0051] In some embodiments, the controller can acquire real-time information about the environment ahead of the vehicle through onboard sensing devices and determine whether the vehicle currently meets preset detour triggering conditions. Detour triggering conditions characterize scenarios where the vehicle needs or is suitable to make lateral movements to avoid targets ahead or improve traffic efficiency. Specifically, the controller can determine at least one of the following three conditions individually or in combination; when any one of these conditions is met, it is determined that the vehicle is detouring around an obstacle.

[0052] In one example, if a static obstacle is detected in front of the vehicle, the obstacle avoidance condition is met. Static obstacles can include, but are not limited to, curbs, guardrails, posts, cones, construction barriers, accident vehicles, illegally parked vehicles, and other fixed or long-term stationary objects. The controller can identify the type and outline of the obstacle using a forward-facing camera, and obtain the longitudinal distance, lateral offset, and lane area occupied by the obstacle from the vehicle using millimeter-wave radar or lidar. When the controller determines that the static obstacle is located in front of the vehicle's current travel path, and the longitudinal distance between the vehicle and the obstacle is less than a preset obstacle avoidance trigger distance, the obstacle avoidance condition can be determined to be met. For example, the obstacle avoidance trigger distance can be set to 50 to 80 meters; if the vehicle speed is high, this distance can be increased accordingly. In addition, the controller can also determine whether the obstacle overlaps with the vehicle laterally: if the lateral area occupied by the obstacle overlaps with the lateral range of the vehicle's current travel trajectory, and the overlap width exceeds a preset threshold, it indicates that the vehicle cannot safely pass along the current path and needs to trigger a lateral detour. At this point, the controller determines that the vehicle is detouring around an obstacle, and that the detouring scenario is obstacle avoidance.

[0053] In another example, if the speed of a target object ahead is detected to be lower than the vehicle's current expected speed, the efficient detour condition is met. The target object ahead is typically a vehicle traveling in the same or adjacent lane. The controller can continuously track the movement of the target object using forward-facing radar or a camera to obtain its absolute speed or relative speed with the vehicle. The vehicle's current expected speed can be determined based on the vehicle's set cruise speed, road speed limits, overall traffic flow speed, or driving mode. When the controller determines that the speed of the target object ahead is lower than the vehicle's expected speed, and the speed difference reaches a preset threshold, the efficient detour condition can be determined. For example, if the target object's speed is less than 80% of the vehicle's expected speed, or the speed difference is greater than 5 m / s, and this state persists for more than a preset duration (e.g., 2 seconds), the controller can determine that the vehicle is detouring around the obstacle, and the detour scenario is an efficient detour. Optionally, the controller can also make a judgment based on the longitudinal distance between the target object ahead and the vehicle. For example, it can trigger an efficient detour only when the longitudinal distance is less than a preset distance (such as 100 meters) and the speed difference meets the conditions, so as to avoid making a detour decision for a low-speed vehicle in the distance too early.

[0054] In another example, if the passable space in front of the vehicle meets the preset space restriction conditions, then the passive avoidance and detour conditions are met. Passable space can refer to the vehicle's current lane or a certain area of ​​road in front of the vehicle that allows for safe passage. The controller can use cameras, radar, or lidar to fuse information such as road boundaries, lane lines, and obstacle distribution in front of the vehicle to calculate the width, length, or geometry of the passable space. When the width of the passable space is less than the minimum width required for safe passage, or when the margin between the boundary of the passable space and the vehicle's current trajectory is less than a preset safety margin, the controller can determine that the space restriction conditions are met. For example, the minimum width can be set to the vehicle's width plus a safety margin of 0.3 to 0.5 meters; if the passable width is less than this minimum value, it indicates that the vehicle cannot pass safely without lateral deviation. For another example, when there are large vehicles driving alongside the vehicle, construction areas encroaching on the lane, or the road narrowing in front of the vehicle, causing partial occupation of the passable area in the current lane, the controller can determine that the vehicle is detouring around the obstacle, and the detouring scenario is a passive avoidance and detour.

[0055] It should be noted that the above three conditions can be judged individually or in any combination. The controller can determine that the vehicle is detouring around an obstacle when any one of the conditions is met, and activate steps such as rear and adjacent lane risk monitoring, risk level determination, and lateral trajectory adjustment generation. Optionally, the controller can also adaptively adjust the thresholds in the above judgment conditions based on factors such as vehicle speed, road type, weather conditions, or perception confidence level to improve the accuracy and robustness of the detouring trigger judgment.

[0056] In one possible implementation, an example of determining the risk level of a target object based on the motion state information of the target object within the area of ​​interest behind the vehicle and in adjacent lanes is as follows: The target objects corresponding to the motion state information are effectively screened to obtain valid target objects for risk assessment. Based on the valid target objects, risk characterization parameters are extracted from both group and individual dimensions. The group-level risk characterization parameters are used to characterize the overall density of traffic flow in the area of ​​interest, while the individual-level risk characterization parameters are used to characterize the urgency of a side collision between the valid target object and a vehicle. The group-level and individual-level risk characterization parameters are then classified and quantified according to preset classification conditions to obtain environmental-level risk indicators and target-level risk indicators. The environmental-level risk indicators and target-level risk indicators are then fused to generate the risk level of the target object.

[0057] In some embodiments, the controller first performs validity screening on the target objects corresponding to the motion state information to obtain valid target objects for risk assessment. Validity screening is used to exclude objects unsuitable as sources of rear detour risk from all detected target objects within the area of ​​interest, such as stationary obstacles, parked vehicles, non-motorized vehicles, pedestrians, and targets with low perception confidence or unstable trajectories. Through validity screening, the controller retains only dynamic vehicles that may pose a side collision risk to vehicle detours as valid target objects, thereby reducing the computational load of risk assessment and improving the accuracy of the assessment results. Validity screening can be based on one or more of the following conditions: the motion state of the target object, target type, lane it is in, relative position to the vehicle, or stability of the perception data; this application does not limit the specific screening conditions.

[0058] Secondly, the controller extracts risk characterization parameters from both group and individual dimensions based on the effective target objects. Group-level risk characterization parameters characterize the overall density of traffic flow within the area of ​​interest, reflecting the macro-level game environment created by rear and adjacent lane traffic flow on vehicle detours. For example, group-level risk characterization parameters could include the number, distribution density, average vehicle distance, and lane occupancy ratio of effective dynamic motor vehicles within the area of ​​interest—parameters that reflect the overall congestion level of the traffic flow. Individual-level risk characterization parameters characterize the urgency of a lateral collision between the effective target object and a vehicle, reflecting the micro-level threat posed by a single rear or adjacent lane vehicle on vehicle detours. For example, individual-level risk characterization parameters could include the relative lateral distance, relative lateral speed, lateral collision duration, and lateral approach trend between the target object and the vehicle—parameters that reflect the lateral danger level of a single effective target object. The group and individual-level risk characterization parameters describe the comprehensive impact of rear and adjacent lane traffic flow on vehicle detours from both macro and micro perspectives, avoiding the limitations of risk assessment based on only a single dimension.

[0059] Then, the controller performs hierarchical quantification on the risk characterization parameters at both the group and individual levels according to preset hierarchical conditions, resulting in environmental and target-level risk indicators. Hierarchical quantification refers to mapping continuous or discrete risk characterization parameters to discrete categories according to preset hierarchical conditions. For group-level risk characterization parameters, the controller can compare their numerical values ​​or distributions with one or more preset thresholds, classifying them into multiple traffic flow conditions, such as first, second, and third traffic flow conditions, where traffic flow density increases sequentially, thus obtaining environmental-level risk indicators. For individual-level risk characterization parameters, the controller can compare their numerical values ​​with one or more preset thresholds, classifying them into multiple collision risk levels, such as first and second collision risks, where the urgency of the second collision risk is higher than that of the first collision risk, thus obtaining target-level risk indicators. Environmental-level risk indicators are used to characterize the overall risk of rear and adjacent lane traffic flow obstructing vehicle movement from a group perspective, while target-level risk indicators are used to characterize the lateral collision risk of a single target object obstructing vehicle movement from an individual perspective. The specific number of categories and thresholds for hierarchical quantification can be determined based on actual calibration, and this application does not limit this.

[0060] Finally, the controller fuses environmental-level risk indicators with target-level risk indicators to generate the risk level of the target object. This fusion can comprehensively consider both the overall traffic flow risk at the group level and the single-target collision risk at the individual level, generating a comprehensive risk level that fully reflects the lateral collision risk posed by rear and adjacent lane traffic flow to the vehicle's detour. In one example, when the target-level risk indicator corresponds to multiple valid target objects, the controller can select the target-level risk indicator with the highest risk level from among the multiple valid target objects as the target-level representative value. This target-level representative value is then fused with the environmental-level risk indicator according to a preset mapping relationship, ensuring that the final generated risk level is not lower than the risk level determined solely by the target-level representative value, and is appropriately adjusted upwards as the overall density of rear and adjacent lane traffic increases. Through this fusion, the controller can transform the complex rear and adjacent lane traffic flow status into a standardized, directly usable risk level.

[0061] In one possible implementation, an example of extracting risk characterization parameters from both the group and individual dimensions based on the effective target objects is as follows: The number of dynamic motor vehicles among the effective target objects within the monitoring area is counted; for vehicles behind the effective target objects, the lateral collision duration between the vehicles behind is calculated, where the lateral collision duration is the ratio of the relative lateral distance to the relative lateral velocity, and the lateral collision duration is calculated when the relative lateral velocity indicates that the vehicles are close to each other.

[0062] In some embodiments, for the group dimension, the controller counts the number of dynamic motor vehicles among valid target objects within the area of ​​interest. Specifically, the controller can determine whether a target object is a dynamic motor vehicle based on its motion state information: if the absolute speed of the target object is greater than a preset speed threshold, for example, greater than 0.5 m / s, and the target type is a motor vehicle, and it is located on a drivable road within the area of ​​interest, then it is identified as a dynamic motor vehicle; if the target object is a stationary obstacle, a roadside parked vehicle, a non-motorized vehicle, a pedestrian, or an animal, it is not included in the number of dynamic motor vehicles. The controller counts all dynamic motor vehicles within the area of ​​interest that meet the above conditions to obtain the number of dynamic motor vehicles. This number can characterize the overall density of traffic flow behind and in adjacent lanes from the group dimension. The larger the number, the more limited the space available for vehicles to detour in the rear and adjacent lanes, and the higher the risk of traffic flow game.

[0063] For individual vehicles, the controller calculates the lateral collision duration between the vehicle and other vehicles within the valid target group. Rear vehicles can include those directly behind the vehicle and those in adjacent lanes. The controller acquires the position and speed information of rear vehicles relative to the vehicle through onboard sensing devices. Using a vehicle coordinate system established with the rear axle center as the origin, the vertical axis pointing forward and the horizontal axis pointing to the left, the relative lateral distance between rear vehicles can be expressed as the absolute value of the difference between the lateral coordinates of the rear vehicle and the vehicle's lateral coordinates, denoted as Δx; the relative lateral velocity can be expressed as the difference between the lateral velocities of the rear vehicle and the vehicle's lateral velocity, denoted as Δv. When Δv indicates that the rear vehicles are approaching each other laterally, i.e., the lateral distance between them is decreasing, the controller calculates the lateral collision duration: Lateral collision duration = Δx / |Δv|. If Δv indicates that the vehicles are moving away from each other, i.e., the lateral distance is increasing, the controller does not calculate the lateral collision duration, or sets it to infinity, indicating no risk of lateral collision.

[0064] For example, a vehicle is traveling near the center line of its current lane, and a vehicle in an adjacent lane is approaching laterally towards the vehicle's lane from behind. The controller obtains that the relative lateral distance between the two vehicles is 1.5m, the relative lateral speed is 0.5m / s, and the direction is towards the approaching vehicle. Therefore, the lateral collision duration is 1.5 / 0.5 = 3s. If the preset collision duration threshold is 3s, then this lateral collision duration equals the threshold and can be classified as a second collision risk, which is more urgent than the first collision risk. If the relative lateral distance is 3m and the relative lateral speed is 0.5m / s, then the lateral collision duration is 6s, which is greater than the threshold and can be classified as a first collision risk.

[0065] In this way, the controller can obtain the number of dynamic motor vehicles from the group dimension and the lateral collision duration of vehicles behind from the individual dimension, which are used to determine environmental-level risk indicators and target-level risk indicators, respectively.

[0066] In one possible implementation, the risk characterization parameters at the group level and the individual level are classified and quantified according to preset classification conditions to obtain environmental-level risk indicators and target-level risk indicators. An example is shown below: Based on the number of dynamic motor vehicles, a preset density threshold, and a first preset classification condition, environmental-level risk indicators are determined. A first traffic flow condition is defined as a number less than a preset threshold; a second traffic flow condition is defined as a number greater than or equal to the preset threshold but less than the preset density threshold; and a third traffic flow condition is defined as a number greater than or equal to the preset density threshold. The traffic density corresponding to the first, second, and third traffic flow conditions increases sequentially. Based on the lateral collision duration, a preset collision duration threshold, and a second preset classification condition, target-level risk indicators are determined. A first collision risk is defined as a lateral collision duration greater than the preset collision duration threshold; a second collision risk is defined as a lateral collision duration less than or equal to the preset collision duration threshold. The urgency of the second collision risk is higher than that of the first collision risk.

[0067] In some embodiments, for environmental risk indicators, the controller counts the number of dynamic motor vehicles among valid target objects within the area of ​​interest. Dynamic motor vehicles refer to motor vehicles in motion, excluding stationary obstacles, parked vehicles, non-motorized vehicles, pedestrians, etc. The controller compares the counted number of dynamic motor vehicles with a preset quantity threshold and a preset density threshold. The preset quantity threshold and preset density threshold can be determined based on actual vehicle calibration and are used to classify traffic flow density. For example, the preset quantity threshold can be set to 1, and the preset density threshold can be set to 3. When the number of dynamic motor vehicles is less than the preset quantity threshold, for example, 0, it indicates that there are basically no dynamic vehicles in the area of ​​interest, and the controller determines it as the first traffic flow condition. When the number of dynamic motor vehicles is greater than or equal to the preset quantity threshold and less than the preset density threshold, for example, 1 or 2, it indicates that there are a small number of dynamic vehicles in the area of ​​interest, and the controller determines it as the second traffic flow condition. When the number of dynamic motor vehicles is greater than or equal to the preset density threshold, for example, 3 or more, it indicates that the traffic flow in the area of ​​interest is dense, and the controller determines it as the third traffic flow condition. The controller uses the determined conditions as traffic flow density categories to obtain the environmental risk indicator. It is understood that the specific values ​​of the preset quantity threshold and the preset density quantity can be calibrated and adjusted according to factors such as road type, vehicle speed, and perception range, and this application does not limit them.

[0068] For target-level risk indicators, the controller calculates the lateral collision duration for vehicles behind valid target objects. The lateral collision duration characterizes the urgency of a lateral collision between two vehicles and can be defined as the ratio of relative lateral distance to relative lateral velocity. The relative lateral distance refers to the distance between the target object and the vehicle along the horizontal axis of the vehicle's coordinate system, and the relative lateral velocity is the difference between the lateral velocity of the target object and the lateral velocity of the vehicle. The controller can calculate the lateral collision duration when the relative lateral velocity indicates that the target object and the vehicle are approaching each other; if they are far apart, the lateral collision duration can be considered infinite or not calculated. The controller compares the calculated lateral collision duration with a preset collision duration threshold. The preset collision duration threshold can be determined based on real-vehicle calibration and is used to distinguish different levels of collision risk. For example, the preset collision duration threshold can be set to 3 seconds. When the lateral collision duration exceeds a preset collision duration threshold, such as more than 3 seconds, it indicates that the target object is approaching laterally at a relatively slow speed, and the probability of a lateral collision occurring within a short period is low. The controller identifies this as a first-level collision risk. When the lateral collision duration is less than or equal to the preset collision duration threshold, such as less than or equal to 3 seconds, it indicates that the target object is approaching laterally at a relatively fast speed, and there is a higher risk of a lateral collision. The controller identifies this as a second-level collision risk, with the second-level collision risk having a higher urgency than the first-level collision risk. The controller uses the identified collision risks as target-level risk indicators.

[0069] Through the above methods, the controller can quantify and classify the risk of traffic flow behind and in adjacent lanes from both group and individual dimensions, obtaining environmental-level risk indicators and target-level risk indicators. It is understood that the specific values ​​of the preset quantity threshold, preset density threshold, and preset collision duration threshold mentioned above are merely examples. In practical applications, they can be calibrated according to vehicle type, perception capabilities, road conditions, etc., and this application does not impose any limitations on them.

[0070] In one possible implementation, an example of fusing environmental-level risk indicators with target-level risk indicators to generate the vehicle's risk level is as follows: When multiple valid target objects correspond to the target-level risk indicators, the risk characterization parameter with the highest risk level is determined from the risk characterization parameters of the multiple valid target objects and used as the target-level representative value. Based on the environmental-level risk indicators and the target-level representative value, the risk level is generated by fusion according to the preset risk mapping relationship. The risk mapping relationship is configured such that the risk level obtained after fusion is adjusted upward as the risk represented by the target-level representative value increases, and the risk level obtained after fusion is not lower than the risk level determined by the target-level representative value.

[0071] In some embodiments, the controller first determines whether a target-level risk indicator corresponds to multiple valid target objects. If multiple valid rear vehicles exist within the area of ​​interest, each vehicle corresponds to a target-level risk indicator. The controller selects the risk characterization parameter with the highest risk level from the risk characterization parameters of multiple valid target objects as the target-level representative value. For example, the controller can compare the lateral collision duration of each valid target object and select the target-level representative value corresponding to the shortest lateral collision duration; or compare parameters such as the lateral approach speed and relative lateral distance of each target object to comprehensively determine the target object with the highest risk level and use its risk characterization parameter as the target-level representative value. By selecting the target-level representative value with the highest risk level, it can be ensured that the fusion result prioritizes the most urgent lateral collision risk, avoiding the neglect of high-risk targets due to multi-target averaging or weakening processing.

[0072] Then, the controller generates a risk level based on environmental-level risk indicators and target-level representative values, according to a preset risk mapping relationship. This risk mapping relationship can be pre-configured in the form of a lookup table, rules, or formula. It is configured such that the risk level obtained after fusion increases as the risk represented by the target-level representative value increases, and the risk level obtained after fusion is not lower than the risk level determined solely by the target-level representative value. For example, the target-level representative value may include the first collision risk and the second collision risk, and the environmental-level risk indicators may include the first traffic flow condition, the second traffic flow condition, and the third traffic flow condition. The controller can generate the risk level according to the following mapping relationship: when the target-level representative value is the first collision risk, if the environmental-level risk indicator is the first traffic flow condition, the risk level is determined to be the first risk level; if the environmental-level risk indicator is the second traffic flow condition, the risk level is determined to be the second risk level; and if the environmental-level risk indicator is the third traffic flow condition, the risk level is determined to be the third risk level. When the target-level representative value is the second collision risk, the risk level is determined to be at least the fourth risk level, regardless of the environmental-level risk indicator's operating condition. If the environmental-level risk indicator is the second or third traffic flow condition, the risk level is further adjusted to the fifth risk level. Therefore, the final risk level reflects both the urgency of a side collision with the most dangerous target object and the impact of the overall traffic density behind and in adjacent lanes on detour safety, and will not be lower than the risk level determined solely by the target-level representative value.

[0073] Through the above fusion method, the controller can uniformly map environmental risks at the group level and target-level risks at the individual level into a standardized risk level.

[0074] 304. The controller generates the vehicle's initial lateral adjustment trajectory based on the risk level and information about obstacles in front of the vehicle.

[0075] In one possible implementation, an example of generating the vehicle's initial lateral adjustment trajectory based on the level of risk and information about obstacles in front of the vehicle is as follows: The lateral adjustment trajectory of the vehicle is optimized using a trajectory optimization model to generate a pending lateral adjustment trajectory; the cost parameters and boundary constraints of the trajectory optimization model are adjusted according to the risk level; and the initial lateral adjustment trajectory is determined based on the adjusted cost parameters, boundary constraints, and obstacle information in front of the vehicle.

[0076] In some embodiments, after determining the risk level of the vehicle and acquiring information about obstacles ahead, the controller can generate an initial lateral adjustment trajectory for the vehicle based on the risk level and the obstacle information. Specifically, the controller can optimize the lateral adjustment trajectory of the vehicle using a preset trajectory optimization model. The trajectory optimization model can be a trajectory planning model based on an optimization algorithm, a trajectory planning model based on a sampling method, a trajectory planning model based on a geometric method, or a trajectory planning model based on a learning method, etc. This embodiment does not limit the specific type of trajectory optimization model. The trajectory optimization model typically includes a cost function and constraints, where the cost function is used to evaluate the merits of candidate lateral adjustment trajectories, and the constraints are used to limit the boundaries that the lateral adjustment trajectory must satisfy, such as maximum lateral offset, lateral velocity, lateral acceleration, curvature, or safe distance from the lane boundary, etc.

[0077] In one example, the controller can first generate a potential lateral adjustment trajectory using the trajectory optimization model. This potential lateral adjustment trajectory can be a candidate trajectory obtained under the current default cost parameters and default boundary constraints, or it can be a pre-selected lateral adjustment trajectory. The controller adjusts the cost parameters and boundary constraints of the trajectory optimization model according to the risk level. Cost parameters can include one or more of the following: lateral safety weight, trajectory smoothness weight, traffic efficiency weight, and lateral offset cost weight; boundary constraints can include one or more of the following: maximum lateral adjustment amplitude, upper limit of lateral adjustment rate, upper limit of lateral acceleration, or minimum distance to the adjacent lane boundary. A higher risk level indicates a greater risk of lateral collision from rear and adjacent lane traffic flow, and the controller can correspondingly increase the lateral safety weight and tighten the boundary constraints to make the generated lateral adjustment trajectory more conservative and with a larger lateral safety margin. Conversely, a lower risk level indicates a smaller risk from rear and adjacent lane traffic flow, and the controller can correspondingly decrease the lateral safety weight and relax the boundary constraints to make the generated lateral adjustment trajectory closer to the path with optimal traffic efficiency.

[0078] After adjusting the cost parameters and boundary constraints, the controller, based on the adjusted cost parameters, adjusted boundary constraints, and information about obstacles ahead of the vehicle, uses the trajectory optimization model to resolve or continue optimizing the undetermined lateral adjustment trajectory, thereby determining the initial lateral adjustment trajectory. Information about obstacles ahead can include the obstacle's location, size, type, lane area it occupies, longitudinal distance from the vehicle, and passable space. This information can serve as a detour target or hard constraint in the trajectory optimization model to ensure that the generated initial lateral adjustment trajectory allows the vehicle to avoid the obstacle. For example, the controller can use the outline or occupied area of ​​the obstacle ahead as an insurmountable constraint region, and the required lateral offset direction and minimum lateral offset as optimization targets. Under the adjusted cost parameters and boundary constraints, it solves for an initial lateral adjustment trajectory that avoids the obstacle ahead while meeting the lateral safety requirements for rear risk levels.

[0079] In this way, the controller can simultaneously consider the detour requirements of obstacles ahead and the risk level of traffic flow behind and in adjacent lanes during the generation of the initial lateral adjustment trajectory. This ensures that the initial lateral adjustment trajectory has a lateral adjustment margin that matches the risk level before collision verification is performed in the predicted driving trajectory envelope area of ​​the target object, thus providing a reasonable basic trajectory for collision verification and trajectory adjustment.

[0080] In one possible implementation, the cost parameters include lateral safety weights, trajectory smoothness weights, and a maximum lateral offset limit. An example of adjusting the cost parameters and boundary constraints of the trajectory optimization model according to the risk level is as follows: If the risk level is below the first risk threshold, the horizontal safety weight is lowered to the first weight, and the maximum horizontal offset limit is increased to the first offset limit. If the risk level is greater than or equal to the first risk threshold and less than the second risk threshold, the horizontal safety weight is maintained at the second weight, and the maximum horizontal offset limit is set to the second offset limit, where the second weight is not less than the first weight and the second offset limit is less than the first offset limit. If the risk level is greater than or equal to the second risk threshold, the horizontal safety weight is raised to the third weight, and the maximum horizontal offset limit is decreased to the third offset limit, where the third weight is greater than the second weight and the third offset limit is less than the second offset limit.

[0081] In some embodiments, the controller can adjust the cost parameters and boundary constraints of the trajectory optimization model in tiers according to the risk level range in which the risk level falls. Lateral safety weights characterize the trajectory optimization model's emphasis on lateral safety margins between the vehicle and targets in the rear and adjacent lanes; a larger weight results in a trajectory that is more inclined to move away from rear and adjacent lane targets. Trajectory smoothness weights characterize the trajectory optimization model's emphasis on trajectory smoothness. The maximum lateral offset upper limit is a boundary constraint used to limit the maximum allowable lateral adjustment range of the vehicle during detours. The controller can pre-set a first risk threshold and a second risk threshold, where the first risk threshold is smaller than the second risk threshold, to divide the risk level into a first risk range, a second risk range, and a third risk range.

[0082] When the risk level is below the first risk threshold, it indicates that the lateral collision risk posed by traffic flow from behind and adjacent lanes to the vehicle's detour is low, and the rear space is relatively sufficient. The controller lowers the lateral safety weight to the first weight and increases the maximum lateral offset limit to the first offset limit. The first weight is a relatively low weight value, used to reduce the constraint of the trajectory optimization model on the lateral safety margin, so that the trajectory optimization model tends to choose a trajectory with higher traffic efficiency and more sufficient lateral adjustment, while meeting the detour requirements of obstacles ahead; the first offset limit is a relatively large offset limit, used to allow the vehicle to make larger lateral adjustments when detouring, thereby making fuller use of adjacent lane space to complete the detour. Optionally, the controller can also adjust the trajectory smoothness weight accordingly, so that the trajectory remains smooth even with a large lateral adjustment.

[0083] When the risk level is greater than or equal to the first risk threshold but less than the second risk threshold, it indicates that there is some traffic flow behind and in adjacent lanes, and vehicles may face slight lateral game risk when detouring. The controller maintains the lateral safety weight at the second weight and sets the maximum lateral offset upper limit to the second offset upper limit. The second weight is not less than the first weight, which is used to appropriately increase the emphasis on lateral safety margin based on the low-risk strategy, or at least maintain a safety constraint level no lower than that in the low-risk scenario; the second offset upper limit is less than the first offset upper limit, which is used to appropriately tighten the lateral adjustment range of the vehicle, reduce the degree of encroachment on the space of adjacent lanes when the vehicle detouring, thereby ensuring that it can avoid obstacles in front while reserving a certain passage space for vehicles behind and in adjacent lanes. Through the above adjustments, the controller can balance detouring safety and traffic efficiency in medium-risk scenarios.

[0084] When the risk level is greater than or equal to the second risk threshold, it indicates that the traffic flow behind and in adjacent lanes is relatively dense, and the vehicle faces a high risk of lateral collision when detouring. The controller increases the lateral safety weight to the third weight and decreases the maximum lateral offset limit to the third offset limit. The third weight is greater than the second weight to enhance the trajectory optimization model's emphasis on lateral safety margin, ensuring that the optimized trajectory is as far away as possible from the target behind and in adjacent lanes; the third offset limit is less than the second offset limit to significantly tighten the vehicle's lateral adjustment range, minimizing the vehicle's encroachment on adjacent lane space when detouring and prioritizing detouring safety. Optionally, the controller can also perform multi-frame progressive updates of the above weights and offset limits through a parameter smoothing iteration mechanism to avoid trajectory jumps caused by sudden parameter changes in a single frame.

[0085] Through the above-mentioned tiered adjustment method, the controller can dynamically adapt the cost parameters and boundary constraints of the trajectory optimization model to the risk level of the traffic flow behind and in adjacent lanes. When the risk is low, it prioritizes traffic efficiency; when the risk is moderate, it balances safety and efficiency; and when the risk is high, it prioritizes safety. This provides optimization objectives and constraints that match the current risk situation for generating the initial lateral adjustment trajectory.

[0086] In one possible implementation, the cost parameters include lateral safety weights, trajectory smoothness weights, and a maximum lateral offset limit. An example of adjusting the cost parameters and boundary constraints of the trajectory optimization model according to the risk level is as follows: The adjustment amounts of the lateral safety weight, trajectory smoothness weight, and maximum lateral offset upper limit are progressively updated over multiple frames. The adjustment amount of any cost parameter between adjacent frames is not greater than the corresponding preset amplitude threshold, so that the lateral adjustment trajectory planned between adjacent frames remains continuous.

[0087] In some embodiments, when the controller determines the target adjustment values ​​for the lateral safety weight, trajectory smoothness weight, and maximum lateral offset upper limit based on the risk level, it does not directly jump the parameter values ​​of the current frame to the target values. Instead, it performs a multi-frame progressive update of the adjustment amounts of the aforementioned parameters. Specifically, the controller can pre-set corresponding preset amplitude thresholds for the lateral safety weight, trajectory smoothness weight, and maximum lateral offset upper limit. The preset amplitude thresholds are used to limit the allowable changes in parameters between two adjacent frames. These preset amplitude thresholds can be determined based on the parameter type, the total parameter adjustment range, the planning period, or actual vehicle calibration. The preset amplitude thresholds for each parameter can be the same or different, and this application does not impose any limitations on this.

[0088] In each planning frame cycle, the controller calculates the target values ​​corresponding to the lateral safety weight, trajectory smoothness weight, and maximum lateral offset upper limit based on the risk level determined in the current frame, and compares these target values ​​with the parameter values ​​actually used in the current frame. If the difference between the target value and the current value is greater than the preset amplitude threshold corresponding to the parameter, the parameter is not directly adjusted to the target value in this frame, but only adjusted towards the target value according to the preset amplitude threshold, with the remaining adjustment amount left to be gradually approximated in subsequent frames. If the difference between the target value and the current value is less than or equal to the preset amplitude threshold, the parameter can be directly adjusted to the target value in this frame. Thus, the adjustment amount of any parameter between two adjacent frames is not greater than its corresponding preset amplitude threshold, making the optimization objectives and constraints of the trajectory optimization model change smoothly between adjacent frames, without any sudden changes in parameters in a single frame.

[0089] For example, in one possible implementation, the preset threshold for the lateral safety weight can be set to 10% of the total adjustment range, and the preset threshold for the maximum lateral offset can also be set to 10% of the total adjustment range. When the risk level jumps from low to high, the target value of the lateral safety weight may need to be increased from the first weight to the third weight, and the maximum lateral offset may need to be compressed from the first offset limit to the third offset limit. If the jump is direct, the optimization target and boundary constraints between adjacent frames will change drastically, which may lead to discontinuities in the planned lateral adjustment trajectory, resulting in steering wheel vibration or vehicle body sway. With multi-frame progressive updates, the controller can adjust the lateral safety weight and the maximum lateral offset limit step by step over multiple consecutive frames, adjusting only about 10% of the total change in each frame. After several frames, it gradually approaches the target value, thus keeping the lateral adjustment trajectory planned in adjacent frames continuous. Optionally, multi-frame progressive updates can be implemented through linear interpolation, first-order filtering, moving average, or a preset number of transition frames. The controller can also combine risk hysteresis filtering to avoid frequent parameter adjustments caused by small fluctuations in the risk level.

[0090] Through the aforementioned multi-frame progressive update method, the controller can smoothly transition the adjustment process of lateral safety weight, trajectory smoothness weight, and maximum lateral offset limit when the risk level changes, ensuring that the lateral adjustment trajectory planned between adjacent frames remains continuous. This improves the smoothness of the vehicle's lateral movement and ride comfort during detours, and avoids trajectory jumps, steering wheel vibrations, or vehicle body swaying caused by sudden parameter changes.

[0091] 305. The controller performs collision verification on the initial lateral adjustment trajectory based on the predicted driving trajectory envelope region of the target object, and adjusts the initial lateral adjustment trajectory if the collision verification fails, so as to obtain the target lateral adjustment trajectory of the vehicle.

[0092] In one possible implementation, an example of performing collision verification on the initial lateral adjustment trajectory based on the predicted driving trajectory envelope region of the target object is as follows: Based on the motion state information of the target object, the predicted driving trajectory envelope area of ​​the target object within the preset prediction time domain is predicted. The preset prediction time domain covers the time look-ahead range defined by the collision duration threshold corresponding to the risk level. Collision verification is performed on each trajectory point of the initial lateral adjustment trajectory to determine whether the minimum distance between each trajectory point and the predicted driving trajectory envelope area meets the preset safety margin.

[0093] In some embodiments, the controller first predicts the predicted driving trajectory envelope region of the target object within a preset prediction time domain based on the target object's motion state information. The motion state information may include the target object's current position, absolute velocity, relative velocity, acceleration, heading angle, lateral velocity, lane assignment information, and lateral movement trend. The controller can use kinematic models, uniform acceleration models, constant steering models, probabilistic prediction models, or learning models to predict the target object's motion over a future period, obtaining its future driving trajectory. Considering the uncertainty of the prediction, the controller can expand the predicted trajectory by combining the target object's size, prediction error range, or motion uncertainty to generate a predicted driving trajectory envelope region. The predicted driving trajectory envelope region characterizes the spatial range that the target object may occupy within the preset prediction time domain in the future. Its shape can be a strip-shaped region, a fan-shaped region, a rectangular region, or other geometric regions suitable for describing the space occupied by the target object in the future, extending along the predicted trajectory.

[0094] The preset prediction time domain can be determined based on the collision duration threshold corresponding to the risk level, and must at least cover the time look-ahead range defined by that collision duration threshold. For example, if the collision duration threshold corresponding to the risk level is 3 seconds, the preset prediction time domain can be set to 3 seconds or slightly greater than 3 seconds, such as 3 to 4 seconds, to ensure that the time range covered by collision verification matches the time window of interest for risk assessment. If the risk level is low, its corresponding collision duration threshold is high, or the risk urgency is low, the controller can shorten the preset prediction time domain accordingly; if the risk level is high, the controller can extend the preset prediction time domain accordingly, or at least ensure that the preset prediction time domain is not less than the time look-ahead range used to determine collision risk under that risk level. In this way, the controller can adapt the time range of the predicted driving trajectory envelope area to the urgency of lateral collision reflected by the risk level, avoiding the situation where the prediction time domain is too short, resulting in high-risk targets not being fully verified, and also avoiding the situation where the prediction time domain is too long, resulting in unnecessary consumption of computing resources.

[0095] Secondly, the controller performs collision verification on each trajectory point of the initial lateral adjustment trajectory. The initial lateral adjustment trajectory may include a series of discrete trajectory points, each corresponding to the vehicle's position or pose at a future moment. For each trajectory point, the controller calculates the minimum distance between that trajectory point and the target object's driving trajectory envelope area. The minimum distance can be the shortest Euclidean distance from the trajectory point to the boundary of the envelope area, or it can be the shortest distance from the trajectory point to the centerline of the envelope area minus the half-width of the envelope area, etc. The controller determines whether the minimum distance between each trajectory point and the driving trajectory envelope area meets the preset safety margin. The preset safety margin can be determined based on vehicle size, target object size, vehicle speed, risk level, or calibration experience. For example, in one possible implementation, the preset safety margin can be set to 0.3 meters to 0.5 meters; the higher the risk level, the larger the preset safety margin can be to provide more sufficient lateral avoidance space.

[0096] If the minimum distance between all trajectory points on the initial lateral adjustment trajectory and the target object's trajectory envelope is greater than or equal to a preset safety margin, the controller determines that the collision check has passed, and the initial lateral adjustment trajectory can be used as the target lateral adjustment trajectory or continue processing. If the minimum distance between one or more trajectory points and the predicted trajectory envelope is less than the preset safety margin, or if a trajectory point falls within the predicted trajectory envelope, the controller determines that the collision check has failed, and the initial lateral adjustment trajectory needs to be adjusted, such as by reducing the lateral adjustment range, changing the lateral offset direction, or regenerating the initial lateral adjustment trajectory. Through the above point-by-point collision check method, the controller can accurately determine whether there is a lateral conflict between the vehicle's future trajectory and the future motion envelope of the target behind and in adjacent lanes during the obstacle avoidance planning stage, thus providing a basis for trajectory adjustment and safety backup.

[0097] In one possible implementation, an example of adjusting the initial lateral adjustment trajectory when the collision check fails is as follows: If there are trajectory segments that do not meet the preset safety margin, the trajectory segments are removed; the initial lateral adjustment trajectory is regenerated according to the adjusted lateral adjustment range until the regenerated initial lateral adjustment trajectory meets the verification or the lateral adjustment range reaches the minimum adjustment range lower limit; if the risk level is within the preset highest risk level range, the lateral adjustment range is determined as the minimum adjustment range lower limit.

[0098] In some embodiments, during collision verification, if the controller detects that the minimum distance between one or more trajectory points in the initial lateral adjustment trajectory and the target object's driving trajectory envelope is less than a preset safety margin, then the trajectory segments to which these trajectory points that do not meet the preset safety margin belong are identified as trajectory segments to be adjusted, and these trajectory segments to be adjusted are removed from the initial lateral adjustment trajectory. After removing the trajectory segments to be adjusted, the controller may retain other trajectory segments in the initial lateral adjustment trajectory that meet the preset safety margin, or it may re-plan the trajectory starting from the vehicle's current pose and ending at the target it is bypassing.

[0099] After removing the trajectory segments to be adjusted, the controller regenerates an initial lateral adjustment trajectory according to the contracted lateral adjustment range. The contracted lateral adjustment range can be a reduced range relative to the original lateral adjustment range, such as reducing the vehicle's lateral offset towards adjacent lanes or obstacles, so that the regenerated initial lateral adjustment trajectory maintains a greater distance from the predicted driving trajectory envelope of the target object. Based on the contracted lateral adjustment range, obstacle information in front of the vehicle, and the current risk level, the controller regenerates an initial lateral adjustment trajectory and performs collision verification on this regenerated initial lateral adjustment trajectory again based on the predicted driving trajectory envelope of the target object.

[0100] If the regenerated lateral adjustment trajectory still contains trajectory segments that do not meet the preset safety margin, the controller can continue to eliminate the corresponding trajectory segments and further reduce the lateral adjustment amplitude, then regenerate the initial lateral adjustment trajectory and perform collision verification. The above process of elimination, reduction, regeneration, and verification can be repeated until the regenerated initial lateral adjustment trajectory meets the collision verification conditions, or the lateral adjustment amplitude reaches the preset minimum adjustment amplitude limit. The minimum adjustment amplitude limit can be determined based on the minimum lateral offset required for the vehicle to safely bypass the obstacle in front, or it can be set to a value of zero offset or near-zero offset based on calibration experience; this application does not impose any limitations on this.

[0101] When the risk level is within the preset highest risk range, indicating that traffic from behind and adjacent lanes poses an extremely high risk of lateral collision to the vehicle's detour, the controller can directly set the lateral adjustment range as the minimum lower limit, without attempting a larger lateral adjustment range. In this case, the controller generates the target lateral adjustment trajectory using the minimum lower limit as a constraint and prohibits the vehicle from making lateral adjustments in a direction deviating from the centerline of the current lane, in order to minimize encroachment on adjacent lane space and prioritize detour safety.

[0102] In this way, when the collision check fails, the controller can gradually obtain the target lateral adjustment trajectory that meets the safety margin requirements or reaches the minimum adjustment range by eliminating conflict trajectory segments, reducing the lateral adjustment range, and performing repeated checks.

[0103] 306. The controller controls the vehicle to perform a bypass maneuver according to the target's lateral adjustment trajectory.

[0104] The controller can output steering control commands based on the target's lateral adjustment trajectory and output longitudinal control commands based on the risk level, driving the vehicle's steering, drive, and braking systems to work together to bypass obstacles in front, while avoiding lateral collision risks from vehicles behind and in adjacent lanes, ensuring the safety and smoothness of the entire detour process.

[0105] Alternatively, this application also provides the following embodiments.

[0106] In one possible implementation, if the collision check fails, a tiered fallback strategy that links lateral and longitudinal control is executed based on the risk level range in which the risk level is located.

[0107] In some embodiments, after the controller performs collision verification on the initial lateral adjustment trajectory within the predicted driving trajectory envelope region of the target object, if it determines that the collision verification fails, it indicates that performing the detour according to the current initial lateral adjustment trajectory will cause a lateral conflict with the future movement of the target objects behind and in adjacent lanes. In this case, the controller does not directly abandon the detour or only adjust the lateral trajectory, but instead executes a hierarchical fallback strategy that links lateral control and longitudinal control according to the current risk level range.

[0108] Specifically, the controller can pre-divide multiple risk level intervals and configure corresponding lateral and longitudinal control strategies for different risk level intervals. A higher priority risk level interval indicates a greater risk of lateral collision from rear and adjacent lane traffic, and the controller's fallback strategy prioritizes safety. Conversely, a lower priority risk level interval allows the controller to balance detour efficiency and ride comfort while ensuring safety. Lateral control strategies can include at least one of the following: continuing to detour, reducing the lateral adjustment range, terminating outward deflection, and controlling the vehicle to return to center. Longitudinal control strategies can include at least one of the following: maintaining speed, gradual deceleration, and uniform deceleration. The lateral and longitudinal control strategies are determined based on the same risk level interval, rather than being executed independently.

[0109] For example, when the risk level is in the lower priority range, the controller can continue to perform detours within the contracted lateral adjustment range and simultaneously implement progressive longitudinal deceleration. This increases the longitudinal clearance and / or collision duration margin between the vehicle and targets in the rear and adjacent lanes by reducing vehicle speed, ensuring that the contracted detour trajectory meets the safety margin requirements of the collision verification. When the risk level is in the higher priority range, the controller can terminate the lateral adjustment in the outward direction, guide the vehicle back to the center line of the current lane, and simultaneously implement longitudinal deceleration to quickly escape the lateral conflict zone with targets in the rear and adjacent lanes. When the risk level is in the preset highest risk range, the controller can use the minimum lateral adjustment range determined in the collision verification as a constraint to prevent the vehicle from continuing to deviate into the adjacent lane, and guide the vehicle to smoothly return to the center line while simultaneously implementing longitudinal deceleration until the risk level decreases or the vehicle speed drops to a safe range.

[0110] Through the aforementioned tiered fallback strategy, the controller can dynamically coordinate lateral and longitudinal control according to the degree of risk when the collision check fails, so that the two are no longer executed separately. This avoids the risk of lateral collisions while taking into account the smoothness and traffic efficiency of the detour process.

[0111] In one possible implementation, when the risk level is within the first risk level range, lateral detour continues within the contracted lateral adjustment range, and longitudinal gradual deceleration is performed in conjunction. The longitudinal gradual deceleration is configured to increase the longitudinal distance between the vehicle and the target object and / or the collision duration margin, so that the contracted detour trajectory meets the safety margin of the collision verification, and the longitudinal deceleration does not exceed the preset comfort deceleration upper limit. When the risk level is within the preset highest risk level range, the minimum adjustment range determined in the collision verification is used as a constraint to control the vehicle to return to the center line of the current driving lane. During the return to center, adjustment to the side away from the center line is prohibited, and uniform deceleration is performed in conjunction until the risk level exits the highest risk level range or the vehicle speed is reduced to the preset safe speed.

[0112] Specifically, for multiple pre-divided risk level intervals, the controller can configure corresponding lateral control actions and longitudinal control actions for different risk level intervals.

[0113] When the risk level is in the first risk level range, it indicates that the traffic flow behind and in adjacent lanes poses a certain degree of risk to the vehicle's detour, but has not yet reached the highest risk level that requires abandoning the detour entirely. At this point, the controller continues to perform lateral detour within the contracted lateral adjustment range; that is, the vehicle still moves towards the target detour direction along the contracted lateral adjustment trajectory, but the lateral offset is limited to a small range to reduce encroachment on adjacent lane space. Simultaneously, the controller performs longitudinal gradual deceleration. Longitudinal gradual deceleration means that the controller gradually reduces the vehicle speed, rather than suddenly decelerating. By reducing the vehicle speed, the longitudinal distance between the vehicle and the target objects behind and in adjacent lanes can be increased, while simultaneously increasing the collision time margin between the vehicle and the target objects. This allows the contracted detour trajectory to obtain a greater safety margin in the time dimension, meeting the safety margin required for collision verification. The controller limits the deceleration during the longitudinal gradual deceleration process to ensure it does not exceed the preset comfort deceleration upper limit. The preset comfort deceleration upper limit can be calibrated according to ride comfort requirements, for example, set to 1.5 m / s², but this application does not limit it in this way. Through the synergy of the aforementioned lateral contraction and longitudinal gradual deceleration, the controller can complete detours or reduce the risk of conflict in a relatively gentle manner within the first risk level range, avoiding driving discomfort caused by sudden deceleration or sharp turns.

[0114] When the risk level is within the preset highest risk range, it indicates that the traffic flow behind and in adjacent lanes poses an extremely high risk of lateral collision to the vehicle's maneuvering. Continuing to maneuver outwards will likely result in a lateral scrape or collision. At this point, the controller, constrained by the minimum adjustment range determined during collision verification, controls the vehicle to return to the center line of the current lane. The minimum adjustment range is the minimum lateral offset determined during the aforementioned collision verification and trajectory adjustment process that allows the vehicle to avoid the predicted trajectory envelope area of ​​the target behind as much as possible. The controller controls the vehicle's steering wheel to turn smoothly towards the lane center line, gradually bringing the vehicle back to the vicinity of the lane center line. During the return-to-center process, the controller prohibits the vehicle from adjusting to the side away from the center line, i.e., prohibits any further lateral movement outwards, to ensure that the vehicle does not further encroach on the space of adjacent lanes. Simultaneously, the controller performs uniform deceleration, i.e., controls the vehicle to reduce its speed at a relatively constant deceleration, allowing the vehicle to quickly leave the lateral conflict area with the target behind and in adjacent lanes. The above return-to-center and uniform deceleration are continuously executed until the risk level exits the highest risk range, or the vehicle speed is reduced to a preset safe speed. The preset safe speed can be determined based on road conditions, traffic flow, or calibration experience. For example, it can be set to a speed lower than the current speed or a fixed low speed. In this way, the controller can promptly terminate the detour and smoothly return to the correct position in the highest risk range, while further reducing the risk of collision through longitudinal deceleration, thus achieving a safety backup.

[0115] Figure 4 This is a schematic diagram of the structure of a vehicle detour control device provided in an embodiment of this application. See also... Figure 4 The device includes: The determination module 401 is used to determine the risk level of a vehicle based on the motion state information of the target object in the area of ​​interest behind the vehicle and in the adjacent lane when the vehicle is going around an obstacle. The risk level is used to indicate the magnitude of the lateral collision risk posed by the traffic flow behind and in the adjacent lane to the vehicle's detour. The generation module 402 is used to generate the initial lateral adjustment trajectory of the vehicle based on the risk level and information about obstacles in front of the vehicle. Processing module 403 is used to perform collision verification on the initial lateral adjustment trajectory based on the predicted driving trajectory envelope region of the target object, and adjust the initial lateral adjustment trajectory when the collision verification fails, so as to obtain the target lateral adjustment trajectory of the vehicle. The control module 404 is used to control the vehicle to perform a bypass operation according to the target lateral adjustment trajectory.

[0116] It should be noted that the vehicle detour control device provided in the above embodiments is only illustrated by the division of the above functional modules when rendering a 3D vehicle model. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above. In addition, the vehicle detour control device and the vehicle detour control method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0117] This application also provides a vehicle. Figure 5 This is a schematic diagram of the structure of a vehicle provided in an embodiment of this application.

[0118] Typically, vehicle 500 includes one or more processors 501 and one or more memories 502.

[0119] Processor 501 may include one or more processing cores, such as a quad-core processor, a hexa-core processor, etc. Processor 501 may be implemented using at least one hardware form selected from DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). Processor 501 may also include a main processor and a coprocessor. The main processor, also known as a CPU (Central Processing Unit), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, processor 501 may integrate a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the screen. In some embodiments, processor 501 may also include an AI (Artificial Intelligence) processor, which is used to handle computational operations related to machine learning.

[0120] The memory 502 may include one or more computer-readable storage media, which may be non-transitory. The memory 502 may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices or flash memory devices. In some embodiments, the non-transitory computer-readable storage media in the memory 502 are used to store at least one computer program, which is executed by the processor 501 to implement the vehicle detour control method provided in the method embodiments of this application.

[0121] Those skilled in the art will understand that Figure 5 The structure shown does not constitute a limitation on vehicle 500 and may include more or fewer components than shown, or combine certain components, or use different component arrangements.

[0122] In addition, the device provided in the embodiments of this application may specifically be a chip, component or module. The chip may include a connected processor and a memory. The memory is used to store instructions. When the processor calls and executes the instructions, the chip can execute a vehicle detour control method provided in the above embodiments.

[0123] This embodiment also provides a computer-readable storage medium storing computer program code. When the computer program code is run on a computer, the computer executes the above-described related method steps to implement the vehicle detour control method provided in the above embodiment.

[0124] This embodiment also provides a computer program product that, when run on a computer, causes the computer to perform the aforementioned steps to implement a vehicle detour control method provided in the above embodiment.

[0125] In this embodiment, the device, computer-readable storage medium, computer program product, or chip are all used to execute the corresponding methods provided above. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods provided above, and will not be repeated here.

[0126] Through the above description of the embodiments, those skilled in the art will understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0127] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another apparatus, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0128] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A vehicle detour control method, characterized in that, The method includes: When the vehicle is bypassing an obstacle, the risk level of the vehicle is determined based on the motion state information of the target object in the area of ​​interest behind the vehicle and in the adjacent lane. The risk level is used to indicate the magnitude of the lateral collision risk posed by the traffic flow behind and in the adjacent lane to the vehicle's bypassing. Based on the risk level and the obstacle information in front of the vehicle, an initial lateral adjustment trajectory for the vehicle is generated; Based on the predicted driving trajectory envelope region of the target object, the initial lateral adjustment trajectory is collided, and if the collision verification fails, the initial lateral adjustment trajectory is adjusted to obtain the target lateral adjustment trajectory of the vehicle. The vehicle is controlled to perform a lateral adjustment trajectory to the target and perform a bypass maneuver.

2. The vehicle detour control method according to claim 1, characterized in that, The determination of the risk level of the vehicle based on the motion state information of target objects within the area of ​​interest behind and adjacent to the vehicle includes: The target objects corresponding to the motion state information are subjected to validity screening to obtain valid target objects for participating in the risk level assessment. Target objects that fail the validity screening are not included in the risk level assessment. Based on the effective target object, risk characterization parameters are extracted from both the group and individual dimensions. The risk characterization parameters at the group level are used to characterize the overall density of traffic flow within the area of ​​interest, while the risk characterization parameters at the individual level are used to characterize the urgency of a side collision between the effective target object and a vehicle. The risk characterization parameters of the group dimension and the risk characterization parameters of the individual dimension are classified and quantified according to preset classification conditions to obtain environmental risk indicators and target risk indicators. The environmental risk indicators are fused with the target risk indicators to generate the risk level of the vehicle.

3. The vehicle detour control method according to claim 2, characterized in that, The step of extracting risk characterization parameters from both group and individual dimensions based on the effective target objects includes: Count the number of dynamic motor vehicles among the valid target objects within the area of ​​interest; For the vehicles behind the effective target objects, calculate the lateral collision duration between the vehicles behind, wherein the lateral collision duration is the ratio of the relative lateral distance to the relative lateral velocity, and the lateral collision duration is calculated when the relative lateral velocities indicate that the vehicles are approaching each other.

4. The vehicle detour control method according to claim 3, characterized in that, The risk characterization parameters of the group dimension and the risk characterization parameters of the individual dimension are classified and quantified according to preset classification conditions to obtain environmental-level risk indicators and target-level risk indicators, including: Based on the number of dynamic motor vehicles, the preset density, and the first preset classification condition, the environmental risk index is determined. When the number is less than a preset number threshold, it is determined as a first traffic flow condition. When the number is greater than or equal to the preset number threshold but less than the preset density, it is determined as a second traffic flow condition. When the number is greater than or equal to the preset density, it is determined as a third traffic flow condition. The traffic density corresponding to the first traffic flow condition, the second traffic flow condition, and the third traffic flow condition increases sequentially. Based on the lateral collision duration, the preset collision duration threshold, and the second preset classification condition, the target-level risk index is determined. The lateral collision duration is determined to be a first collision risk when it is greater than the preset collision duration threshold, and the lateral collision duration is determined to be a second collision risk when it is less than or equal to the preset collision duration threshold. The urgency of the second collision risk is higher than that of the first collision risk.

5. The vehicle detour control method according to claim 2, characterized in that, The step of fusing the environmental-level risk indicators with the target-level risk indicators to generate the risk level of the vehicle includes: When the target-level risk indicator corresponds to multiple valid target objects, the risk characterization parameter with the highest risk level is determined from the risk characterization parameters of the multiple valid target objects and used as the target-level representative value. Based on the environmental-level risk indicators and the target-level representative values, a risk level is generated by fusing them according to a preset risk mapping relationship. The risk mapping relationship is configured such that the risk level obtained after fusion increases as the risk represented by the target-level representative value increases, and the risk level obtained after fusion is not lower than the risk level determined by the target-level representative value.

6. The vehicle detour control method according to claim 1, characterized in that, The step of generating the initial lateral adjustment trajectory of the vehicle based on the risk level and the obstacle information in front of the vehicle includes: The lateral adjustment trajectory of the vehicle is optimized using a trajectory optimization model to generate a lateral adjustment trajectory to be determined. The cost parameters and boundary constraints of the trajectory optimization model are adjusted according to the risk level. The initial lateral adjustment trajectory is determined based on the adjusted cost parameters, the boundary constraints, and the obstacle information in front of the vehicle.

7. The vehicle detour control method according to claim 6, characterized in that, The cost parameters include lateral safety weight, trajectory smoothness weight, and maximum lateral offset limit; The step of adjusting the cost parameters and boundary constraints of the trajectory optimization model according to the risk level includes: If the risk level is lower than the first risk threshold, the lateral safety weight is reduced to the first weight, and the maximum lateral offset limit is increased to the first offset limit. When the risk level is greater than or equal to the first risk threshold and less than the second risk threshold, the lateral safety weight is maintained at the second weight, and the maximum lateral offset upper limit is set to the second offset upper limit, wherein the second weight is not less than the first weight and the second offset upper limit is less than the first offset upper limit; If the risk level is greater than or equal to the second risk threshold, the lateral safety weight is increased to the third weight, and the maximum lateral offset limit is reduced to the third offset limit. The third weight is greater than the second weight, and the third offset limit is less than the second offset limit.

8. The vehicle detour control method according to claim 1, characterized in that, The collision verification of the initial lateral adjustment trajectory based on the predicted driving trajectory envelope region of the target object includes: Based on the motion state information of the target object, the predicted driving trajectory envelope region of the target object within a preset prediction time domain is predicted, and the preset prediction time domain covers the time look-ahead range defined by the collision duration threshold corresponding to the risk level. Collision verification is performed on each trajectory point of the initial lateral adjustment trajectory to determine whether the minimum distance between each trajectory point and the envelope area of ​​the predicted driving trajectory meets the preset safety margin.

9. The vehicle detour control method according to claim 8, characterized in that, The step of adjusting the initial lateral adjustment trajectory when the collision check fails includes: If there are trajectory segments that do not meet the preset safety margin, the trajectory segments are eliminated. The initial lateral adjustment trajectory is regenerated according to the adjusted lateral adjustment range until the regenerated initial lateral adjustment trajectory meets the verification or the lateral adjustment range reaches the minimum adjustment range lower limit. When the risk level is within the preset maximum risk level range, the horizontal adjustment range is determined as the lower limit of the minimum adjustment range.

10. A vehicle, characterized in that, The vehicles include: Memory, used to store executable program code; A processor is configured to call and run the executable program code from the memory, causing the vehicle to perform the vehicle detour control method as described in any one of claims 1 to 9.