Vehicle control methods, vehicles, and computer-readable storage media

By acquiring vehicle motion information and obstacle spatiotemporal boundary information for accurate motion prediction, the system can identify obstacle scenarios that can be avoided by accelerating, thus solving the problem of high false trigger rate of automatic emergency braking system and achieving higher driving comfort and system reliability.

CN123104637APending Publication Date: 2026-11-17CHERY AUTOMOBILE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611063416.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-16
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing automatic emergency braking systems have a high rate of false triggering when facing obstacles, and cannot accurately distinguish between scenarios where acceleration is possible to avoid them, which affects the driving experience and increases the risk of rear-end collisions.

Method used

By acquiring vehicle motion information, dynamic constraint parameters, and spatiotemporal boundary information of candidate objects, motion prediction is performed, feasibility is identified, target objects are screened, and braking or warning control commands are generated, accurately identifying scenarios that can be avoided by accelerating.

Benefits of technology

It significantly reduces the false triggering rate of the emergency braking system, improves driving comfort and system reliability, and ensures that effective braking or warning actions are performed when necessary.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123104637A_ABST
    Figure CN123104637A_ABST
Patent Text Reader

Abstract

This application provides a vehicle control method, a vehicle, and a computer-readable storage medium. The method includes: acquiring the vehicle's motion information, dynamic constraint parameters, and spatiotemporal boundary information of multiple candidate objects, wherein a predicted intersection point exists between the vehicle's trajectory and the candidate object's trajectory; performing motion prediction based on the vehicle's motion information, dynamic constraint parameters, and spatiotemporal boundary information to obtain a prediction result; determining a target object from the multiple candidate objects based on the prediction result; and generating a control command for the current vehicle based on the relative motion information between the target object and the current vehicle, wherein the control command instructs the current vehicle to perform braking and / or warning actions. This application solves the technical problems of high false triggering rate and low control strategy accuracy in existing automatic emergency braking systems.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of vehicle technology, and more specifically, to a vehicle control method, a vehicle, and a computer-readable storage medium. Background Technology

[0002] Automatic emergency braking (AEP) detects obstacles ahead and triggers braking to ensure driving safety. Current technologies, when dealing with obstacles, often rely on simple kinematic models to calculate collision time or required deceleration, only considering slowing down to avoid them. However, the trajectory of crossing obstacles is complex, and vehicles can often pass the intersection by accelerating appropriately. Current technologies cannot distinguish between scenarios where acceleration is possible to avoid obstacles, leading to false triggering of AEP, which not only affects the driving experience but also increases the risk of rear-end collisions.

[0003] There is currently no good solution to the above problems. Summary of the Invention

[0004] This application provides a vehicle control method, a vehicle, and a computer-readable storage medium to at least solve the technical problems of high false triggering rate and low control strategy accuracy in the prior art of automatic emergency braking systems.

[0005] According to one aspect of the embodiments of this application, a vehicle control method is provided, comprising: acquiring the vehicle motion information, dynamic constraint parameters, and spatiotemporal boundary information of a plurality of candidate objects of the current vehicle, wherein there is a predicted intersection point between the vehicle motion trajectory of the current vehicle and the object motion trajectory of the candidate objects; performing motion prediction based on the vehicle motion information, dynamic constraint parameters, and spatiotemporal boundary information to obtain a prediction result, wherein the prediction result is used to characterize the feasibility of the current vehicle passing the predicted intersection point before the candidate objects by accelerating; determining a target object from the plurality of candidate objects according to the prediction result; and generating a control command for the current vehicle based on the relative motion information between the target object and the current vehicle, wherein the control command is used to instruct the current vehicle to perform braking and / or warning actions.

[0006] Furthermore, the vehicle motion information includes the vehicle speed; the dynamic constraint parameters include the current vehicle's response delay time and longitudinal acceleration upper limit; the spatiotemporal boundary information includes the upper boundary point set, which is used to characterize the spatiotemporal occupancy of the candidate object's leading edge boundary at the prediction intersection.

[0007] Furthermore, motion prediction is performed based on vehicle motion information, dynamic constraint parameters, and spatiotemporal boundary information. The prediction results include: traversing the set of upper boundary points, calculating the predicted longitudinal position of the current vehicle based on the target timestamp, vehicle motion information, and dynamic constraint parameters corresponding to the current upper boundary point during the traversal; and determining the prediction result based on the predicted longitudinal position corresponding to each upper boundary point in the set of upper boundary points and the longitudinal position of the boundary point.

[0008] Furthermore, the calculation of the predicted longitudinal position based on the target timestamp, vehicle motion information, and dynamic constraint parameters includes: if the interval between the current timestamp and the target timestamp is less than or equal to the response delay time, the predicted longitudinal position is calculated based on the vehicle speed and the interval time; if the interval time is greater than the response delay time, the predicted longitudinal position is calculated based on the vehicle speed, the response delay time, and the upper limit of longitudinal acceleration.

[0009] Furthermore, based on the predicted longitudinal position corresponding to each upper boundary point in the upper boundary point set and the longitudinal position of the boundary point, the prediction result is determined as follows: if for all upper boundary points in the upper boundary point set, the predicted longitudinal position is greater than the longitudinal position of the boundary point, then the prediction result is determined to be feasible; if for at least one boundary point in the upper boundary point set, the predicted longitudinal position is less than or equal to the longitudinal position of the boundary point, then the prediction result is determined to be infeasible.

[0010] Furthermore, based on the prediction results, determining the target object from multiple candidate objects includes: filtering candidate objects whose prediction results are infeasible from multiple candidate objects to obtain filtering results; and selecting the candidate object with the smallest predicted collision time with the current vehicle from the filtering results as the target object.

[0011] Furthermore, the relative motion information includes the predicted collision time. Based on the relative motion information between the target object and the current vehicle, the control command for the current vehicle is generated as follows: if the predicted collision time is less than a first threshold, an emergency braking control command is generated; if the predicted collision time is greater than or equal to the first threshold and less than a second threshold, a warning control command is generated.

[0012] Furthermore, the method also includes: for candidate objects whose prediction results are feasible, calculating the longitudinal safety distance of the current vehicle passing through the predicted intersection based on the unexpected acceleration of the candidate object; if the longitudinal safety distance is less than a preset distance threshold, adjusting the prediction result corresponding to the candidate object to infeasible.

[0013] According to another aspect of the embodiments of this application, a vehicle control device is also provided, comprising: an acquisition module, configured to acquire the vehicle motion information, dynamic constraint parameters, and spatiotemporal boundary information of a plurality of candidate objects of the current vehicle, wherein there is a predicted intersection point between the vehicle motion trajectory of the current vehicle and the object motion trajectory of the candidate objects; a prediction module, configured to perform motion prediction based on the vehicle motion information, dynamic constraint parameters, and spatiotemporal boundary information to obtain a prediction result, wherein the prediction result is used to characterize the feasibility of the current vehicle passing the predicted intersection point before the candidate objects by accelerating; a determination module, configured to determine a target object from the plurality of candidate objects according to the prediction result; and a generation module, configured to generate control commands for the current vehicle based on the relative motion information between the target object and the current vehicle, wherein the control commands are used to instruct the current vehicle to perform braking actions and / or warning actions.

[0014] According to another aspect of the embodiments of this application, a vehicle is also provided, including: a memory storing an executable program; and a processor for running the program, wherein the program executes the methods in various embodiments of this application when it runs.

[0015] According to another aspect of the embodiments of this application, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the computer-readable storage medium is located to perform the methods of various embodiments of this application.

[0016] In this embodiment, a vehicle control approach is adopted. This involves acquiring the vehicle's motion information, dynamic constraint parameters, and spatiotemporal boundary information of multiple candidate objects. A predicted intersection point exists between the vehicle's trajectory and the candidate object's trajectory. Motion prediction is performed based on the vehicle's motion information, dynamic constraint parameters, and spatiotemporal boundary information to obtain a prediction result. This prediction result characterizes the feasibility of the vehicle accelerating past the candidate object to pass the predicted intersection point. Based on the prediction result, a target object is determined from the multiple candidate objects. Control commands are generated for the vehicle based on the relative motion information between the target object and the vehicle. These commands instruct the vehicle to perform braking and / or warning actions. This achieves the goal of accurately identifying scenarios where the vehicle can accelerate to avoid crossing obstacles, thus significantly reducing the false trigger rate of the emergency braking system, improving driving comfort and system reliability. This solves the technical problems of high false trigger rate and low control strategy accuracy in existing automatic emergency braking systems. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0018] Figure 1 This is a schematic diagram of an optional vehicle control method according to an embodiment of this application;

[0019] Figure 2 This is a schematic flowchart of an optional vehicle control method according to an embodiment of this application;

[0020] Figure 3 This is a flowchart illustrating another optional vehicle control method according to an embodiment of this application;

[0021] Figure 4 This is a schematic diagram of an optional vehicle control device according to an embodiment of this application;

[0022] Figure 5 This is a structural block diagram of an optional vehicle control device according to an embodiment of this application. Detailed Implementation

[0023] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0024] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0025] With the rapid development of intelligent driving technology, Automatic Emergency Braking (AEB) has become a crucial technology for ensuring driving safety. AEB systems detect obstacles ahead using sensors (such as millimeter-wave radar, cameras, or other onboard sensors). When a collision risk is detected, the system automatically triggers braking to avoid or mitigate the collision. In real-world road scenarios, besides the vehicle in front in the same lane, there are numerous obstacles crossing the road, such as pedestrians, bicycles, and other vehicles crossing from the side. The trajectories of these obstacles intersect with the vehicle's path, but not all obstacles will collide with the vehicle. For example, when crossing an obstacle at a high speed, the vehicle may have already passed the intersection before reaching the intersection; or when the vehicle is traveling at a low speed, it can accelerate to pass the intersection before the obstacle.

[0026] Existing technologies have significant problems in handling obstacles crossing the road. Traditional collision assessment methods are typically based on simple kinematic prediction trajectories, predicting collision risk by calculating the kinematic models of the obstacle and the vehicle. While simple to implement, this method often generates numerous false triggers when dealing with obstacles crossing the road. For example, when a pedestrian quickly crosses the road from the right, although simple kinematic predictions indicate a collision risk, the pedestrian may actually cross before the vehicle arrives. In this case, triggering emergency braking is unnecessary, affecting the driving experience and potentially causing secondary accidents such as rear-end collisions. Currently, the mainstream existing technologies mainly include three methods: The first is based on the predicted time-to-collision (TTC) threshold, which is the most commonly used method in current AEB systems. This method is based on a simple kinematic model, assuming that both the obstacle and the vehicle are moving at a constant linear speed at the current speed, calculating the collision time between the vehicle and the obstacle, and triggering emergency braking if the TTC is less than a certain threshold. The problem with this method is that the TTC calculation usually assumes that the vehicle maintains its current speed or decelerates, without considering the possibility that the vehicle can accelerate moderately to cross the intersection before the obstacle, leading to false triggers. The second method is based on braking deceleration demand. This method also relies on simple kinematic prediction to calculate the braking deceleration required to avoid a collision. If the deceleration demand exceeds a certain threshold, emergency braking is triggered. While this method considers braking capacity, it also assumes the vehicle can only brake and does not consider the possibility of avoiding a collision by accelerating. Furthermore, the predicted trajectory is overly simplistic. The third method is collision detection based on obstacle trajectory prediction. This method predicts the future position of the obstacle using a simple kinematic model and then determines whether the vehicle's trajectory intersects with the predicted obstacle trajectory. While this method considers the obstacle's motion, the prediction model is overly simplistic. It typically assumes the obstacle moves in a uniform straight line and does not fully utilize the upper and lower boundary information of the obstacle in the space-time coordinate system (ST). It also cannot accurately determine whether the vehicle can avoid a collision by accelerating.

[0027] The shortcomings and deficiencies of existing technologies are mainly reflected in three aspects. First, existing AEB technologies generally only use simple kinematic trajectory prediction, assuming that both the obstacle and the vehicle are moving at a constant linear speed at the current speed. Although this prediction method is simple to implement, it is overly simplistic and does not consider the complexity of the obstacle's actual trajectory. When crossing an obstacle, the trajectory is often not a simple constant linear motion, and simple kinematic prediction cannot accurately reflect the actual motion state of the obstacle, leading to inaccurate collision judgments and a high rate of false triggers. Second, existing technologies do not consider vehicle delays and acceleration capabilities. When judging collision risk, they usually assume that the vehicle maintains its current speed or can only decelerate, without considering system response delays and the vehicle's acceleration capabilities. In actual systems, there is a certain delay between detecting an obstacle and executing braking, which affects the accuracy of the judgment. More importantly, when crossing an obstacle, the vehicle can often avoid a collision by accelerating moderately to pass the intersection before the obstacle, but existing methods do not consider the possibility of such acceleration to avoid collision. In actual testing, when crossing obstacles at a slow speed or over a long distance, the vehicle can safely pass the intersection by moderate acceleration without triggering emergency braking. However, existing methods cannot identify this situation, leading to false triggering. Finally, existing technology cannot distinguish between avoidable and unavoidable scenarios. When crossing obstacles, if the vehicle can accelerate to pass the intersection before the obstacle, emergency braking is unnecessary; however, if it cannot avoid the obstacle by accelerating, emergency braking is required. Existing methods cannot make this distinction, resulting in the application of the same handling strategy to all potential collision scenarios, failing to fully utilize the vehicle's avoidance capabilities.

[0028] According to an embodiment of this application, a method embodiment for vehicle control is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0029] This embodiment provides a vehicle control method. Figure 1 This is a flowchart of method A according to an embodiment of this application, as follows: Figure 1 As shown, the process includes the following steps S101 to S104.

[0030] Step S101: Obtain the current vehicle's motion information, dynamic constraint parameters, and spatiotemporal boundary information of multiple candidate objects, wherein there is a predicted intersection point between the current vehicle's motion trajectory and the object motion trajectories of the candidate objects.

[0031] In this embodiment, the vehicle control system acquires the current vehicle's motion information, dynamic constraint parameters, and spatiotemporal boundary information of multiple candidate objects. There is a predicted intersection point between the current vehicle's motion trajectory and the object motion trajectory of the candidate objects.

[0032] Vehicle motion information refers to physical quantities that reflect the real-time motion state of the vehicle, mainly including data such as longitudinal speed, lateral speed, yaw rate, steering angle, and positioning attitude. For example, vehicle motion information could specifically be a current vehicle speed of 40 km / h and a yaw rate of 0.1 rad / s.

[0033] Dynamic constraints are system boundary conditions that limit a vehicle's motion capabilities. They include system response delay time (the time interval between detecting an obstacle and the system executing an action) and longitudinal acceleration limit (the maximum allowable acceleration for the vehicle). Specifically, dynamic constraints can be defined as a system response delay of 0.3 seconds and a longitudinal acceleration limit of 0.3 m / s².

[0034] Candidate objects refer to surrounding obstacles or traffic participants within the current field of vision whose predicted trajectories intersect with the vehicle's trajectory and are included in the vehicle control system's subsequent collision risk assessment or target selection process. For example, pedestrians crossing the road from the side, cyclists, vehicles suddenly changing lanes or crossing, goods that have fallen from a stationary position in the middle of the road, or even other vehicles waiting at an intersection but potentially starting to move, are all considered candidate objects as long as their predicted trajectories intersect with the vehicle's trajectory.

[0035] Spatiotemporal boundary information refers to the occupancy description of candidate objects in a space-time coordinate system. It is typically composed of a set of upper boundary points representing the leading edge of the obstacle and a set of lower boundary points representing the trailing edge of the obstacle. It is used to characterize the longitudinal position range of the obstacle along the vehicle's driving direction at different prediction times. For example, at prediction times of 0.5 seconds, 1.0 seconds, and 1.5 seconds, the longitudinal occupancy range of a pedestrian relative to the vehicle can be represented as [10m, 12m], [15m, 17m], etc., respectively. The acquisition of this spatiotemporal boundary information depends on the obstacle category, position, speed, heading angle, bounding box size, and confidence level data output by onboard perception sensors (such as cameras, millimeter-wave radar, lidar, etc.). At the same time, it is necessary to simultaneously acquire the vehicle's current speed, yaw rate, steering angle, positioning attitude information, and extrinsic parameter data of each sensor relative to the vehicle's coordinate system in order to perform unified coordinate system transformation and projection calculation.

[0036] A predicted intersection point refers to a geometric area in space where the vehicle and a candidate object may overlap or be close to each other, based on their current motion and trajectory. Specifically, the predicted intersection point can be the center position of the intersection 20 meters away from the vehicle, where the vehicle's straight-ahead trajectory and the pedestrian's crossing trajectory are located.

[0037] As can be seen, the vehicle control system performs data initialization and key parameter acquisition, namely, acquiring real-time motion state data of the vehicle, preset vehicle dynamics limit parameters, and trajectory prediction data of multiple potential hazard candidates through onboard sensors and the vehicle control system. Based on this, the vehicle control system analyzes the relative relationship between the vehicle's trajectory and the candidate object's trajectory to identify and confirm the existence of predicted intersection points in their motion paths, thereby providing the necessary data foundation for subsequent feasibility assessments of these specific intersection points.

[0038] For example, assuming a vehicle is traveling at an intersection, the vehicle control system first reads its own speed as 8 m / s, the preset maximum allowable acceleration as 0.25 m / s², and a delay of 0.3 s. Simultaneously, the perception system detects a bicycle crossing from the right, whose predicted trajectory indicates it will reach the center of the intersection in 1.2 seconds. The vehicle control system calculates that it will take the vehicle 1.5 seconds to reach the center of the intersection; since there is a temporal overlap between the two at the center, the existence of the predicted intersection point is confirmed, and the spatiotemporal boundary data of the bicycle near that intersection point is recorded.

[0039] Therefore, by standardizing the acquisition of multi-source heterogeneous perception and vehicle status data and confirming trajectory intersections, a unified and necessary data foundation is provided for subsequent accurate collision risk assessment.

[0040] Step S102: Motion prediction is performed based on the vehicle's motion information, dynamic constraint parameters, and spatiotemporal boundary information to obtain prediction results. The prediction results are used to characterize the feasibility of the current vehicle passing the predicted intersection point before the candidate object after acceleration.

[0041] In this embodiment, the vehicle control system, based on the vehicle's own motion state, physical limitations, and spatiotemporal occupancy data of obstacles, uses a mathematical model to deduce the feasibility of the vehicle accelerating through an intersection. Specifically, the vehicle control system traverses the upper boundary points of obstacles, and for each time point, combines the vehicle's current speed, the acceleration start lag caused by the vehicle control system's response delay, and the maximum allowable acceleration to calculate the theoretically achievable longitudinal position of the vehicle at that time point. This position is then compared with the longitudinal position of the obstacle at the same moment, thereby outputting a feasibility result characterizing whether the vehicle can pass through ahead of schedule.

[0042] Vehicle motion information refers to parameters that reflect the current motion state of the vehicle, such as longitudinal speed. For example, vehicle motion information could specifically be a vehicle speed of 40 km / h.

[0043] Dynamic constraints are conditions that limit a vehicle's motion capabilities, including system response delay time (the time interval from perception to execution) and longitudinal acceleration limit (the maximum allowable acceleration). Examples of dynamic constraints include a response delay of 0.2 seconds and a maximum acceleration of 0.3 m / s².

[0044] Spatiotemporal boundary information refers to the occupancy of candidate objects in a time-space coordinate system. Specifically, it is represented by the longitudinal position interval corresponding to different prediction times, usually consisting of the leading edge point (upper boundary point) representing the head of the obstacle and the trailing edge point (lower boundary point) representing the tail. For example, at prediction time t=1.0s, the longitudinal position interval of the obstacle is [20m, 22m], where 22m is the upper boundary point.

[0045] It can be seen that the vehicle control system uses the acquired vehicle speed, vehicle control system delay time, and maximum acceleration limit, combined with the longitudinal position data of the candidate object at different prediction times (especially the upper boundary point), to calculate the predicted longitudinal position that the vehicle can achieve after considering the delay and accelerating. Then, the calculated predicted longitudinal position of the vehicle is compared with the longitudinal position of the candidate object at the corresponding time. If the vehicle position is always greater than the candidate object position, it is determined to be feasible; otherwise, it is determined to be infeasible, thus obtaining the prediction result that characterizes the feasibility.

[0046] For example, assuming the vehicle's current speed is 40 km / h, the system delay is 0.2 s, the maximum acceleration is 0.3 m / s², and the longitudinal position of the upper boundary point when a pedestrian crosses the road is 25 m at t=1.0 s; the vehicle control system calculates the predicted position of the vehicle at t=1.0 s. If the remaining 0.8 s acceleration time after the 0.2 s delay is taken into account, the calculated position of the vehicle is 26 m. Since 26 m is greater than 25 m, and the vehicle's position is greater than the pedestrian's position at other time points, the prediction result is "feasible", that is, the vehicle can accelerate and pass first.

[0047] Therefore, by quantitatively assessing the likelihood of a vehicle accelerating to avoid an obstacle, scenarios that do not require emergency braking can be accurately identified, thus providing a preliminary risk classification basis for subsequent target selection.

[0048] Step S103: Based on the prediction results, determine the target object from multiple candidate objects.

[0049] In this embodiment, the vehicle control system determines the target object from multiple candidate objects based on the prediction results. The target object refers to a specific obstacle that is ultimately selected from the candidate objects and requires triggering vehicle control commands (such as braking or warning). For example, the target object could be the pedestrian crossing to the left among all the "infeasible" candidate objects, with the shortest predicted collision time.

[0050] It can be seen that the vehicle control system first filters the prediction results of all candidates, eliminates the candidates whose prediction results are "feasible", retains only the candidates whose prediction results are "infeasible" to form a filter set, and selects one of the candidates as the target object.

[0051] For example, the vehicle control system detects three candidate objects: pedestrian A, vehicle B, and vehicle C. Pedestrian A is predicted as "feasible," while vehicles B and C are predicted as "infeasible." The vehicle control system eliminates pedestrian A, calculates the predicted collision time for vehicle B as 1.5 seconds, and the predicted collision time for vehicle C as 2.0 seconds. Since 1.5 seconds is less than 2.0 seconds, the vehicle control system identifies vehicle B as the target object.

[0052] Thus, by prioritizing the elimination of avoidable targets and filtering based on the urgency of the collision, the vehicle control system ensures that it only responds to the highest-risk targets that cannot be avoided by accelerating.

[0053] Step S104: Based on the relative motion information between the target object and the current vehicle, generate control commands for the current vehicle, wherein the control commands are used to instruct the current vehicle to perform braking and / or warning actions.

[0054] In this embodiment, the vehicle control system generates control commands for the current vehicle based on the relative motion information between the target object and the current vehicle. These control commands instruct the current vehicle to perform braking and / or warning actions. Examples of such control commands include "triggering the automatic emergency braking system's first-level braking" or "issuing a forward collision warning (FCW) alarm."

[0055] Relative motion information refers to dynamic data such as the relative distance and relative speed between the target object and the vehicle, as well as the predicted collision time calculated from them. For example, relative motion information might include a distance of 50 meters between the vehicle and the pedestrian, a relative speed of 15 m / s, and a predicted collision time of 1.2 seconds.

[0056] Braking refers to the physical process of a vehicle decelerating or stopping, such as the brake pads clamping the brake disc to slow the vehicle down. Warning actions refer to the physical process of alerting the driver to risks through audible and visual signals, such as a flashing red light on the dashboard accompanied by a buzzer sound.

[0057] As can be seen, the vehicle control system calculates or reads the relative motion information between the target object and the vehicle (such as the collision time), determines the risk level, and generates control commands accordingly to instruct the current vehicle to perform braking actions to avoid a collision or to perform warning actions to remind the driver.

[0058] For example, assuming that after processing, the vehicle control system determines that the pedestrian's time-to-committee (TTC) ahead is 1.2 seconds, which is less than the emergency braking threshold. The vehicle control system then generates a braking command and sends it to the actuator, while simultaneously generating a warning command to trigger a buzzer, instructing the vehicle to begin braking and alerting the driver.

[0059] Thus, specific control commands are generated from the identified risk targets, realizing closed-loop control from risk identification to vehicle execution, ensuring that braking or warning measures can be taken in a timely manner when facing collision risks.

[0060] In summary, this embodiment employs vehicle control by acquiring the current vehicle's motion information, dynamic constraint parameters, and spatiotemporal boundary information of multiple candidate objects. A predicted intersection point exists between the current vehicle's trajectory and the candidate object's trajectory. Motion prediction is performed based on the vehicle's motion information, dynamic constraint parameters, and spatiotemporal boundary information to obtain a prediction result. This prediction result characterizes the feasibility of the current vehicle accelerating past the candidate object to pass the predicted intersection point. Based on the prediction result, a target object is determined from multiple candidate objects. Control commands for the current vehicle are generated based on the relative motion information between the target object and the current vehicle. These control commands instruct the current vehicle to perform braking and / or warning actions. This achieves the goal of accurately identifying scenarios where the vehicle can accelerate to avoid crossing obstacles, thereby significantly reducing the false trigger rate of the emergency braking system, improving driving comfort and system reliability, and solving the technical problems of high false trigger rate and low control strategy accuracy in existing automatic emergency braking systems.

[0061] Optionally, the vehicle motion information includes the vehicle speed; the dynamic constraint parameters include the current vehicle's response delay time and the upper limit of longitudinal acceleration; the spatiotemporal boundary information includes a set of upper boundary points, which characterizes the spatiotemporal occupancy of the candidate object's leading edge boundary at the prediction intersection. Specifically, the set of upper boundary points (upper_points) refers to a set of data points representing the leading edge (head) position of the candidate object in the time-space coordinate system, used to describe the farthest longitudinal range occupied by the obstacle within the prediction time period. For example, the set of upper boundary points can be specifically {(0.5s, 12m), (1.0s, 18m), (1.5s, 24m)}, indicating that the obstacle's leading edge is at 12 meters at 0.5 seconds, at 18 meters at 1.0 seconds, and so on.

[0062] Spatiotemporal boundary information can also include a set of lower boundary points (lower_points). The lower boundary point set refers to a series of data points representing the trailing edge position of a candidate object in the spatiotemporal coordinate system (ST coordinate system). In the space-time dimension, these points describe the longitudinal position of the obstacle's tail as it changes over time. Together with the upper boundary points, they constitute the complete occupancy range of the obstacle in both time and space, used to define the innermost longitudinal range occupied by the obstacle.

[0063] In this embodiment of the application, the vehicle speed It refers to the instantaneous linear velocity of the vehicle in the longitudinal direction of travel, and is the basic parameter for calculating vehicle displacement.

[0064] Response latency This refers to the inherent time interval between when the vehicle control system detects an obstacle and when the vehicle executes a action, encompassing the time required for perception, decision-making, and the actions of the actuators. For example, a response delay time of 0.4 seconds means that the vehicle control system needs 0.4 seconds to react.

[0065] The upper limit of longitudinal acceleration, i.e., the maximum permissible acceleration. Longitudinal acceleration refers to the maximum physical acceleration threshold allowed for a vehicle's longitudinal acceleration or deceleration, used to limit the vehicle's extreme motion capabilities. For example, the upper limit of longitudinal acceleration can be specifically 0.3 m / s², indicating that the vehicle can only accelerate at this maximum.

[0066] It can be seen that the vehicle motion information mainly includes the vehicle speed; the dynamic constraint parameters specifically include response delay time and longitudinal acceleration upper limit; the spatiotemporal boundary information specifically includes the upper boundary point set, and it is pointed out that the role of this set is to characterize the spatiotemporal occupancy of the candidate object's leading edge boundary at the predicted intersection point, that is, to provide the longitudinal position sequence of the obstacle on the time axis.

[0067] Therefore, by clearly defining the specific composition of key parameters, a clear and standardized data foundation is provided for subsequent accurate prediction calculations based on dynamic constraints and ST boundaries.

[0068] Optionally, motion prediction is performed based on the vehicle's motion information, dynamic constraint parameters, and spatiotemporal boundary information, yielding prediction results including:

[0069] Step S1021: Traverse the set of upper boundary points, and calculate the predicted longitudinal position of the current vehicle based on the target timestamp, vehicle motion information and dynamic constraint parameters corresponding to the current upper boundary point during the traversal.

[0070] Step S1022: Determine the prediction result based on the predicted longitudinal position of each upper boundary point in the upper boundary point set and the longitudinal position of the boundary point.

[0071] In this embodiment, the vehicle control system first traverses the set of upper boundary points and calculates the predicted longitudinal position of the current vehicle based on the target timestamp, vehicle motion information, and dynamic constraint parameters corresponding to the current upper boundary point during the traversal.

[0072] The target timestamp refers to the time corresponding to the specific upper boundary point currently being processed, representing the predicted time when the obstacle exists at that location.

[0073] Predicted longitudinal position, i.e., the longitudinal position of the vehicle at that point in time. This refers to the current speed of the vehicle, system response delay, and maximum allowable acceleration. The longitudinal coordinate value that the vehicle can reach at the target timestamp is calculated, reflecting the theoretical displacement capability of the vehicle under acceleration or constant speed conditions.

[0074] As can be seen, by iteratively visiting each point in the upper boundary point set, extracting the corresponding timestamp for each point, and combining it with the vehicle's current speed, system response delay time, and longitudinal acceleration upper limit, the longitudinal position that the vehicle can reach at that specific time point is calculated using kinematic formulas.

[0075] For example, the vehicle control system retrieves the first point from the set at a time of 0.5 seconds, with a position of 12 meters. The vehicle's position at that moment is calculated: after a 0.2-second delay of constant speed, followed by an acceleration of 0.3 m / s² for the next 0.3 seconds, the vehicle's position is approximately 5.01 meters. Then, the second point is retrieved at a time of 1.0 second, with a position of 18 meters. The vehicle's position at that moment is calculated: after a 0.2-second delay of constant speed, followed by an acceleration of 0.8 seconds, the vehicle's position is approximately 10.96 meters. This process continues until all points have been traversed.

[0076] Therefore, this step achieves a refined and quantitative assessment of the vehicle's acceleration and obstacle avoidance capabilities by calculating the theoretical position of the vehicle at each moment at the leading edge of the obstacle.

[0077] Then, the vehicle control system determines the prediction result based on the predicted longitudinal position corresponding to each upper boundary point in the set of upper boundary points and the longitudinal position of the boundary point.

[0078] Among them, the longitudinal position of the boundary point This refers to the longitudinal coordinate of the specific upper boundary point currently being processed in the ST coordinate system, i.e., the actual position of the leading edge of the obstacle at that moment. By comparing the magnitudes of these two coordinates, it is determined whether the vehicle is in front of the obstacle at that moment.

[0079] As can be seen, the vehicle control system compares the predicted longitudinal position of the vehicle with the longitudinal position of the currently traversed upper boundary points one by one. By checking the comparison of all points in the set of upper boundary points, it comprehensively determines whether the vehicle is ahead of the leading edge of the obstacle at all key time points, thereby determining whether the overall prediction result supports the vehicle's ability to pass ahead.

[0080] Therefore, by verifying that the vehicle is behind the trailing edge of the obstacle at all critical time points, the accuracy of the judgment that the vehicle has completely overtaken the obstacle is ensured, preventing the risk of rear collision from being ignored due to only focusing on the front overtaking, and improving the safety and reliability of the leading decision.

[0081] Optionally, the predicted longitudinal position is calculated based on the target timestamp, vehicle motion information, and dynamic constraint parameters, including:

[0082] Step S10211: If the interval between the current timestamp and the target timestamp is less than or equal to the response delay time, calculate the predicted longitudinal position based on the vehicle speed and the interval time.

[0083] Step S10212: If the interval time is greater than the response delay time, calculate the predicted longitudinal position based on the vehicle speed, the response delay time, and the upper limit of the longitudinal acceleration.

[0084] In this embodiment, if the interval between the current timestamp and the target timestamp is less than or equal to the response delay time, the vehicle control system calculates and predicts the longitudinal position based on the vehicle speed and the interval. Here, t refers to the starting time of the calculation action in this embodiment, typically marked as t=0, serving as the reference point for the calculation time interval. In the vehicle control scenario, this corresponds to the latest sensor data moment or the moment the algorithm triggers the calculation.

[0085] It can be seen that when the time interval from the current timestamp to the target timestamp is less than or equal to the response delay time, the vehicle control system determines that the vehicle cannot perform an acceleration action during this period. Therefore, it assumes that the vehicle maintains its current speed and moves at a constant speed. By multiplying the vehicle speed by the interval time, the predicted longitudinal position of the vehicle at the target timestamp is obtained.

[0086] For example, assume the vehicle's current speed is 10 m / s and the vehicle control system's response delay is 0.4 seconds. If it is necessary to calculate the vehicle's position 0.2 seconds later (i.e., the target timestamp is the current time + 0.2 seconds), since 0.2 seconds is less than 0.4 seconds, it falls within the response delay period. At this time, the longitudinal position is predicted. .

[0087] If the interval is greater than the response delay time, the vehicle control system will calculate and predict the longitudinal position based on the vehicle speed, the response delay time, and the upper limit of longitudinal acceleration.

[0088] It can be seen that when the time interval between the predicted moment and the current moment is greater than the response delay time, the vehicle control system assumes that the vehicle will begin to accelerate at the upper limit of longitudinal acceleration after the delay period. The calculation logic is divided into two parts: the first part is the distance traveled at the current speed during the delay period, and the second part is the distance traveled at the maximum allowable acceleration after the delay period. The sum of the two is the predicted longitudinal position after considering acceleration capability.

[0089] For example, the vehicle's speed is 10 m / s, the response delay is 0.3 seconds, and the maximum longitudinal acceleration is 0.3 m / s². If the position needs to be calculated after 0.5 seconds (0.5 seconds is greater than 0.3 seconds), then in the first 0.3 seconds, the vehicle travels at a constant speed for 10 × 0.3 = 3 meters; in the following 0.2 seconds, it accelerates at 0.3 m / s², increasing the distance by 0.5 × 0.3 × 0.2² = 0.006 meters. The total predicted longitudinal position is 3.006 meters.

[0090] Therefore, by distinguishing the motion state inside and outside the delay period, the actual physical process of the vehicle from perception to execution is accurately simulated, avoiding the leading judgment bias caused by ignoring the delay of the vehicle control system or overestimating the acceleration capability.

[0091] Optionally, based on the predicted longitudinal position of each upper boundary point in the set of upper boundary points and the longitudinal position of the boundary point, the prediction result is determined to include:

[0092] Step S10221: If for all upper boundary points in the set of upper boundary points, the predicted longitudinal position is greater than the longitudinal position of the boundary point, then the prediction result is determined to be feasible.

[0093] Step S10222: If for at least one boundary point in the set of upper boundary points, the predicted longitudinal position is less than or equal to the longitudinal position of the boundary point, then the prediction result is determined to be infeasible.

[0094] In this embodiment of the application, if the predicted longitudinal position of all upper boundary points in the set of upper boundary points is greater than the longitudinal position of the boundary point, then the prediction result is determined to be feasible.

[0095] As can be seen, the algorithm iterates through each upper boundary point in the set of upper boundary points, comparing the predicted longitudinal position of the vehicle at that point with the longitudinal position of the obstacle's boundary point at each corresponding moment. Since the current vehicle's driving direction is the positive longitudinal direction, if the predicted longitudinal position corresponding to all upper boundary points in the set is strictly greater than the corresponding longitudinal position of the boundary point, it indicates that the vehicle is in front of the obstacle at all key time points (i.e., the vehicle's coordinates are greater than the obstacle's coordinates), thus determining that the prediction result is feasible.

[0096] For example, suppose the upper boundary point set contains two points: the obstacle position is 10 meters at time point t_1 and 20 meters at time point t_2. If the predicted longitudinal position of the vehicle at time t_1 is calculated to be 12 meters (12>10), and the predicted longitudinal position at time t_2 is 25 meters (25>20), since all points satisfy the condition that the vehicle position is greater than the obstacle position, the prediction result is determined to be feasible. That is, if for all checked upper boundary points, there is... If the vehicle's position is greater than the obstacle's position, then it can be determined that a leading decision can be made, meaning that the vehicle can accelerate to pass the intersection before the obstacle.

[0097] If, for at least one boundary point in the set of upper boundary points, the predicted longitudinal position is less than or equal to the longitudinal position of the boundary point, then the prediction result is determined to be infeasible.

[0098] It can be seen that the vehicle control system traverses the set of upper boundary points. If the predicted longitudinal position corresponding to any upper boundary point is less than or equal to the longitudinal position of the corresponding boundary point, it indicates that the vehicle failed to pass the leading edge of the obstacle at a certain moment, and there is a potential collision or following state. At this time, the check is immediately terminated and the prediction result is determined to be infeasible.

[0099] For example, in the set of upper boundary points The obstacle was located 10 meters away. The obstacle was located 20 meters away. If the calculation shows the vehicle was... The predicted longitudinal position at that time was 9 meters (9 meters is less than 10 meters), although The position is 25 meters (25>20), but there is at least one point ( If the condition is not met, the prediction result is determined to be infeasible. That is, if there exists some upper boundary point such that... If it is determined that a leading decision cannot be made, the obstacle needs to be treated as the target object for collision risk assessment.

[0100] Furthermore, if a preliminary assessment indicates that a leading decision can be made, subsequent safety redundancy verification is required. This verification includes at least obstacle-related acceleration verification, obstacle-related emergency braking or stalling verification, perception error compensation, and control execution error compensation. If, in any verification scenario, the vehicle's leading distance or leading time is less than a preset safety threshold, the leading decision is revoked, and the obstacle crossing is treated as a candidate target.

[0101] Therefore, by using a rigorous comparison logic across all points, it accurately determined whether the vehicle could complete the crossing before the leading edge of the obstacle, providing a clear basis for subsequent leading decisions.

[0102] Optionally, the method of determining the target object from multiple candidate objects based on the prediction results includes:

[0103] Step S1031: From multiple candidate objects, filter out candidate objects whose prediction results are infeasible to obtain the filtering results;

[0104] Step S1032: From the filtering results, select the candidate object with the smallest predicted collision time with the current vehicle as the target object.

[0105] In this embodiment, the vehicle control system first filters candidate objects whose predictions are deemed infeasible from a pool of candidates, obtaining a filtering result. The filtering result refers to the set of all remaining candidate objects whose predictions are deemed infeasible after removing those predicted as "feasible" (i.e., the vehicle can safely pass ahead). This is the candidate pool that enters the final target selection process.

[0106] As can be seen, the vehicle control system iterates through all candidate objects, checking the prediction results of each object obtained in the previous steps. Only those objects predicted as "infeasible" are retained and added to the filter result set; objects predicted as "feasible" are ignored and no longer participate in the subsequent target selection process.

[0107] For example, suppose the vehicle control system detects three candidate objects: object A (pedestrian), object B (vehicle in front), and object C (vehicle to the side). The calculation shows that object A is predicted as "feasible" (the vehicle can accelerate through), while objects B and C are both predicted as "infeasible".

[0108] Then, the vehicle control system selects the candidate vehicle with the shortest predicted collision time from the screening results as the target vehicle. The predicted collision time, based on the relative speed, relative distance, and motion model between the current vehicle and the candidate vehicle, is the predicted time required for physical contact or entry into a safe threshold range between the two vehicles, typically denoted by TTC. For example, if the two vehicles are 100 meters apart and their relative speed is 20 meters per second, the predicted collision time is approximately 5 seconds.

[0109] As can be seen, the algorithm calculates the predicted collision time between each candidate object in the filtered result set and the current vehicle, compares these time values, and selects the candidate object with the smallest value as the final target object. This follows the principle of "prioritizing the most urgent risks".

[0110] For example, the result set includes object B and object C. The predicted collision time for object B is calculated to be 3 seconds, and the predicted collision time for object C is 5 seconds. Comparing the two, 3 seconds is less than 5 seconds, therefore object B is selected as the target object, and control instructions will be generated for object B subsequently.

[0111] Therefore, by first filtering targets that can be safely passed and then prioritizing them according to collision urgency, the system ensures that the vehicle only takes braking or warning measures against targets that truly pose an imminent threat.

[0112] Optionally, the relative motion information includes predicted collision time. Based on the relative motion information between the target object and the current vehicle, the control commands for the current vehicle are generated, including:

[0113] Step S1041: If the predicted collision time is less than the first threshold, generate an emergency braking control command.

[0114] Step S1042: If the predicted collision time is greater than or equal to the first threshold and the predicted collision time is less than the second threshold, generate a warning control command.

[0115] In this embodiment, if the predicted collision time is less than a first threshold, an emergency braking control command is generated. The first threshold refers to the upper limit of the predicted collision time used to define the emergency braking trigger condition, typically corresponding to the critical point at which the system needs to immediately execute full braking to avoid a collision. For example, it can be set to 1.5 seconds, meaning that when the TTC is less than 1.5 seconds, the vehicle control system considers the situation critical.

[0116] Emergency braking control commands are instructions to control the vehicle's braking system to decelerate at or near its maximum capacity, with the aim of minimizing the collision speed or avoiding a collision.

[0117] As can be seen, this step is used to handle high-risk scenarios. The vehicle control system compares the predicted collision time between the target object and the current vehicle with a first threshold. If the predicted collision time is less than the first threshold, it indicates that the collision risk is extremely high and time is of the essence. The vehicle control system directly generates an emergency braking control command, requiring the vehicle to immediately perform high-intensity braking.

[0118] If the predicted collision time is greater than or equal to a first threshold and less than a second threshold, a warning control command is generated. The second threshold refers to the upper limit of the predicted collision time used to define the warning trigger condition; its value is greater than the first threshold and typically corresponds to the critical point at which the vehicle control system needs to alert the driver or pre-brake. For example, setting it to 2.5 seconds means that when the TTC (Traffic Tactical Control) is between 1.5 and 2.5 seconds, the vehicle control system considers there to be a potential risk but still has a margin of safety.

[0119] Warning control commands refer to commands that alert the driver through sound, light, vibration, or other means, or trigger pre-braking, with the aim of warning of risks or putting the braking system in a standby state.

[0120] As can be seen, this step is used to handle low-to-medium risk scenarios. The vehicle control system checks whether the predicted collision time is between a first threshold and a second threshold. Specifically, the judgment condition is that the predicted collision time is greater than or equal to the first threshold and simultaneously less than the second threshold. If this condition is met, it indicates that there is a collision risk but there is still a buffer time. The vehicle control system generates a warning control command to remind the driver or prepare for pre-braking, rather than immediately performing emergency braking.

[0121] Therefore, by setting dual thresholds to differentiate risk levels, a graded response to collision risks of different urgency levels is achieved, taking into account both driving safety and driving comfort.

[0122] Optionally, the method further includes:

[0123] Step S1033: For candidate objects whose prediction results are feasible, calculate the longitudinal safe distance for the current vehicle to pass through the predicted intersection based on the unexpected acceleration of the candidate object;

[0124] Step S1034: If the longitudinal safety distance is less than the preset distance threshold, adjust the prediction result corresponding to the candidate object to infeasible.

[0125] In this embodiment, firstly, for candidate objects whose prediction results are feasible, the vehicle control system calculates the longitudinal safe distance for the current vehicle to pass through the predicted intersection based on the unexpected acceleration of the candidate object.

[0126] Unexpected acceleration refers to changes in acceleration that may occur in a candidate object (such as a pedestrian, bicycle, or vehicle) beyond its current motion trend, including unpredictable motion behaviors such as sudden acceleration, rapid deceleration, or sudden stop. For example, a pedestrian suddenly accelerates across the road or a bicycle suddenly stops.

[0127] Longitudinal safety distance refers to the minimum longitudinal distance that a vehicle needs to maintain with a candidate object when passing through a predicted intersection in order to ensure safety. This distance is a redundancy space calculated based on the unexpected acceleration capability of the candidate object.

[0128] As can be seen, this step further validates candidate objects whose predictions are feasible. The vehicle control system reassesses the required safe distance for the vehicle to pass the predicted intersection based on the unexpected acceleration parameters of the candidate object. Specifically, the vehicle control system considers possible unexpected acceleration or sudden braking of the candidate object and calculates the longitudinal safety distance that the vehicle must maintain when passing the intersection.

[0129] For example, suppose a pedestrian is predicted to be "feasible," but their unexpected acceleration is large (such as suddenly accelerating across). The vehicle control system calculates based on this unexpected acceleration that the vehicle needs to maintain a longitudinal distance of at least 5 meters from the pedestrian when passing through the intersection to ensure safety; that is, the calculated longitudinal safety distance is 5 meters.

[0130] Then, if the longitudinal safety distance is less than a preset distance threshold, the vehicle control system adjusts the prediction result corresponding to the candidate object to infeasible. The preset distance threshold refers to a standard value set by the vehicle control system to determine whether the longitudinal safety distance is sufficient, typically calibrated based on sensor error, control execution error, and safety margin.

[0131] As can be seen, this step is used to correct potential misjudgments. The vehicle control system compares the longitudinal safety distance with a preset distance threshold. If the longitudinal safety distance is less than the preset distance threshold, it means that even if the vehicle passes ahead, the safety margin is insufficient, and there is a risk of collision due to unforeseen circumstances. Therefore, the vehicle control system adjusts the original "feasible" prediction to "infeasible," thereby re-including the candidate in the risk candidate pool.

[0132] For example, the vehicle control system sets a preset distance threshold of 6 meters. Since the calculated longitudinal safety distance of 5 meters is less than the preset distance threshold of 6 meters, indicating insufficient safety margin, the vehicle control system adjusts the pedestrian prediction result from "feasible" to "infeasible", and will subsequently perform emergency braking or warning processing on the pedestrian.

[0133] Therefore, by introducing unexpected acceleration verification and dynamically adjusting the prediction results, the risk of brake leakage caused by theoretical advancement but insufficient actual safety margin is effectively reduced.

[0134] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0135] Furthermore, when acquiring the spatiotemporal boundary information of crossing obstacles, this application first performs a coordinate transformation based on the sensor extrinsic parameters, transforming the target point from the sensor coordinate system to the vehicle coordinate system. The specific transformation relationship follows the formula... ,in and These are the rotation matrix and translation vector from the sensor coordinate system to the vehicle coordinate system, respectively. These are the coordinates of the target point in the sensor coordinate system. These are the coordinates of the target point in the vehicle's coordinate system; subsequently, the planned trajectory of the vehicle or the center line of the lane is used as a reference line. Predicting the trajectory of obstacles and the corner points of the bounding box Projection processing is performed by solving for the minimum distance function. Determine the projected arc length, and then... Defined as the longitudinal coordinate of the corner point along the vehicle's trajectory; based on this, for each prediction time point... The corner points of the obstacle bounding boxes that fall within the vehicle's driving corridor or collision detection corridor are selected to calculate the longitudinal occupancy range. This will further determine the maximum value of the interval. Marked as upper boundary point, minimum value The points are marked as lower boundary points and accumulated in chronological order to form upper boundary point sets and lower boundary point sets. Finally, the generated point sets are subjected to data cleaning and verification operations such as time sorting, duplicate point merging, abnormal jump removal, and confidence verification, thereby outputting high-precision and reliable ST boundary information for leading decision-making.

[0136] Furthermore, this application provides an anomaly handling mechanism designed to ensure the functional safety of the vehicle control system in the event of incomplete or unreliable perception data. Specifically, when the vehicle control system detects anomalies in the spatiotemporal boundary information, it determines to enter a conservative safety state. Anomalies in the spatiotemporal boundary information include, but are not limited to, any of the following: missing upper boundary point set, missing lower boundary point set, insufficient number of boundary points in the spatiotemporal boundary information, abnormal timestamps, target detection confidence output by the perception sensor being lower than a preset threshold, or failure to convert the sensor coordinate system to the vehicle coordinate system. Once in this conservative safety state, the vehicle control system prohibits skipping operations based on leading decisions. That is, it does not allow ignoring a crossing obstacle based on a preliminary judgment that it can pass ahead; instead, it forcibly retains the crossing obstacle in the candidate target object set and generates control commands based on the relative motion information between the candidate target object and the current vehicle, following conventional automatic emergency braking or forward collision warning collision risk logic. This allows for the execution of warning or braking actions when preset trigger conditions are met, thereby avoiding safety omissions due to missing or incorrect data.

[0137] Figure 2 This is a schematic flowchart of an optional vehicle control method according to an embodiment of this application, such as... Figure 2 As shown, the vehicle control system first acquires the boundary ST of the obstacle crossing. and The algorithm determines whether a leading decision can be made. If so, the obstacle is skipped and not selected as a target, and the process ends. If a leading decision cannot be made, the obstacle's Time-to-Collision (TTC) collision time is calculated and compared with other candidate targets; the one with the smallest TTC is selected as the final target.

[0138] According to an embodiment of this application, a vehicle control method is provided. Figure 3 This is a schematic flowchart of another optional vehicle control method according to an embodiment of this application, such as... Figure 3 As shown, the vehicle control method provided in this embodiment includes the following five steps.

[0139] The first step is for the vehicle control system to acquire ST boundary information that crosses the obstacle.

[0140] In this step, the vehicle control system acquires the ST boundary information of the obstacle. The ST boundary is the boundary representation of the obstacle in a space-time coordinate system, where S represents the longitudinal coordinate of the obstacle along the vehicle's reference line or planned trajectory, and T represents the prediction time relative to the current moment. This boundary is jointly determined by the obstacle's perception results, the predicted trajectory, and its spatial occupancy within the vehicle's driving corridor. Specifically, the acquired information includes the upper boundary point set and the lower boundary point set.

[0141] The upper boundary point set represents the set of trajectory points of the obstacle's head (leading edge) in the ST coordinate system, with each point containing two dimensions: time t and longitudinal position s. The upper boundary points typically correspond to the positions of the obstacle's front end at different times. The lower boundary point set represents the set of trajectory points of the obstacle's tail (trailing edge) in the ST coordinate system.

[0142] In practical implementation, the upper and lower boundary points can be obtained in the following ways: data acquisition, acquiring obstacle information output by cameras, millimeter-wave radar, lidar or other vehicle perception sensors, including target category, center point position, relative speed, heading angle, bounding box length and width, predicted trajectory and detection confidence, while acquiring the vehicle speed v_ego, yaw rate, steering angle, positioning attitude, planned trajectory or lane center line, and the external parameters of each sensor relative to the vehicle coordinate system;

[0143] Coordinate transformation, which transforms the obstacle points in the sensor coordinate system. The conversion formula to the vehicle coordinate system is as follows: For historical or predicted trajectory points, further integration with vehicle pose transformation can be achieved. This unifies points at different times into the current vehicle coordinate system or the global coordinate system;

[0144] Longitudinal coordinate calculation uses the vehicle's planned trajectory or lane centerline as a reference line. s is the reference line arc length parameter, used for predicting the trajectory points and bounding box corners of obstacles. Find the arc length of the nearest reference line The lateral offset can be set according to Calculation, where As a reference line The heading angle at that location. The point is used as the longitudinal coordinate along the vehicle's direction of travel, and the difference between the predicted timestamp and the current time is used as... .

[0145] Obtain the set of upper and lower boundary points of the obstacle: for each prediction time Calculate the longitudinal coordinates of all corner points of the obstacle bounding box or occupancy polygon within the vehicle's driving corridor to obtain the longitudinal occupancy range. ;Will As the upper boundary point, As the lower boundary point, add respectively and If an obstacle only partially enters the vehicle's driving corridor, only the boundary of the part entering the corridor is projected to reflect the actual collision-related occupancy, and then subsequent initialization operations are performed.

[0146] Point set verification and filtering, for and According to time Sort in ascending order and remove points with backward timestamps, vertical coordinate abrupt changes exceeding a preset threshold, confidence scores below a preset threshold, or significantly inconsistent with the direction of obstacle movement. For slight jitter between adjacent time points, a moving average or Kalman filter can be used for smoothing.

[0147] Confidence level assignment: The ST boundary confidence level is calculated based on sensor detection confidence, sensor fusion consistency, coordinate transformation residual, and predicted trajectory stability. .when t is lower than the preset threshold At that time, leading release decisions are not allowed based on the ST boundary.

[0148] The second step is for the vehicle control system to determine whether it can make a priority decision to pass the obstacle.

[0149] The specific implementation includes the following sub-steps S301 to S306.

[0150] Step S301: Obtain the vehicle's current speed and system response delay time. The vehicle control system obtains the vehicle's current speed. (Unit: m / s) and system response delay time (Unit: s). This represents the time delay from obstacle detection to the system's ability to execute braking. The delay includes: sensor data processing time (typically 0.05 to 0.1 seconds), algorithm calculation time (typically 0.05-0.1 seconds), and braking system response time (typically 0.1-0.3 seconds). In actual testing, the vehicle control system detects... The typical value is between 0.2 and 0.5 seconds, and the specific value needs to be calibrated according to the actual system configuration.

[0151] Step S302: The vehicle control system sets the maximum permissible acceleration. After testing, setting the maximum permissible acceleration to 0.3 m / s² effectively reduces false triggering while ensuring safety.

[0152] Step S303: Traverse the upper boundary points of obstacles and calculate the vehicle's position. Traverse the set of upper boundary points of obstacles. For each upper boundary point The longitudinal position of the vehicle at that point in time is calculated after taking into account system latency and acceleration capability. .

[0153] Here we need to distinguish between the following two situations.

[0154] Case 1 occurs when the time at the upper boundary point is greater than the system latency. In this case, system latency and acceleration capability need to be considered. The lead time is calculated as follows: This indicates that after the system delay, the vehicle still has... The time can be accelerated; calculate the longitudinal position of the vehicle at that point. ,in For the vehicle to travel at its current speed Position after the time. In order to be in Within a time period, with The distance increased by acceleration is added together to obtain the total position after taking acceleration into account.

[0155] Case 2 is This means that the time at the upper boundary point is less than or equal to the system delay time. At this point, the vehicle control system has not yet responded, cannot accelerate, and can only travel at the current speed. The calculation process is as follows: .

[0156] Step S304: The vehicle control system performs position comparison and decision-making. For the upper boundary point, the comparison... and :if This indicates that the vehicle's position at that point is greater than the obstacle's position, meaning the vehicle is already ahead of the obstacle and a collision will not occur, allowing for the next point to be assessed. If This indicates that the vehicle's position at that point is less than or equal to the obstacle's position, meaning the vehicle has not yet passed the obstacle and there is a risk of collision.

[0157] During the traversal, a special case also needs to be considered: if the time of a certain upper boundary point... (in It is the first upper boundary point. If the condition is met, it will return true, indicating that a leading decision can be made.

[0158] The basis for this judgment is that if an obstacle has passed for a sufficiently long time (0.5 seconds), then even if there was previous overlap, the obstacle has moved away and there will be no longer a risk of collision. Therefore, the 0.5-second threshold was determined through actual testing and can effectively filter out obstacles that have already moved away.

[0159] Step S305: Comprehensive judgment result. If for all checked upper boundary points, there is... If a leading decision can be made, then return true. If there exists an upper boundary point such that... If the condition is not met, it is determined that a leading decision cannot be made, and false is returned.

[0160] Step S306: Safety Redundancy Verification. After the preliminary determination in Step S305 that a leading decision can be made, the vehicle control system does not immediately skip the obstacle crossing, but further performs a safety redundancy verification to cover the risks caused by sudden acceleration, abrupt stopping, or stalling while crossing the obstacle, as well as perception control errors. Specifically, for each upper boundary point involved in the verification... Calculate the lead distance margin And calculate the safety redundancy distance. . At least including sensor measurement error Coordinate transformation error Control execution error Preset time margin converted distance and the distance of uncertainty in obstacle behavior .

[0161] in, It can be calibrated according to the type of obstacle. For example, for pedestrians, it can be based on the maximum accidental acceleration. and emergency stop time Generate conservative boundaries; for bicycles or vehicles, extended boundaries can be generated based on the corresponding maximum acceleration, maximum deceleration, and prediction confidence. The vehicle control system constructs unexpected acceleration trajectories and emergency stop / stagnation trajectories respectively, and assigns the extended upper boundary points... Used for verification.

[0162] When all checkpoints are satisfied If the vehicle passes through the intersection area at least kMinLeadTimes earlier than the earliest risk time when the obstacle may occupy the area, the safety redundancy check passes; otherwise, the safety redundancy check fails, the vehicle control system cancels the leading decision, and continues to perform AEB and FCW risk assessments on the obstacle as a candidate target.

[0163] Through this safety redundancy check, even if theoretical calculations show that the vehicle can pass through first, if there is a sudden acceleration, abrupt stop, or insufficient safety margin due to detection error when crossing the obstacle, the vehicle control system will still trigger conservative processing, thereby reducing the risk of missed braking.

[0164] Third, the vehicle control system classifies obstacles based on the leading decision judgment and the safety redundancy check results. If the system determines that the vehicle can accelerate to pass the intersection before the obstacle, and this judgment passes the safety redundancy check, which covers unexpected acceleration, sudden stop or stagnation, perception errors, and control execution errors related to the obstacle, then the vehicle is confirmed to be in a safe passage state and does not need to trigger emergency braking. Therefore, the obstacle is skipped and not selected as a target. Conversely, if the system cannot make a leading decision, or if it initially determines that it can pass but the safety redundancy check fails, the obstacle is marked as a candidate target and the subsequent target selection process begins.

[0165] Fourth, for all traversing obstacles that enter the candidate list (including obstacles for which a leading decision cannot be made, which fail the safety redundancy check, or which are in a conservative safety state), the vehicle control system calculates the time-to-collision (TTC) with the vehicle. The calculation formula is as follows: The vehicle control system compares the Time-to-Cross (TTC) of all candidate targets, selects the target with the lowest TTC as the final decision target, and triggers the corresponding level of forward collision warning or automatic emergency braking command to ensure timely response to the highest-risk target.

[0166] Fifth, in actual operation, if data anomalies or edge cases are encountered, such as missing upper / lower boundary points, insufficient ST boundary points, abnormal timestamps, insufficient sensor confidence, inconsistent multi-sensor fusion, unavailable reference lines, or failed coordinate transformation, the vehicle control system immediately enters a conservative safety state. In this state, the vehicle control system prohibits the execution of lead-through logic, forcibly retaining the crossing obstacle as a candidate target to prevent missed detections due to missing data. Simultaneously, the vehicle control system can output corresponding degradation flags (such as ST_BOUNDARY_INVALID, SENSOR_CONF_LOW, or TRANSFORM_FAILED) for subsequent diagnostic recording, data review, and function degradation control. This ensures the safety of the vehicle control system while preventing the erroneous release of potentially risky targets due to incomplete ST boundary extraction.

[0167] In the specific implementation process, the vehicle control system can flexibly configure the upper boundary point inspection strategy for crossing obstacles based on computing resources and real-time requirements. In addition to traversing all upper boundary points, the vehicle control system can adopt a time-interval-based sampling inspection strategy, such as selecting a point for verification every fixed interval (e.g., 0.1 seconds) to reduce the computational load; or adopt a key point inspection strategy, only judging the start point, end point, and intermediate key points identified based on motion features in the sequence, thereby focusing on potential risks while ensuring efficiency; in addition, the vehicle control system can also introduce a time window limitation mechanism, only checking upper boundary points within a preset time period (e.g., 2 seconds) in the future, to focus on the rapid assessment of near-term risks.

[0168] In the safety redundancy verification stage, the parameter settings for safety redundancy distance and minimum lead time have a high degree of scene adaptability, and can be dynamically calibrated according to vehicle dynamics characteristics, sensor configuration, and target object type. For example, for pedestrian targets, a larger abrupt stop time redundancy can be set to cope with the risk of sudden stop; for bicycle targets, a larger lateral acceleration uncertainty redundancy can be introduced to cover the instability of its trajectory; and for vehicle targets, a larger longitudinal velocity uncertainty redundancy can be set to adapt to its complex acceleration and deceleration behavior, thereby ensuring verification accuracy under different scenarios.

[0169] When the vehicle control system faces edge cases such as missing ST boundary information, insufficient confidence, or abnormal coordinate transformation, the conservative safety state handling strategy can be configured hierarchically according to the safety level set by the vehicle control system. Optional strategies include defaulting to the candidate target selection process, defaulting to triggering a forward collision warning, or defaulting to triggering emergency braking when the collision time is below a threshold. Preferably, this application adopts a strategy of not performing lead-through clearance and retaining obstacles as candidate target objects. This strategy effectively balances the safety redundancy and false triggering control of the vehicle control system by maintaining the conventional risk judgment logic, avoiding the risk of missed braking due to data anomalies, and providing clear status indicators for subsequent diagnostic recording and function degradation.

[0170] Figure 4 This is a schematic diagram of an optional vehicle control device according to an embodiment of this application, such as... Figure 4 As shown, this includes an ST boundary acquisition module, a leading decision judgment module, and a security redundancy verification unit (not included in...). Figure 4 (Demonstration), abnormal conservative handling unit (not in) Figure 4 The system includes a demonstration module and a target selection module. The ST boundary acquisition module obtains the ST boundary information crossing the obstacle and transmits it to the leading decision-making module. The leading decision-making module further includes a velocity delay acquisition unit, an acceleration setting unit, a position calculation unit, and a decision-making unit. These four units process the acquired ST boundary information crossing the obstacle in sequence and output the leading decision to the target selection module. The target selection module then selects the target object based on the leading decision result.

[0171] According to an embodiment of this application, an apparatus embodiment for a vehicle control method is provided. It should be noted that the apparatus can be used to execute the above-described vehicle control method.

[0172] Figure 5 This is a structural block diagram of an optional vehicle control device according to an embodiment of this application, such as... Figure 5 The vehicle control method 500 includes: an acquisition module 501, used to acquire the current vehicle's own motion information, dynamic constraint parameters, and spatiotemporal boundary information of multiple candidate objects, wherein there is a predicted intersection point between the current vehicle's own motion trajectory and the candidate object's motion trajectory; a prediction module 502, used to perform motion prediction based on the vehicle's own motion information, dynamic constraint parameters, and spatiotemporal boundary information to obtain a prediction result, wherein the prediction result is used to characterize the feasibility of the current vehicle accelerating past the candidate object to pass the predicted intersection point; a determination module 503, used to determine the target object from multiple candidate objects based on the prediction result; and a generation module 504, used to generate control commands for the current vehicle based on the relative motion information between the target object and the current vehicle, wherein the control commands are used to instruct the current vehicle to perform braking and / or warning actions.

[0173] Embodiments of this application also provide a vehicle, including: a memory storing an executable program; and a processor for running the program, wherein the program executes the methods described in various embodiments of this application when it runs.

[0174] Embodiments of this application also provide a computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the computer-readable storage medium is located to perform the methods of various embodiments of this application.

[0175] Embodiments of this application also provide a computer program product, including a computer program that, when executed by a processor, implements the methods of various embodiments of this application.

[0176] Embodiments of this application also provide a computer program product, including a non-volatile computer-readable storage medium for storing a computer program that, when executed by a processor, implements the methods in various embodiments of this application.

[0177] Embodiments of this application also provide a computer program that, when executed by a processor, implements the methods described in the various embodiments of this application.

[0178] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0179] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0180] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0181] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0182] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0183] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A vehicle control method characterized by, include: The vehicle's motion information, dynamic constraint parameters, and spatiotemporal boundary information of multiple candidate objects are obtained. The vehicle's motion trajectory and the candidate objects' motion trajectories have a predicted intersection point. Motion prediction is performed based on the vehicle motion information, the dynamic constraint parameters, and the spatiotemporal boundary information to obtain a prediction result, wherein the prediction result is used to characterize the feasibility of the current vehicle accelerating to pass the predicted intersection point before the candidate object; Based on the prediction results, a target object is determined from the plurality of candidate objects; Based on the relative motion information between the target object and the current vehicle, a control command for the current vehicle is generated, wherein the control command is used to instruct the current vehicle to perform braking and / or warning actions.

2. The method according to claim 1, characterized in that, The vehicle motion information includes the vehicle speed; The dynamic constraint parameters include: the response delay time and the upper limit of longitudinal acceleration of the current vehicle; The spatiotemporal boundary information includes an upper boundary point set, wherein the upper boundary point set is used to characterize the spatiotemporal occupancy of the leading edge boundary of the candidate object at the predicted intersection point.

3. The method of claim 2, wherein, Motion prediction is performed based on the vehicle motion information, the dynamic constraint parameters, and the spatiotemporal boundary information. The prediction results include: Traverse the set of upper boundary points, and calculate the predicted longitudinal position of the current vehicle based on the target timestamp corresponding to the current upper boundary point, the vehicle motion information, and the dynamic constraint parameters. The prediction result is determined based on the predicted longitudinal position and the longitudinal position of the boundary point corresponding to each upper boundary point in the set of upper boundary points.

4. The method of claim 3, wherein, Calculating the predicted longitudinal position based on the target timestamp, the vehicle motion information, and the dynamic constraint parameters includes: If the interval between the current timestamp and the target timestamp is less than or equal to the response delay time, the predicted longitudinal position is calculated based on the vehicle speed and the interval time. If the interval time is greater than the response delay time, the predicted longitudinal position is calculated based on the vehicle speed, the response delay time, and the upper limit of the longitudinal acceleration.

5. The method of claim 3, wherein, Based on the predicted longitudinal position and the longitudinal position of each upper boundary point in the set of upper boundary points, the prediction result is determined as follows: If for all upper boundary points in the set of upper boundary points, the predicted longitudinal position is greater than the longitudinal position of the boundary point, then the prediction result is determined to be feasible. If, for at least one boundary point in the set of upper boundary points, the predicted longitudinal position is less than or equal to the longitudinal position of the boundary point, then the prediction result is determined to be infeasible.

6. The method according to claim 1, characterized in that, Determining the target object from the plurality of candidate objects based on the prediction results includes: From the multiple candidate objects, the candidate objects whose prediction results are infeasible are filtered to obtain the filtering results; From the filtering results, the candidate object with the smallest predicted collision time with the current vehicle is selected as the target object.

7. The method according to claim 1, characterized in that, The relative motion information includes a predicted collision time. Based on the relative motion information between the target object and the current vehicle, the control commands for the current vehicle are generated, including: If the predicted collision time is less than the first threshold, an emergency braking control command is generated. If the predicted collision time is greater than or equal to the first threshold and the predicted collision time is less than the second threshold, an early warning control command is generated.

8. The method according to claim 6, characterized in that, The method further includes: For the candidate object whose prediction result is feasible, the longitudinal safe distance for the current vehicle to pass through the predicted intersection is calculated based on the unexpected acceleration of the candidate object; If the longitudinal safety distance is less than a preset distance threshold, the prediction result corresponding to the candidate object will be adjusted to infeasible.

9. A vehicle, characterized in that, include: Memory, which stores executable programs; A processor for running the program, wherein the program, when running, performs the method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored executable program, wherein, when the executable program is executed, it controls the device on which the storage medium is located to perform the method according to any one of claims 1 to 8.