Vehicle pre-crash control and severe pre-crash trajectory verification method and apparatus

Abstract

The application relates to the technical field of collision safety, in particular to a vehicle pre-collision control and trajectory verification method and device before a serious collision. The method comprises the following steps: collecting environment information in front of a vehicle through a laser radar sensor, a millimeter wave radar and a forward-looking camera mounted on the vehicle; fusing the environment information collected by different sensors to obtain a final contour of an obstacle; predicting a first collision position, a relative approaching angle and an overlap rate function of the vehicle and the obstacle according to the final contour and a motion state of the vehicle; detecting a first position change trajectory of a contact point in an early contact stage according to a contact sensor mounted on the vehicle; simulating a second position change trajectory of the contact point; and obtaining a safe operation according to a predicted collision type and a collision severity if a difference between the first position change trajectory and the second position change trajectory is less than a set threshold. The application can realize a high-confidence vehicle safety protection scheme.

Description

Technical Field

[0001] This application relates to the field of collision safety technology, and more specifically, to a method and device for vehicle pre-collision control and trajectory verification before a severe collision. Background Technology

[0002] Vehicle safety technologies and systems are evolving from "in-collision protection" to "pre-collision prevention." Pre-collision systems, by activating restraint systems (such as multi-stage airbags and reversible pretensioners) before a collision occurs, aim to buy occupants precious milliseconds of time, significantly reducing the risk of injury. The core bottleneck in the performance of this system lies in the accuracy and reliability of its collision scenario prediction. Existing technologies suffer from three inherent flaws:

[0003] 1. Distortion of the perception model: Mainstream solutions rely on millimeter-wave radar or monocular cameras and use a simple "cuboid bounding box" to abstract the target vehicle. This model is severely distorted and cannot represent the curved features of the vehicle (such as bumper curvature and wheel arches). This results in an extremely high false alarm rate in "near-distance miss" scenarios and a significant underestimation of the overlap rate in oblique collisions, leading to missed detections or inappropriate protection strategy selection.

[0004] 2. Algorithm Limitations: Although some studies have attempted to employ more accurate contour estimation algorithms (such as convex hull, polynomial fitting, and circular arc models), each individual algorithm has its limitations. For example, convex hull algorithms are robust at large angles but lack sufficient accuracy; the three-circular-arc model has extremely high accuracy in head-on scenes but is sensitive to point cloud quality and angles; B-spline or polynomial fitting offers high accuracy but is computationally complex and prone to overfitting. Current technologies lack a mechanism that can adaptively select or fuse the optimal results based on the real-time scene.

[0005] 3. Lack of Verification Mechanism: Existing systems lack the technical means to independently and quickly verify prediction results after a collision. Decision-making relies on the single output of the prediction algorithm; if the prediction is incorrect, it will lead to false triggering. The system usually has to wait for physical contact to occur before a collision is confirmed by traditional collision sensors (accelerometers), wasting the valuable time window of the "pre-collision" phase.

[0006] In view of the above, this application is hereby submitted. Summary of the Invention

[0007] The purpose of this application is to provide a method and device for vehicle pre-collision control and trajectory verification before severe collisions. By using parallel computing and dynamic weighted fusion of multiple heterogeneous contour estimation algorithms, a high-precision target representation can be obtained. Based on this, the trajectory of the contact point can be simulated and predicted before the actual collision of the vehicle. By comparing the trajectory with the collision sensing sensor signal in real time, a high-confidence vehicle safety protection scheme can be achieved.

[0008] To achieve the above objectives, this application adopts the following technical solution:

[0009] Firstly, this application provides a method for vehicle pre-collision control and trajectory verification before a severe collision, including:

[0010] The vehicle collects environmental information in front of it using lidar sensors, millimeter-wave radar, and a forward-facing camera.

[0011] By fusing environmental information collected from different sensors, the final outline of the obstacle is obtained;

[0012] Based on the final contour and the vehicle's motion state, the first collision position, relative approach angle, and overlap rate function between the vehicle and the obstacle are predicted.

[0013] Based on the contact sensors mounted on the vehicle, the trajectory of the initial position change of the contact point in the early stage of contact is detected;

[0014] In the simulation environment, the trajectory of the second position change at the contact point is obtained by simulating the first collision position of the vehicle and the obstacle, the relative approach angle, the overlap rate function and the final contour.

[0015] If the difference between the first position change trajectory and the second position change trajectory is less than a set threshold, then a safe operation is obtained based on the predicted collision type and collision severity, and the safe operation is executed.

[0016] Secondly, this application provides an electronic device, comprising:

[0017] At least one processor, and a memory communicatively connected to at least one of the processors;

[0018] The memory stores instructions that can be executed by at least one of the processors, which enable the at least one processor to perform any of the vehicle pre-collision control and severe collision pre-trajectory verification methods.

[0019] Compared with the prior art, the beneficial effects of this application are as follows:

[0020] In this embodiment, regarding timeliness, the system utilizes trajectory information from the early contact stage (micro-strain phase) for verification and triggering, significantly advancing the decision-making and execution nodes of the safety system from the traditional "after the vehicle body structure has undergone plastic deformation" to "when contact energy has just begun to transfer." This provides a valuable time window of tens of milliseconds for restraint systems such as seatbelt pretensioning and precise airbag deployment. Regarding accuracy and reliability, adaptive weighted fusion of multi-contour estimation algorithms ensures optimal geometric descriptions are obtained under different types of obstacles and complex scenarios, laying a solid foundation for prediction. When the simulated trajectory generated based on the predicted parameters highly matches the early deformation trajectory measured by the contact sensor, a safety operation is ultimately executed. In terms of overall performance optimization, end-to-end collaboration from perception, prediction, verification to execution is achieved, making the "second trajectory" used for final decision-making highly valuable. This deep collaboration enables the entire system to not only act earlier but also more accurately and stably, improving occupant protection while significantly enhancing user trust in advanced active safety features. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0022] Figure 1 This is a flowchart illustrating a method for vehicle pre-collision control and trajectory verification before a severe collision, provided in an embodiment of this application.

[0023] Figure 2 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0024] The following description, in conjunction with the accompanying drawings, illustrates exemplary embodiments of this application, including various details to aid understanding. These should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this application. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0025] Figure 1This is a flowchart illustrating a vehicle pre-collision control and severe collision pre-collision trajectory verification method provided in this application embodiment. This embodiment is applicable to situations where collision information is predicted after the vehicle comes into contact with an obstacle and before a certain deformation (greater than minor deformation) occurs, in order to perform pre-collision control. Existing pre-collision control strategies generally trigger collision safety strategies, such as seat belt pretensioning and airbag pre-inflation, only after sensors at locations such as the B-pillar detect a certain deformation of the vehicle body. However, by the time sensors detect a certain deformation, the vehicle body may have already suffered irreversible damage, and the collision safety strategy cannot guarantee driving safety. This application determines the corresponding pre-collision strategy before the vehicle undergoes a certain deformation, i.e., when minor deformation occurs, to intervene in control as early as possible and ensure driving safety.

[0026] The method provided in this embodiment relies on the following hardware components: a lidar sensor, millimeter-wave radar, a forward-facing camera, vehicle motion state sensors (such as IMU, wheel speed sensors, etc.), contact sensors (such as a high-sensitivity piezoelectric film or micro-strain sensor array arranged under the front bumper or side panel skin of the vehicle), and a pre-collision safety actuator assembly (such as an airbag controller, seat belt pretensioner, etc.). The control unit (ECU) is responsible for running this method.

[0027] See Figure 1 The method provided in this embodiment includes:

[0028] S110 uses lidar sensors, millimeter-wave radar, and forward-facing cameras mounted on the vehicle to collect environmental information in front of the vehicle.

[0029] First, during vehicle operation, raw environmental data in front of the vehicle is simultaneously collected using LiDAR, millimeter-wave radar, and a forward-facing camera. LiDAR provides high-precision point clouds, millimeter-wave radar provides target relative velocity and relative distance, and the forward-facing camera provides rich semantic information.

[0030] S120. The environmental information collected by different sensors is fused to obtain the final outline of the obstacle.

[0031] First, the raw point cloud data acquired by the lidar sensor is preprocessed. Preprocessing operations include: noise reduction, which uses methods such as Statistical Outlier Removal or radius filtering to remove noisy points from the point cloud data; ground point filtering, which uses the Random Sample Consensus (RANSAC) algorithm or plane fitting methods to remove ground point clouds; and clustering segmentation, which uses Euclidean distance-based clustering algorithms (such as DBSCAN) or deep learning-based 3D instance segmentation networks to segment the remaining point cloud into several independent target point cloud clusters, Clusteri(t), where each target point cloud cluster represents an obstacle with a potential collision risk.

[0032] Then, multiple contour estimation algorithms are used to process the preprocessed point cloud data in parallel, yielding various contour estimation results. These algorithms include: convex hull completion, three-circular-arc fitting, and B-spline curve fitting. The convex hull completion algorithm uses computational geometry algorithms (such as Andrew's monotone chain or Grahamscan) to calculate the sequence of 2D convex hull vertices of the point set; and based on the symmetry prior and statistical size model of the vehicle contour (such as the prior distribution of vehicle width and length), it intelligently completes the contour missing due to occlusion through methods such as mirroring and interpolation; that is, it completes the 2D convex hull vertex sequence according to the vehicle size and symmetry. The three-circular-arc fitting algorithm is suitable for fitting contours with smooth curved surface features, such as the front end and bumper of a vehicle, using three tangent circular arcs to approximate the vehicle contour. The B-spline curve fitting algorithm uses non-uniform rational B-splines (NURBS) or cubic B-splines to fit the point cloud. It iteratively optimizes the control points and node vectors to parameterize a smooth, continuous target contour curve. This method can accurately represent complex surfaces.

[0033] In this embodiment, three contour estimation algorithms are executed in parallel for each preprocessed target point cloud cluster to obtain the contour estimation results output by each contour estimation algorithm, i.e., continuous contour curves.

[0034] Next, the multiple contour estimation results are fused to obtain the final contour. This involves the following four steps:

[0035] Step 1: Determine the confidence score for each contour estimation algorithm based on its reconstruction error and scene fit.

[0036] First, the contour estimation results obtained by each contour estimation algorithm are re-discreteed into point sets, and the mean square error between the point sets and the preprocessed point cloud data is calculated. Specifically, for the continuous contour curve estimated by the j-th algorithm, it is re-discreteed into a point set P at fixed arc length intervals. est Calculate the point set P. est Compared with the preprocessed point cloud data P raw Mean square error (MSE) between the nearest points:

[0037] MSE j =mean(min(dist(p est ,P raw ) 2 ));

[0038] Among them, MSE j It is the mean square error of the j-th contour estimation algorithm, dist(p) est ,Praw ) is the point set P est Compared with the preprocessed point cloud data P raw The mean square error is the distance between corresponding points, where min is the minimum value and mean is the average value. The smaller the mean square error, the closer the contour reconstructed by the algorithm is to the original point cloud data.

[0039] Based on the contour estimation results (i.e., the continuous contour curve), the relative distance, relative deflection angle, and point cloud density between the vehicle and the obstacle are determined. Specifically, the distance between the vehicle and the nearest point on the continuous contour curve (from the vehicle itself) is calculated as the relative distance. The azimuth angle of the nearest point on the continuous contour curve relative to the vehicle's direction of travel (usually the vehicle coordinate system + X-axis direction) is the relative deflection angle. The point cloud density reflects the abundance of observational information about the obstacle surface from the contour estimation results; it is obtained by dividing the total number of preprocessed point clouds by the projected area of ​​the continuous contour curve on the horizontal plane.

[0040] The confidence score for each contour estimation algorithm is determined based on the relative distance, relative skew angle, point cloud density, and mean square error.

[0041] To standardize the units, relative distance, relative deflection angle, point cloud density, and mean square error were normalized. Confidence score. c j The goal is to evaluate the contour estimation quality of the j-th algorithm in the current frame and scene in real time. It is a comprehensive score based on multi-dimensional indicators, with values ​​ranging from (0,1). Optionally, the first factor can be obtained from the normalized mean square error using the following formula:

[0042] ;

[0043] in, It is the first factor corresponding to the j-th contour estimation algorithm. It is the attenuation coefficient; the first factor reflects the degree of fit between the contour and the point cloud data. The smaller the value, the higher the first factor.

[0044] relative distance The closer the point cloud, the denser it is, resulting in clearer observations and higher confidence levels. The second factor is calculated based on the relative distance, as shown in the following formula:

[0045] ;

[0046] in, It is the second factor corresponding to the j-th contour estimation algorithm. and These are the calibrated minimum and maximum relative distances, respectively.

[0047] A frontal view (smaller relative deflection angle) yields a more complete profile and higher confidence; a side view (larger relative deflection angle) may result in an incomplete profile and lower confidence. The third factor is calculated based on the relative deflection angle.

[0048] ;

[0049] in, It is the third factor corresponding to the j-th contour estimation algorithm. It is the relative deflection angle corresponding to the j-th contour estimation algorithm.

[0050] Point cloud density directly determines the amount of information; the higher the density, the higher the confidence level. The fourth factor is calculated based on point cloud density.

[0051] ;

[0052] in, It is the fourth factor corresponding to the j-th contour estimation algorithm. It is the point cloud density corresponding to the j-th contour estimation algorithm. The calibrated density threshold.

[0053] The confidence score is obtained by taking a weighted average of the first, second, third, and fourth factors. A higher confidence score indicates a higher reliability of the contour estimation algorithm.

[0054] Then, based on the semantic information from the forward-looking camera, the prior weights for each contour estimation algorithm are obtained. Using the semantic information provided by the camera (such as "car" and "truck" classification labels), a prior weight w is assigned to the corresponding point cloud cluster. type For example, for the target "truck", the prior weights of the convex hull algorithm... w type,ch Higher; for the "car" target, the prior weight w of the three circular arcs or B-spline algorithm is higher. type,3arc ,w type,spline higher.

[0055] Based on the prior weights and confidence scores, the combined weights for each contour estimation algorithm are obtained. See the following formula:

[0056] ;

[0057] in, c j It is the confidence score of the j-th contour estimation algorithm. w j It is the combined weight of the j-th contour estimation algorithm. w type,j It is the prior weight of the j-th contour estimation algorithm.

[0058] The final contour is obtained by weighted summation of multiple contour estimation results using a comprehensive weighting method. Specifically, since the representations (discrete points / continuous parameters) and the number of points of the various contour estimation results are inconsistent, they must first be converted into discrete point sequences with the same number of points. After obtaining M aligned point sequences with the same number of points, a weighted average is performed on the position of each sampling point from 1 to M using a comprehensive weighting method to obtain the final contour.

[0059] S130. Based on the final profile and the vehicle's motion state, predict the first collision position, relative approach angle, and overlap rate function between the vehicle and the obstacle.

[0060] The vehicle's motion state includes the relative velocity, acceleration, and yaw rate between the vehicle (or itself) and the obstacle. Motion prediction algorithms (such as Extended Kalman Filter (EKF), Unscented Kalman Filter (UKF), or physics-based motion models) are used to predict the vehicle's trajectory in the short term (e.g., within 5 seconds). If the obstacle is dynamic, its relative velocity and distance are collected using millimeter-wave radar, and its trajectory is predicted in the short term. A tight bounding box is calculated for the vehicle profile, and the final profile obtained from the previous steps is used. Within the prediction time domain (e.g., within the next 5 seconds), discrete sampling is performed at fixed time steps (e.g., 10 ms) to determine whether the bounding box and the final profile intersect at each moment, the intersection position (i.e., the first collision position), the relative approach angle, and the overlap rate function. The relative approach angle is the angle between the relative motion direction of the vehicle and the obstacle and the normal direction of the obstacle's contact surface at the instant of collision, used to determine the collision type (e.g., frontal, oblique, scrape). The overlap rate function is a function that varies with time t, describing the proportion of overlap between the vehicle and obstacle outlines in the lateral direction before and after a collision.

[0061] S140. Based on the contact sensor mounted on the vehicle, detect the trajectory of the first position change of the contact point in the early stage of contact.

[0062] The contact sensor is a high-frequency dynamic pressure sensor array based on carbon nanotube composite materials, embedded in the inner side of the front bumper skin or the B-pillar of the vehicle. Its sampling frequency is no less than 10kHz, and its response delay is less than 0.1ms. It can output the contact force distribution across the entire bumper surface in real time, and by calculating the centroid of the force distribution, it obtains the trajectory of the contact point's position change in the early stage of contact (referred to as the first position change trajectory). The early stage of contact is defined as the range from the moment of contact to 200ms. The contact point can be the vehicle's B-pillar.

[0063] S150. In the vehicle controller, the second position change trajectory at the contact point is simulated based on the initial collision position of the vehicle and the obstacle, the relative approach angle, the overlap rate function, and the final contour.

[0064] The final outline of the obstacle is treated as a rigid body, and this rigid body is converted into a geometric model and imported into the simulation environment; the vehicle's geometric model is also imported into the simulation environment. Based on the initial collision position and relative approach angle, the geometric models of the obstacle and the vehicle are positioned in the onboard controller. Initial linear and angular velocities provided by the perception system are assigned to the vehicle and obstacle. The overlap rate function describes the continuous change of the lateral overlap ratio over time from before the collision, to the instant of collision, and during the collision process. The collision between the two geometric models and the vehicle's deformation and intrusion are controlled according to the overlap rate function.

[0065] Optionally, the vehicle can be simplified into a model consisting of multiple rigid or flexible components connected by spring-damping units, forming a vehicle structural dynamics model, as shown in the following equation:

[0066] ;

[0067] ;

[0068] Where q(t) is the generalized coordinate vector of the vehicle at time t, and M is the mass matrix of the vehicle. It is the vehicle's damping matrix. It is the vehicle's stiffness matrix. It is the external collision force vector at time t. It is the normal penetration depth at the contact point. It is the rate of change of the normal penetration depth. It is a nonlinear damping coefficient function that simulates energy dissipation during a collision. It is a nonlinear contact stiffness function that simulates the material hardening effect and can be calibrated through material testing. e is the force-displacement exponent, usually 1≤e≤1.5, and for metal skin, e≈1.2.

[0069] It should be noted that in the simulation model, the collision process between the vehicle and the obstacle is simulated using the initial collision position, relative approach angle, overlap rate function, and final profile as constraints, and the trajectory of the second position change at the contact point (i.e., the installation position of the contact sensor) is calculated. The specific simulation process and structural equations can be constructed according to actual needs, and this embodiment does not limit them.

[0070] Solve the vehicle structure dynamics model to obtain the second position change trajectory at the contact point, that is, obtain the second position change trajectory based on the coordinate vector sequence of q(t) at the contact point.

[0071] S160. If the difference between the first position change trajectory and the second position change trajectory is less than a set threshold, then a safe operation is obtained based on the predicted collision type and collision severity, and the safe operation is executed.

[0072] Optionally, the discrete Fraser distance between the first position change trajectory and the second position change trajectory is calculated; if the discrete Fraser distance is less than a set threshold, a safe operation is obtained based on the collision type and collision severity according to the preset rule engine.

[0073] Specifically, if the discrete Friesian distance is less than a set threshold, it indicates that the first position change trajectory has been verified and is credible. Therefore, the collision type (e.g., frontal collision, offset collision, oblique collision, pole collision, etc.) is determined based on the location of the contact point, and the collision severity is determined based on the first position change trajectory; the deeper the intrusion of the first position change trajectory along the vehicle's interior, the more severe the collision. A rule engine is pre-configured to establish the correspondence between collision type, severity, and safety operation. Based on this, a matching process is performed in the rule engine based on the collision type and severity to obtain the safety operation, which is then executed by the pre-collision safety actuator.

[0074] Optional pre-collision safety actuators include: a multi-stage ignition airbag controller, a reversible pretensioning seatbelt motor, an active headrest actuator, and a pedestrian protection hood lift device. The multi-stage ignition airbag controller precisely controls the multi-stage ignition sequence and propellant dosage of the airbag gas generator, enabling "soft" airbag deployment or adjusting the deployment speed and inflation level according to the impact intensity to match the occupant's specific position and posture in the early stages of a collision (the prediction has been verified to be accurate), achieving optimal protection and reducing potential injury from the airbag itself.

[0075] The reversible pretensioning seat belt motor receives commands regarding the magnitude and duration of the pretension force. In the instant after a crash test is passed, the motor drives the retractor to quickly retract a certain length of webbing, eliminating any slack between the seat belt and the occupant's torso. This ensures that the occupant is firmly secured to the seat before the vehicle undergoes significant deformation or violent deceleration, maximizing the restraint effect of the seat belt.

[0076] When the obstacle is a pedestrian, the pedestrian protection hood lift device quickly raises the rear of the hood to a certain height (e.g., 8-10 cm), creating a buffer space between the pedestrian and the hard hood components (such as the camshaft cover), increasing the deformation travel of the head impact area, thereby significantly reducing the risk of head injury to pedestrians.

[0077] Active headrest actuators rapidly propel the headrest forward and upward, shortening the gap between the headrest and the occupant's head before or in the early stages of a rear-end collision, providing timely support for potential whiplash injuries and effectively reducing neck damage.

[0078] This step integrates and controls multiple advanced active safety actuators to collaboratively activate multi-dimensional safety measures, from restraint systems (airbags, seat belts) to external protection (pedestrian protection), for specific, verified collision scenarios, constructing a comprehensive and sophisticated occupant and pedestrian protection network. Because of their extremely early intervention, these actuators have ample time to complete their actions (such as headrest movement or hood lifting) or can activate at the optimal moment (such as multi-stage airbag ignition), significantly improving the vehicle's overall passive safety level.

[0079] In this embodiment, regarding timeliness, the system utilizes trajectory information from the early contact stage (micro-strain phase) for verification and triggering, significantly advancing the decision-making and execution nodes of the safety system from the traditional "after the vehicle body structure has undergone plastic deformation" to "when contact energy has just begun to transfer." This provides a valuable time window of tens of milliseconds for restraint systems such as seatbelt pretensioning and precise airbag deployment. Regarding accuracy and reliability, adaptive weighted fusion of multi-contour estimation algorithms ensures optimal geometric descriptions are obtained under different types of obstacles and complex scenarios, laying a solid foundation for prediction. When the simulated trajectory generated based on the predicted parameters highly matches the early deformation trajectory measured by the contact sensor, a safety operation is ultimately executed. In terms of overall performance optimization, end-to-end collaboration from perception, prediction, verification to execution is achieved, making the "second trajectory" used for final decision-making highly valuable. This deep collaboration enables the entire system to not only act earlier but also more accurately and stably, improving occupant protection while significantly enhancing user trust in advanced active safety features.

[0080] This embodiment provides an electronic device, see [link / reference] Figure 2 It includes at least one processor 301 and a memory 302 communicatively connected to at least one of the processors 301;

[0081] The memory 302 stores instructions that can be executed by at least one of the processors 301, which enable at least one of the processors 301 to perform the above-described vehicle pre-collision control and severe collision trajectory verification method, thus having at least the same advantages as the above-described method.

[0082] Optionally, the electronic device also includes interfaces for connecting the various components, including high-speed interfaces and low-speed interfaces. The components are interconnected using different buses and can be mounted on a common motherboard or otherwise installed as needed. The processor can process instructions executed within the electronic device, including instructions stored in or on memory to display graphical information of a GUI (Graphical User Interface) on an external input / output device (such as a display device coupled to the interface). In other embodiments, multiple processors can be used with multiple memories, and / or multiple buses can be used with multiple memories, if desired. Similarly, multiple electronic devices (e.g., as a server array, a group of blade servers, or a multiprocessor system) can be connected, each providing some of the necessary operations.

[0083] The memory 302, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the vehicle pre-collision control and severe collision pre-trajectory verification method in the embodiments of this application. The processor 301 executes various functional applications and data processing of the device by running the software programs, instructions, and modules stored in the memory 301, thereby realizing the aforementioned vehicle pre-collision control and severe collision pre-trajectory verification method.

[0084] Memory 301 may primarily include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on terminal usage. Furthermore, memory 302 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some instances, memory 302 may further include memory remotely configured relative to the processor, which can be connected to the device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0085] The electronic device may also include an input device 303 and an output device 304. The processor 301, memory 301, input device 303, and output device 304 may be connected via a bus or other means.

[0086] Input device 303 can receive input digital or character information, and output device 304 may include a display device, an auxiliary lighting device (e.g., an LED), and a haptic feedback device (e.g., a vibration motor). The display device may include, but is not limited to, a liquid crystal display (LCD), a light-emitting diode (LED) display, and a plasma display. In some embodiments, the display device may be a touchscreen.

[0087] This embodiment provides a medium storing computer instructions for instructing a computer to perform the methods described above. The computer instructions on this medium, used to instruct the computer to perform the methods described above, thus possess at least the same advantages as the methods described above.

[0088] The medium in this application may be any combination of one or more computer-readable media. The medium may be a computer-readable signal medium or a computer-readable storage medium. The medium may be, for example,, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of the medium (a non-exhaustive list) include: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this document, the medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0089] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of sending, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0090] Program code contained on a computer-readable medium may be transmitted using any suitable medium, including but not limited to wireless, wire, optical fiber, RF (Radio Frequency), or any suitable combination thereof.

[0091] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0092] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this application can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this application can be achieved, and this is not limited herein.

[0093] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for vehicle pre-collision control and trajectory verification before a severe collision, characterized in that, include: The vehicle collects environmental information in front of it using lidar sensors, millimeter-wave radar, and a forward-facing camera. By fusing environmental information collected from different sensors, the final outline of the obstacle is obtained; Based on the final contour and the vehicle's motion state, the first collision position, relative approach angle, and overlap rate function between the vehicle and the obstacle are predicted. Based on the contact sensors mounted on the vehicle, the trajectory of the initial position change of the contact point in the early stage of contact is detected; In the simulation environment, based on the initial collision position, relative approach angle, overlap rate function, and final profile of the vehicle and the obstacle, the trajectory of the second position change at the contact point is simulated. Among them, the relative approach angle is the angle between the relative motion direction of the vehicle and the obstacle and the normal direction of the obstacle contact surface at the moment of collision; the overlap rate function is a function that changes with time t, describing the overlap ratio of the vehicle and obstacle profiles in the lateral direction before and after the collision. If the difference between the first position change trajectory and the second position change trajectory is less than a set threshold, a safety operation is obtained based on the predicted collision type and collision severity, and the safety operation is executed. This includes: calculating the discrete Friesian distance between the first position change trajectory and the second position change trajectory; if the discrete Friesian distance is less than the set threshold, determining the collision type based on the position of the contact point, determining the collision severity based on the first position change trajectory, pre-setting a rule engine, and constructing a correspondence between collision type, collision severity, and safety operation; obtaining a safety operation by matching the collision type and collision severity in the rule engine; the safety operation is operated by a pre-collision safety actuator; the pre-collision safety actuator includes: a multi-stage ignition airbag controller, a reversible pretensioning seat belt motor, an active headrest actuator, and a pedestrian protection hood lifting device.

2. The vehicle pre-collision control and severe collision pre-trajectory verification method according to claim 1, characterized in that, By fusing environmental information collected from different sensors, the final outline of the obstacle is obtained, including: Preprocess the raw point cloud data collected by the lidar sensor; Multiple contour estimation algorithms are used to process the preprocessed point cloud data in parallel to obtain various contour estimation results. The multiple contour estimation results are fused to obtain the final contour.

3. The vehicle pre-collision control and severe collision pre-trajectory verification method according to claim 2, characterized in that, Multiple contour estimation algorithms, including: convex hull completion algorithm, three-circular-arc fitting algorithm and B-spline curve fitting algorithm; The multiple contour estimation results are fused to obtain the final contour, including: Based on the reconstruction error and scene conformity of each contour estimation algorithm, a confidence score for each contour estimation algorithm is determined. Based on the semantic information from the forward-looking camera, the prior weights of each contour estimation algorithm are obtained; Based on the prior weights and confidence scores, the comprehensive weights for each contour estimation algorithm are obtained; The final contour is obtained by weighting and summing the results of various contour estimations using comprehensive weights.

4. The vehicle pre-collision control and severe collision pre-trajectory verification method according to claim 3, characterized in that, Based on the reconstruction error and scene fit of each contour estimation algorithm, a confidence score is determined for each contour estimation algorithm, including: The contour estimation results obtained by each contour estimation algorithm are re-discretized into a point set, and the mean square error between the point set and the preprocessed point cloud data is calculated. Based on the contour estimation results, determine the relative distance, relative deflection angle, and point cloud density between the vehicle and the obstacle; The confidence score for each contour estimation algorithm is determined based on the relative distance, relative skew angle, point cloud density, and mean square error.

5. The vehicle pre-collision control and severe collision pre-trajectory verification method according to claim 4, characterized in that, Based on the prior weights and confidence scores, the comprehensive weights for each contour estimation algorithm are obtained, including: The combined weight of each contour estimation algorithm is calculated using the following formula: ； in, c j It is the confidence score of the j-th contour estimation algorithm. w j It is the combined weight of the j-th contour estimation algorithm. w type,j It is the prior weight of the j-th contour estimation algorithm.

6. An electronic device, characterized in that, include: At least one processor, and a memory communicatively connected to at least one of the processors; The memory stores instructions executable by at least one of the processors, which are executed to enable the at least one processor to perform the vehicle pre-collision control and severe collision trajectory verification method according to any one of claims 1-5.

Images