Vehicle pre-collision control and track verification method and device before serious collision

Abstract

The invention relates to the technical field of collision safety, in particular to a method and equipment for vehicle pre-collision control and track verification before serious collision. The method comprises the following steps: respectively acquiring environmental information in front of a vehicle through a laser radar sensor, a millimeter wave radar and a foresight camera carried on the vehicle; fusing the environment information collected by different sensors to obtain a final contour of the obstacle; according to the final contour and the motion state of the vehicle, predicting a first collision position, a relative approach angle and an overlapping rate function of the vehicle and the obstacle; detecting a first position change track of a contact point in the early stage of contact according to a contact sensor carried on the vehicle; simulating to obtain a second position change track at the contact point; if the difference value between the first position change track and the second position change track is smaller than a set threshold value, safe operation is obtained according to the predicted collision type and collision severity. According to the invention, a high-confidence vehicle safety protection scheme can be realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of collision safety, in particular to a vehicle pre-crash control and trajectory verification method and device before a serious collision. BACKGROUND

[0002] Vehicle safety technology and system are developing from "protecting in collision" to "preventing before collision". Pre-crash system aims to win valuable milliseconds for passengers by starting the restraint system (such as multi-stage airbag, reversible pre-tensioner) in advance before the collision occurs, so as to significantly reduce the risk of injury. The core bottleneck of the performance of the system is the accuracy and reliability of the prediction of the collision scene. There are three inherent defects in the prior art: 1. Perception model distortion: the mainstream scheme relies on millimeter wave radar or monocular camera, and uses 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 radius, wheel arch). This leads to a high false positive rate in the "close miss" scenario, and a serious underestimation of the overlap rate in the oblique collision, resulting in missed reports or inappropriate protection strategy selection.

[0003] 2. Algorithm limitations: although some research attempts to use more accurate contour estimation algorithms (such as convex hull, polynomial fitting, circular arc model), any single algorithm has its applicable boundary. For example, the convex hull algorithm is robust at large angles but lacks precision; the three circular arc model has high precision in head-on scenarios but is sensitive to point cloud quality and angle; B-spline or polynomial fitting has high precision but is computationally complex and prone to overfitting. The prior art lacks a mechanism that can adaptively select or fuse the optimal result according to the real-time scene.

[0004] 3. Lack of verification mechanism: existing systems lack independent and rapid verification of prediction results after a collision occurs. The decision relies on the single output of the prediction algorithm, and once the prediction is wrong, it will lead to false triggering. The system usually waits for physical contact to occur and confirms the collision by traditional collision sensors (acceleration sensors), wasting the valuable time window in the "pre-crash" phase.

[0005] In view of this, the present application is proposed. SUMMARY

[0006] The purpose of the present application is to provide a vehicle pre-crash control and trajectory verification method and device before a serious collision, to obtain high-precision target characterization by parallel computing and dynamically weighted fusion of multiple heterogeneous contour estimation algorithms, and to simulate the predicted contact point trajectory before the actual collision of the vehicle based on this, and to compare the similarity with the collision sensing sensor signal in real time, thereby realizing a high-confidence vehicle safety protection scheme.

[0007] In order to achieve the above object, the following technical solutions are adopted in the present application. In a first aspect, the present application provides a vehicle pre-collision control and severe collision trajectory verification method, comprising: The laser radar sensor, the millimeter wave radar and the forward-looking camera carried on the vehicle are used to collect the environmental information in front of the vehicle respectively; The environmental information collected by different sensors is fused to obtain the final contour of the obstacle; According to the final contour and the motion state of the vehicle, the first collision position, the relative approach angle and the overlap rate function of the vehicle and the obstacle are predicted; According to the contact sensor carried on the vehicle, the first position change trajectory of the contact point in the early contact stage is detected; In the simulation environment, according to the first collision position, the relative approach angle, the overlap rate function and the final contour of the vehicle and the obstacle, the second position change trajectory of the contact point is simulated; If the difference between the first position change trajectory and the second position change trajectory is less than a set threshold, the safe operation is obtained according to the predicted collision type and the collision severity, and the safe operation is executed.

[0008] In a second aspect, the present application provides an electronic device, comprising: At least one processor, and a memory connected in communication with the at least one processor; The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute any of the vehicle pre-collision control and severe collision trajectory verification methods.

[0009] Compared with the prior art, the present application has the following beneficial effects: In the embodiments of the present application, in terms of timeliness, the trajectory information in the early contact (micro-strain stage) is used for verification and triggering, the decision and execution node of the safety system is greatly advanced from the traditional "vehicle body structure has occurred plastic deformation" to "contact energy just starts to transfer", and the valuable time window of tens of milliseconds is obtained for the restraint system such as seat belt pretensioning and airbag accurate point explosion. In terms of accuracy and reliability, through adaptive weighted fusion of multi-outline estimation algorithm, the optimal geometric description can be obtained under different types of obstacles and complex scenes, laying a solid foundation for prediction; when the simulation trajectory generated based on the prediction parameters is highly consistent with the early deformation trajectory measured by the contact sensor, the safety operation is finally executed. In the overall performance optimization, the end-to-end cooperation from perception, prediction, verification to execution is realized, so that the "second trajectory" used for decision-making has very high reference value. This deep cooperation makes the whole system not only act earlier, but also act more accurately and stably, which not only improves the passenger protection effect, but also significantly enhances the user's trust in advanced active safety functions. BRIEF DESCRIPTION OF DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the specific embodiments or prior art of the present application, the drawings needed to be used in the specific embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0011] Figure 1 is a flowchart of a vehicle pre-crash control and severe pre-crash trajectory verification method provided by the embodiments of the present application; Figure 2 is a structural schematic diagram of an electronic device provided by the embodiments of the present application. DETAILED DESCRIPTION

[0012] The exemplary embodiments of the present application are described below in conjunction with the drawings, which include various details of the embodiments of the present application to help understanding, and should be considered only as exemplary. Therefore, those skilled in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present application. Similarly, in order to be clear and concise, the description in the following description omits the description of known functions and structures.

[0013] Figure 1A flowchart of a vehicle pre-crash control and severe crash before trajectory verification method is provided in the embodiment of the present application. The embodiment is applicable to the case where the collision information is predicted after the vehicle contacts the obstacle and before a certain deformation (greater than a slight deformation) occurs, so as to perform pre-crash control. The existing pre-crash control strategy is generally that the sensor at the B-pillar of the vehicle detects a certain deformation of the vehicle body, and then triggers the collision safety strategy, such as safety belt pretensioning, airbag pre-pressurization, etc. However, after the sensor detects the certain deformation of the vehicle body, the vehicle body may have already suffered irreversible damage, and the collision safety strategy cannot guarantee the driving safety. The present application determines the corresponding pre-crash strategy when the vehicle has a slight deformation, so as to intervene in the control as early as possible and guarantee the driving safety.

[0014] The method provided in the embodiment relies on the following hardware components: a laser radar sensor, a millimeter wave radar, a forward-looking camera, a vehicle motion state sensor (such as an IMU, a wheel speed sensor, etc.), a contact sensor (for example, a high-sensitivity piezoelectric film or a micro-strain sensor array arranged under the front bumper or side skin of the vehicle), and a pre-crash safety actuator assembly (such as an airbag controller, a safety belt pretensioner, etc.). A control unit (ECU) is responsible for running the method.

[0015] Referring to Figure 1 The method provided in the embodiment includes the following steps. S110, collecting the environment information in front of the vehicle through the laser radar sensor, the millimeter wave radar and the forward-looking camera mounted on the vehicle.

[0016] Firstly, during the driving of the vehicle, the laser radar, the millimeter wave radar and the forward-looking camera synchronously collect the original environment data in front of the vehicle. The laser radar provides high-precision point cloud, the millimeter wave radar provides the relative speed and relative distance of the target, and the forward-looking camera provides rich semantic information.

[0017] S120, fusing the environment information collected by different sensors to obtain the final contour of the obstacle.

[0018] Firstly, the original point cloud data collected by the laser radar sensor is preprocessed. The preprocessing operation includes: noise reduction, that is, filtering out the noise points of the point cloud data by using a statistical outlier removal or a radius filtering method; ground point filtering, that is, removing the ground point cloud by using a random sample consensus (RANSAC) algorithm or a plane fitting method; clustering segmentation, that is, segmenting the remaining point cloud into a plurality of independent target point cloud clusters Clusteri(t) by using a clustering algorithm based on Euclidean distance (such as DBSCAN) or a three-dimensional instance segmentation network based on deep learning, and each target point cloud cluster represents an obstacle with potential collision risk.

[0019] Then, the preprocessed point cloud data is processed in parallel by using multiple contour estimation algorithms to obtain multiple contour estimation results. The multiple contour estimation algorithms include a convex hull completion algorithm, a three-arc fitting algorithm, and a B-spline curve fitting algorithm. The convex hull completion algorithm calculates a two-dimensional convex hull vertex sequence of the point set by using a computational geometry algorithm (such as Andrew's monotone chain or Graham scan), and intelligently completes the contour that is partially missing due to occlusion by using mirroring and interpolation methods based on the symmetry prior of the vehicle contour and a statistical size model (such as the prior distribution of the vehicle width and length). That is, the two-dimensional convex hull vertex sequence is completed according to the vehicle size and symmetry. The three-arc fitting algorithm is suitable for fitting contours such as the front end of a vehicle and a bumper that have smooth curved surface characteristics, and uses three tangent arcs to approximate the vehicle contour. The B-spline curve fitting algorithm uses non-uniform rational B-spline (NURBS) or cubic B-spline to fit the point cloud. A smooth and continuous target contour curve is parameterized by iteratively optimizing the control points and node vectors. This method can accurately represent complex curved surfaces.

[0020] The embodiment performs the three contour estimation algorithms in parallel for each preprocessed target point cloud cluster to obtain contour estimation results output by each contour estimation algorithm, that is, a continuous contour curve.

[0021] Next, the multiple contour estimation results are fused to obtain a final contour. Specifically, the following four steps are included: First step: Determine the confidence score of each contour estimation algorithm according to the reconstruction error and scene compliance of each contour estimation algorithm.

[0022] First, the contour estimation result obtained by each contour estimation algorithm is discretized into a point set again, and the mean square error between the point set and the preprocessed point cloud data is calculated. Specifically, for the continuous contour curve estimated by the jth algorithm, the point set P est is discretized again at a fixed arc length interval. The mean square error (MSE) between the point set P est and the nearest point in the preprocessed point cloud data P raw is calculated as follows: MSE j = mean(min(dist(p est ,P raw ) 2 )); where MSE j is the mean square error of the jth contour estimation algorithm, and dist(p est ,P raw ) is the distance between the point set P est and the preprocessed point cloud data P rawThe distance between the corresponding points, min is the minimum value, and mean is the average value. The smaller the mean square error, the more the reconstructed contour of the algorithm conforms to the original point cloud data.

[0023] According to the contour estimation result (i.e., the continuous contour curve), the relative distance, the relative angle, and the point cloud density of the vehicle and the obstacle are determined. Specifically, the distance between the nearest point on the continuous contour curve (from the vehicle) and the vehicle is calculated as the relative distance. The azimuth angle of the nearest point on the continuous contour curve relative to the vehicle advancing direction (usually the +X axis direction of the ego vehicle coordinate system) is the relative angle. The point cloud density reflects the observation information abundance of the obstacle surface by the contour estimation result. The point cloud density is obtained by dividing the total number of the preprocessed point clouds by the projection area of the continuous contour curve on the horizontal plane.

[0024] According to the relative distance, the relative angle, the point cloud density, and the mean square error, the confidence score of each contour estimation algorithm is determined.

[0025] In order to unify the dimension, the relative distance, the relative angle, the point cloud density, and the mean square error are normalized. The confidence score c j The confidence score is aimed at evaluating the contour estimation quality of the jth algorithm in the current frame and the current scene in real time, which is a comprehensive score based on multiple indicators, and the value range is (0, 1]. Optionally, the first factor is obtained according to the normalized mean square error by using the following formula: ; wherein, is the first factor corresponding to the jth contour estimation algorithm, is the decay coefficient, and the first factor reflects the fitting degree of the contour and the point cloud data, the smaller the first factor, the higher the first factor.

[0026] The relative distance is closer, the point cloud is usually denser, the observation is clearer, and the confidence is higher. The second factor is calculated according to the relative distance, as shown in the following formula: ; wherein, is the second factor corresponding to the jth contour estimation algorithm, and are the minimum relative distance and the maximum relative distance of the calibration, respectively.

[0027] When the observation is frontal (the relative angle is small), the contour is more complete, and the confidence is high. When the observation is lateral (the relative angle is large), the contour may be incomplete, and the confidence is low. The third factor is calculated according to the relative angle: ; wherein, is the third factor corresponding to the jth contour estimation algorithm, is the relative angle of the jth contour estimation algorithm.

[0028] The point cloud density directly determines the amount of information, the higher the density, the higher the confidence. According to the point cloud density, the fourth factor is calculated: ; wherein, is the fourth factor corresponding to the jth contour estimation algorithm, is the point cloud density corresponding to the jth contour estimation algorithm, is the density threshold value of the calibration.

[0029] The first factor, the second factor, the third factor and the fourth factor are weighted and averaged to obtain a confidence score. The higher the confidence score, the higher the reliability of the contour estimation algorithm.

[0030] Then, according to the semantic information of the front-view camera, the prior weight of each contour estimation algorithm is obtained. The semantic information provided by the camera (such as "car" and "truck" classification labels) is used to give a prior weight w type to the corresponding point cloud cluster. For example, for a "truck" target, the prior weight w w type,ch of the convex hull algorithm is higher; for a "car" target, the prior weight w type,3arc ,w type,spline of the three-arc or B-spline algorithm is higher.

[0031] According to the prior weight and the confidence score, a comprehensive weight of each contour estimation algorithm is obtained. Referring to the following formula: ; wherein, c j is the confidence score of the jth contour estimation algorithm, w j is the comprehensive weight of the jth contour estimation algorithm, w type,j is the prior weight of the jth contour estimation algorithm.

[0032] The comprehensive weight is used to weighted sum the results of multiple contour estimations to obtain the final contour. Specifically, since the representation forms (discrete points / continuous parameters) and the number of points of the multiple contour estimation results are inconsistent, they must be first converted into discrete point sequences with the same number of points. After obtaining the aligned point sequences with the same number of points, the comprehensive weight is used to weighted average the position of each sampling point from 1 to M to obtain the final contour.

[0033] S130, predicting the first collision position, the relative approaching angle and the overlap ratio function between the vehicle and the obstacle according to the final profile and the motion state of the vehicle.

[0034] The motion state of the vehicle includes the relative speed, acceleration and yaw rate between the vehicle (or called ego vehicle) and the obstacle. The motion trajectory of the vehicle in the near future (e.g. 5s) is predicted by a motion prediction algorithm (e.g. extended Kalman filter EKF, unscented Kalman filter UKF or physics-based motion model). If the obstacle is dynamic, the relative speed and distance of the obstacle are collected by the millimeter wave radar, and the motion trajectory of the obstacle in the near future is predicted. A tight oriented bounding box is calculated for the vehicle profile, which is the final profile calculated in the previous step. In the prediction time domain (e.g. in the next 5 seconds), the bounding box and the final profile are discretely sampled at fixed time steps (e.g. 10ms), and it is determined whether the bounding box and the final profile intersect at each time, the intersection position (i.e. the first collision position), the relative approaching angle and the overlap ratio function, etc. The relative approaching angle is the angle between the relative motion direction of the vehicle and the obstacle at the moment of collision and the normal direction of the contact surface of the obstacle, which is used to determine the type of collision (e.g. head-on, oblique angle, scraping). The overlap ratio function is a function of time t, which describes the overlap ratio of the vehicle and the obstacle profile in the lateral direction before and after the collision.

[0035] S140, detecting the first position change trajectory of the contact point in the early contact stage according to the contact sensor mounted on the vehicle.

[0036] The contact sensor is a high-frequency dynamic pressure sensor array based on carbon nanocomposite material, which is embedded in the inner side of the vehicle front bumper skin or on the vehicle B-pillar, with a sampling frequency not less than 10kHz and a response delay less than 0.1ms, which can output the contact force distribution on the entire bumper surface in real time, and obtain the position change trajectory (called first position change trajectory) of the contact point in the early contact stage by calculating the centroid of the force distribution. The early contact stage is defined as the range from the contact to 200ms. The contact point can be the vehicle B-pillar.

[0037] S150, simulating the second position change trajectory of the contact point in the vehicle-mounted controller according to the first collision position, the relative approaching angle, the overlap ratio function and the final profile between the vehicle and the obstacle.

[0038] The final profile of the obstacle is taken as a rigid body, and the rigid body is converted into a geometric model and imported into the simulation environment; the geometric model of the vehicle is also imported into the simulation environment. The geometric models of the obstacle and the vehicle are placed in the vehicle-mounted controller according to the first collision position and the relative approach angle. The initial linear velocity and angular velocity of the vehicle and the obstacle are given by the perception system. The overlap rate function describes the continuous change relationship of the lateral overlap ratio with time from before the collision occurs, the instant the collision occurs to the collision process. The two geometric models are controlled to collide and the vehicle is controlled to deform and intrude according to the overlap rate function.

[0039] Optionally, the vehicle is simplified into a model connected by a plurality of rigid or flexible components through spring-damping units to form a vehicle structure dynamics model, as shown in the following formula: ; ; wherein q(t) is a generalized coordinate vector of the vehicle at time t, M is a mass matrix of the vehicle, is a damping matrix of the vehicle, is a stiffness matrix of the vehicle, is an external collision force vector at time t. is a normal intrusion depth at the contact point, is a change rate of the normal intrusion depth, is a nonlinear damping coefficient function, simulating energy dissipation in the collision. is a nonlinear contact stiffness function, simulating the material hardening effect, which can be calibrated by material testing. e is a force-displacement index, usually 1≤e≤1.5, and for a metal skin, e≈1.2.

[0040] It should be noted that in the simulation model, the collision process of the vehicle and the obstacle is simulated and the second position change trajectory at the contact point (i.e. the installation position of the contact sensor) is calculated under the constraint conditions of the first collision position, the relative approach angle, the overlap rate function and the final profile. The specific simulation process and the structure equation can be constructed according to actual needs, and the present embodiment does not limit this.

[0041] The vehicle structure dynamics model is solved to obtain the second position change trajectory at the contact point, i.e. the second position change trajectory is obtained according to the coordinate vector sequence of q(t) at the contact point.

[0042] S160, if the difference between the first position change trajectory and the second position change trajectory is less than a set threshold, the safety operation is obtained according to the predicted collision type and collision severity, and the safety operation is performed.

[0043] Optionally, the discrete Fréchet distance of the first position change trajectory and the second position change trajectory is calculated; if the discrete Fréchet distance is less than a set threshold, a safe operation is obtained according to the collision type and the collision severity based on a preset rule engine.

[0044] Specifically, if the discrete Fréchet distance is less than the set threshold, it indicates that the first position change trajectory is verified and has credibility. Therefore, the collision type (for example, a head-on collision, a side collision, an oblique collision, a column collision, etc.) is determined according to the position of the contact point, and the collision severity is determined according to the first position change trajectory. The deeper the intrusion of the first position change trajectory along the inside of the vehicle, the more severe the collision. The rule engine is preset to build a corresponding relationship between the collision type, the severity, and the safe operation. Based on this, the safe operation is obtained by matching the collision type and the collision severity in the rule engine, and the safe operation is operated by a pre-crash safety actuator.

[0045] Optionally, the pre-crash safety actuator includes a multi-stage ignition airbag controller, a reversible pre-tensioning seatbelt motor, an active headrest actuator, and a pedestrian protection hood lifting device. The multi-stage ignition airbag controller can accurately control the multi-stage ignition timing and the amount of the airbag gas generator, realize the "soft" deployment of the airbag or adjust the deployment speed and fullness according to the collision intensity, match the specific position and posture of the occupant in the early stage of the collision (verified prediction is correct), realize the best protection effect and reduce the harm caused by the airbag itself.

[0046] The reversible pre-tensioning seatbelt motor receives instructions of the pre-tensioning force size and time. At the moment when the collision is verified, the motor drives the retractor to quickly recover a certain length of the webbing, eliminating the slack between the seatbelt and the occupant's torso, so that the occupant is firmly fixed on the seat before the vehicle body is deformed and decelerated, and the restraining effect of the seatbelt is fully played.

[0047] When the obstacle is a pedestrian, the pedestrian protection hood lifting device quickly lifts the rear part of the hood to a certain height (such as 8-10 cm), creates a buffer space between the pedestrian and the hard engine compartment cover part (such as the camshaft cover), increases the deformation stroke of the head impact area, and thus significantly reduces the risk of head injury to the pedestrian.

[0048] The active headrest actuator quickly pushes the headrest to move forward and upward, shortens the gap between the headrest and the occupant's head before or in the early stage of the rear-end collision, provides timely support for the possible "whiplash injury", and effectively reduces the neck injury.

[0049] The step can control multiple advanced active safety actuators in an integrated manner, can start multiple-dimension safety measures from restraint systems (airbags, seat belts) to external protection (pedestrian protection) in coordination for specific collision scenarios verified, and constructs a stereoscopic and refined occupant and pedestrian protection network. Since the intervention time is very early, these actuators have enough time to complete their actions (such as headrest movement, hood lifting), or can be started at the best time (such as multi-stage airbag ignition), which significantly improves the overall passive safety level of the vehicle.

[0050] In the embodiment of the application, in terms of timeliness, trajectory information in the early contact (micro-strain stage) is used for verification and triggering, the decision and execution nodes of the safety system are greatly advanced from "after the vehicle body structure has occurred plastic deformation" to "when the contact energy starts to transfer", and valuable time window of tens of milliseconds is obtained for the restraint system such as seat belt pretensioning and airbag accurate point explosion. In terms of accuracy and reliability, through adaptive weighted fusion of multiple contour estimation algorithms, the optimal geometric description can be obtained under different types of obstacles and complex scenes, laying a solid foundation for prediction; when the simulation trajectory generated based on the prediction parameters is highly consistent with the early deformation trajectory measured by the contact sensor, the safety operation is finally executed. In terms of overall performance optimization, end-to-end cooperation from perception, prediction, verification to execution is realized, so that the "second trajectory" used for decision making has very high reference value. This deep cooperation enables the entire system to act earlier, more accurately and stably, improves the occupant protection effect, and significantly enhances the user's trust in advanced active safety functions.

[0051] The embodiment provides an electronic device, referring to Figure 2 , comprising at least one processor 301 and a memory 302 connected with the at least one processor 301 in communication; The memory 302 stores instructions executable by the at least one processor 301, and the instructions are executed by the at least one processor 301 to enable the at least one processor 301 to perform the vehicle pre-collision control and severe pre-collision trajectory verification method described above, thus having at least the same advantages as the above method.

[0052] Optionally, an interface can be included to enable the various components to communicate with one another and with other devices or systems. The various components can be mounted on a common motherboard or in other manners as appropriate. In some embodiments, the processor can execute instructions stored in the memory to perform the various functions of the electronic device, including the functions of the vehicle pre-crash control and severe pre-crash trajectory verification method. In some embodiments, multiple processors and / or multiple buses can be employed as appropriate, as will be appreciated by those skilled in the art. Additionally, other components, such as storage devices and cooling devices, can also be used as appropriate. Similarly, the electronic device can be a personal computer, a server, a portable electronic device, or any other device suitable for implementation of the vehicle pre-crash control and severe pre-crash trajectory verification method.

[0053] The memory 302, as a computer readable storage medium, can be used to store software programs, computer executable programs and modules, such as program instructions / modules corresponding to the vehicle pre-crash control and severe pre-crash trajectory verification method in the embodiments of the present application. The processor 301 performs various functional applications and data processing of the device by running the software programs, instructions and modules stored in the memory 301, that is, implements the vehicle pre-crash control and severe pre-crash trajectory verification method described above.

[0054] The memory 301 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system and application programs required by at least one function; the data storage area can store data created according to the use of the terminal, etc. In addition, the memory 302 can include a high-speed random access memory, and can also include a non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other non-volatile solid-state memory device. In some examples, the memory 302 can further include a memory remotely disposed relative to the processor, which can be connected to the device through a network. Examples of the above network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0055] The electronic device can also include an input device 303 and an output device 304. The processor 301, the memory 301, the input device 303 and the output device 304 can be connected through a bus or other means.

[0056] The input device 303 can receive input digital or character information, and the output device 304 can include a display device, an auxiliary lighting device (e.g., an LED), a tactile feedback device (e.g., a vibration motor), and the like. The display device can 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 can be a touch screen.

[0057] The embodiment provides a medium, and the medium stores computer instructions for causing a computer to execute the method described above. The computer instructions on the medium are used to cause the computer to execute the method described above, and thus at least have the same advantages as the method described above.

[0058] The medium in the present application can adopt any combination of one or more computer-readable media. The medium can be a computer-readable signal medium or a computer-readable storage medium. The medium may, for example, be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or apparatus, or any combination of the above. More specific examples (non-exhaustive list) of the medium include an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, the medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, device or apparatus.

[0059] The computer-readable signal medium can include a data signal propagated in a baseband or as a part of a carrier wave, in which a computer-readable program code is borne. Such a propagated data signal can take multiple forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The computer-readable signal medium can also be any computer-readable medium other than the computer-readable storage medium that can send, propagate or transmit a program for use by or in connection with an instruction execution system, device or apparatus.

[0060] The program code contained on the computer-readable medium can be transmitted by any suitable medium, including but not limited to wireless, wire, optical cable, RF (Radio Frequency, radio frequency), and the like, or any suitable combination of the above.

[0061] Computer program code for carrying out operations of the present application can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0062] It should be understood that the steps of the various embodiments recited can be reordered, added to, or removed from the various forms of flowcharts illustrated above. For example, the steps recited in the present application can be performed in parallel, in series, or in a different order, without limitation herein, so long as the desired results of the technology disclosed in the present application are achieved.

[0063] The specific embodiments have been shown and described for purposes of illustrating the embodiments, and not for purposes of limitation. It will be clear to those skilled in the art that various modifications, combinations, sub-combinations and alternatives can be made to the specific embodiments without departing from the spirit and principles of the application. Any further modifications, changes, improvements, combinations, sub-combinations, alternatives, and the like made to the specific embodiments relate, by way of example only, to further implementations of the technology disclosed in the present application.

Claims

1. A method of vehicle pre-crash control and severe pre-crash trajectory verification, characterized by, The application relates to a method for predicting a collision between a vehicle and an obstacle, comprising: collecting environment information in front of the vehicle by means of a laser radar sensor, a millimeter wave radar and a front-view camera mounted on the vehicle; fusing the environment information collected by different sensors to obtain a final contour of the 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 by means of a contact sensor mounted on the vehicle; in a simulation environment, simulating a second position change trajectory of the contact point according to the first collision position, the relative approaching angle, the overlap rate function and the final contour of the vehicle and the obstacle; if a difference between the first position change trajectory and the second position change trajectory is less than a set threshold, obtaining a safe operation according to a predicted collision type and a collision severity, and executing the safe operation.

2. The vehicle pre-crash control and severe pre-crash trajectory verification method according to claim 1, characterized by, The method for fusing the environment information collected by different sensors to obtain the final contour of the obstacle comprises: preprocessing original point cloud data collected by the laser radar sensor; parallelly processing the preprocessed point cloud data by means of multiple contour estimation algorithms to obtain multiple contour estimation results; fusing the multiple contour estimation results to obtain the final contour.

3. The vehicle pre-crash control and severe pre-crash trajectory verification method according to claim 2, characterized by, The multiple contour estimation algorithms comprise a convex hull completion algorithm, a three-arc fitting algorithm and a B-spline curve fitting algorithm. The method for fusing the multiple contour estimation results to obtain the final contour comprises: determining a confidence score of each contour estimation algorithm according to a reconstruction error and a scene compliance degree of the contour estimation algorithm; obtaining a prior weight of each contour estimation algorithm according to semantic information of the front-view camera; obtaining a comprehensive weight of each contour estimation algorithm according to the prior weight and the confidence score; adopting the comprehensive weight to perform weighted summation on the multiple contour estimation results to obtain the final contour.

4. The vehicle pre-crash control and severe pre-crash trajectory verification method according to claim 3, characterized by, The method for determining the confidence score of each contour estimation algorithm according to the reconstruction error and the scene compliance degree of the contour estimation algorithm comprises: re-discretizing the contour estimation result obtained by each contour estimation algorithm into a point set, and calculating a mean square error between the point set and the preprocessed point cloud data; determining a relative distance, a relative deflection angle and a point cloud density of the vehicle and the obstacle according to the contour estimation result; determining the confidence score of each contour estimation algorithm according to the relative distance, the relative deflection angle, the point cloud density and the mean square error.

5. The vehicle pre-crash control and severe pre-crash trajectory verification method according to claim 4, characterized by, The method for obtaining the comprehensive weight of each contour estimation algorithm according to the prior weight and the confidence score comprises: adopting the following formula to calculate the comprehensive weight of each contour estimation algorithm: ； wherein, c j is a confidence score of the jth contour estimation algorithm, w j is a comprehensive weight of the jth contour estimation algorithm, w type,j is a prior weight of the jth contour estimation algorithm.

6. The vehicle pre-crash control and severe pre-crash trajectory verification method according to claim 5, characterized by, if the difference between the first position change trajectory and the second position change trajectory is less than the set threshold, obtaining the safe operation according to the predicted collision type and the collision severity, and executing the safe operation, which comprises: calculating a discrete Fréchet distance of the first position change trajectory and the second position change trajectory; if the discrete Fréchet distance is less than the set threshold, obtaining the safe operation according to the collision type and the collision severity based on a preset rule engine; the safe operation is operated by a pre-collision safety executor.

7. The vehicle pre-crash control and severe pre-crash trajectory verification method according to claim 6, characterized in that, the pre-crash safety actuators include: a multi-stage ignition airbag controller, a reversible pre-tensioning seatbelt motor, an active headrest actuator, and a pedestrian protection hood lifting device.

8. An electronic device, comprising: comprise: at least one processor, and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the vehicle pre-crash control and severe pre-crash trajectory verification method of any one of claims 1-7.

Images