Active collision avoidance and collision mitigation integrated control method, system, device and medium

Abstract

The application provides a kind of active collision avoidance and collision mitigation integrated control method, system, equipment and medium, belong to vehicle technical field, this method includes obtaining the fusion state of multiple source sensors of ego vehicle and obstacle vehicle and prediction sequence, to construct vehicle pre-collision overlap domain analysis model;The intersection relationship between candidate trajectory and obstacle vehicle is analyzed based on vehicle pre-collision overlap domain analysis model, to obtain completely avoidable set and existing intersection set;According to the set state of completely avoidable set and existing intersection set, determine target trajectory;Integrated model predictive control is carried out based on target trajectory, and the steering and longitudinal acceleration control sequence of ego vehicle is output.The application constructs pre-collision overlap domain analysis model, improves the accuracy and integrity of pre-collision risk analysis, introduces the characteristics of pre-collision overlap domain analysis into candidate trajectory screening, so that the target trajectory represents the potential collision relationship and collision contact characteristics, and makes up for the lack of collision contact geometric feature constraints in trajectory planning.

Description

Technical Field

[0001] This invention relates to the field of vehicle technology, and in particular to an integrated control method, system, device and medium for active collision avoidance and collision mitigation. Background Technology

[0002] With the continuous development of intelligent driving technology, intelligent driving vehicles can now assist drivers to a certain extent in tasks such as lane keeping, emergency braking, and obstacle avoidance, providing technical support for improving road traffic safety. However, with the continuous increase in vehicle ownership and the increasingly complex road traffic environment, traffic accidents still occur frequently. In extreme scenarios, vehicles often face extremely short reaction times and limited maneuverability; improper handling may result in vehicle damage, occupant injury, or even more serious consequences.

[0003] Active collision avoidance systems (such as Automatic Emergency Braking, AEB) and collision mitigation systems are typically implemented as independent functional modules, deployed on different onboard controllers or by separate control units. They aim to prevent the vehicle from contacting obstacles before a collision occurs, thereby reducing the probability of an accident; or to minimize the intensity and consequences of a collision by controlling vehicle speed, attitude, or relative contact state, thus improving the safety of the vehicle and its occupants. However, active safety systems implemented by distributed or independent controllers still have shortcomings. Summary of the Invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art, and proposes an integrated control method, system, device and medium for active collision avoidance and collision mitigation.

[0005] In a first aspect, embodiments of the present invention provide an integrated control method for active collision avoidance and collision mitigation, comprising:

[0006] The multi-source sensor fusion states and prediction sequences of the vehicle and the obstacle vehicle are obtained to construct a vehicle pre-collision overlap domain analysis model;

[0007] Based on the vehicle pre-collision overlap domain analysis model, the overlap relationship between candidate trajectories and obstacle vehicles is analyzed to obtain a completely avoidable set and a set with overlap.

[0008] The target trajectory is determined based on the set state of the completely avoidable set and the set that overlaps with it;

[0009] Based on the target trajectory, an integrated model predictive control is performed, outputting the vehicle's steering and longitudinal acceleration control sequences.

[0010] In some embodiments, acquiring the multi-source sensor fusion states and prediction sequences of the vehicle and the obstacle vehicle to construct a vehicle pre-collision overlap domain analysis model includes:

[0011] Acquire the state information of the vehicle and the obstacle vehicle from the multi-source sensor fusion output, as well as the predicted state sequence of the obstacle vehicle in the prediction time domain;

[0012] Based on the state information of the autonomous vehicle and the obstacle vehicle and the predicted state sequence, the occupancy polygons of the autonomous vehicle and the obstacle vehicle at each prediction step length are constructed respectively.

[0013] Based on the polygon intersection clipping operation, the overlapping region polygons and the area of ​​the overlapping region are calculated according to the occupied polygons, and the time step when the occupied regions first overlap is determined according to the area of ​​the overlapping region.

[0014] The first contact time, first contact point, and first contact angle are calculated based on the time step in which the occupied domains first overlap.

[0015] Based on the occupied polygon, overlapping region polygon, overlapping region area, time step of the first overlap, first contact time, first contact point, and first contact angle, a vehicle pre-collision overlapping region analysis model is constructed.

[0016] In some embodiments, calculating the first contact time, first contact point, and first contact angle based on the time step in which the occupied domains first overlap includes:

[0017] The sampling interval corresponding to the time step in which the occupied domains first overlap is taken as the preset sampling interval;

[0018] Within the preset sampling interval, continuous-time interpolation is performed on the poses of the vehicle and the obstacle vehicle to obtain the continuous-time occupied region polygon of the vehicle and the obstacle vehicle.

[0019] The minimum distance between the boundaries is defined based on the polygonal domain occupied by the autonomous vehicle and the obstacle vehicle during continuous time.

[0020] The minimum moment within the preset sampling interval when the minimum distance at the boundary first falls below the preset threshold is defined as the first contact moment.

[0021] The first contact point and the first contact angle are calculated based on the first contact time.

[0022] In some embodiments, the analysis of the overlap relationship between candidate trajectories and obstacle vehicles based on the vehicle pre-collision overlap domain analysis model to obtain a completely avoidable set and a set with overlap includes:

[0023] A candidate trajectory set is constructed based on the horizontal decision set, the vertical decision set, and the time series set.

[0024] Generate a discrete state sequence of each candidate trajectory in the prediction time domain based on the candidate trajectory set;

[0025] Based on the vehicle pre-collision overlap domain analysis model and discrete state sequence, the overlap relationship between each candidate trajectory and the obstacle vehicle is calculated.

[0026] Based on the overlap relationship, the candidate trajectory set is divided into a completely avoidable set and a set with overlap.

[0027] In some embodiments, determining the target trajectory based on the set state of the completely avoidable set and the set with overlapping elements includes:

[0028] When the set state of the completely avoidable set is not empty, trajectory filtering is performed within the completely avoidable set to determine the target trajectory;

[0029] When the set of completely avoidable sets is empty, a comprehensive screening index is constructed within the set of overlapping sets, and a target trajectory is selected based on the comprehensive screening index.

[0030] In some embodiments, the integrated model predictive control based on the target trajectory, outputting a steering and longitudinal acceleration control sequence for the vehicle, includes:

[0031] Fuzzy reasoning is performed based on the target trajectory to obtain the three-state membership degree;

[0032] An integrated control optimization model is constructed based on the target trajectory;

[0033] The weights of each objective item in the integrated control optimization model are scheduled according to the three-state membership degree, and the steering control sequence and longitudinal acceleration control sequence of the vehicle are output.

[0034] In some embodiments, the step of performing fuzzy reasoning based on the target trajectory to obtain the three-state membership degree includes:

[0035] Based on the target trajectory and the overlap region area sequence of the target trajectory, input quantities are constructed; wherein, the input quantities include a temporal proximity representation quantity, an overlap intensity representation quantity, and an overlap persistence representation quantity;

[0036] The input quantity is fuzzified based on a preset membership function and a fuzzy rule base to obtain the fuzzy inference result.

[0037] The fuzzy reasoning results are normalized to obtain a three-state membership degree; wherein the three-state membership degree includes an avoidable state, a transitional state, and a pre-collision state.

[0038] Secondly, embodiments of the present invention provide an integrated active collision avoidance and collision mitigation control system, comprising:

[0039] The pre-collision overlap domain analysis module is used to obtain the multi-source sensor fusion state and prediction sequence of the vehicle and the obstacle vehicle in order to build a vehicle pre-collision overlap domain analysis model.

[0040] The set partitioning module is used to analyze the overlap relationship between candidate trajectories and obstacle vehicles based on the vehicle pre-collision overlap domain analysis model, and obtain the completely avoidable set and the overlapping set.

[0041] The target trajectory filtering module is used to determine the target trajectory based on the set state of the completely avoidable set and the set with overlapping elements;

[0042] An integrated predictive control module is used to perform integrated model predictive control based on the target trajectory, and output the vehicle's steering and longitudinal acceleration control sequences.

[0043] Thirdly, embodiments of the present invention provide an electronic device, including:

[0044] One or more processors;

[0045] Memory, used to store one or more programs;

[0046] When the one or more programs are executed by the one or more processors, the one or more processors implement any of the methods described above.

[0047] Fourthly, embodiments of the present invention provide a computer-readable medium on which a computer program is stored, the computer program being executed by a processor to implement the steps of any of the methods described above.

[0048] The present invention provides an integrated active collision avoidance and collision mitigation control method, comprising: acquiring the multi-source sensor fusion state and prediction sequence of the vehicle and the obstacle vehicle to construct a vehicle pre-collision overlap domain analysis model; analyzing the overlap relationship between candidate trajectories and obstacle vehicles based on the vehicle pre-collision overlap domain analysis model to obtain a fully avoidable set and a set with overlap; determining the target trajectory based on the set state of the fully avoidable set and the set with overlap; and performing integrated model predictive control based on the target trajectory to output the steering and longitudinal acceleration control sequences of the vehicle. The present invention, by constructing a pre-collision overlap domain analysis model, integrates the vehicle and obstacle vehicles into a unified analysis framework, improving the accuracy and completeness of pre-collision risk analysis and providing a unified geometric analysis basis for subsequent trajectory decision-making and control. Introducing the features obtained from the pre-collision overlap domain analysis into the candidate trajectory screening process enables the target trajectory to characterize potential collision relationships and collision contact features, compensating for the lack of collision contact geometric feature constraints in trajectory planning. Attached Figure Description

[0049] Figure 1A flowchart illustrating an integrated active collision avoidance and collision mitigation control method provided in an embodiment of the present invention;

[0050] Figure 2 This is a flowchart illustrating an optional specific implementation method involved in an embodiment of the present invention;

[0051] Figure 3 This is a schematic diagram of the pre-collision overlap domain analysis in an embodiment of the present invention;

[0052] Figure 4 This is a schematic diagram of the process involving target trajectory filtering in an embodiment of the present invention;

[0053] Figure 5 This is a schematic diagram of the integrated model predictive control process in an embodiment of the present invention;

[0054] Figure 6 This is a structural block diagram of an integrated active collision avoidance and collision mitigation control system provided in an embodiment of the present invention;

[0055] Figure 7 This is a structural block diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0056] To enable those skilled in the art to better understand the technical solutions of the present invention, exemplary embodiments of the present invention are described below in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to aid understanding. These should be considered merely 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 invention. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0057] Where there is no conflict, the various embodiments of the present invention and the features thereof may be combined with each other.

[0058] As used herein, the term “and / or” includes any and all combinations of one or more related enumerated entries.

[0059] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used herein, the singular forms “a” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will also be understood that when the terms “comprising” and / or “made of” are used in this specification, the presence of the stated feature, integral, step, operation, element, and / or component is specified, but the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or groups thereof is not excluded. Terms such as “connected” or “linked” are not limited to physical or mechanical connections but can include electrical connections, whether direct or indirect.

[0060] Unless otherwise specified, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art. It will also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having the meaning consistent with their meaning in the context of the relevant art and the invention, and will not be interpreted as having an idealized or overly formal meaning unless expressly so defined herein.

[0061] In the technical solution of this invention, the collection, storage, use, processing, transmission, provision, and disclosure of user personal information all comply with relevant laws and regulations and do not violate public order and good morals. The use of user data in this technical solution follows relevant national laws and regulations (e.g., the "Information Security Technology - Personal Information Security Specification"). For example: appropriate measures are taken for personal information access control; restrictions are imposed on the display of personal information; the purpose of using personal information does not exceed the scope of direct or reasonable association; and explicit identity targeting is eliminated when using personal information to avoid precisely locating a specific individual.

[0062] The key terms involved in this invention are defined as follows:

[0063] Pre-collision overlap region: The intersection region of the polygons occupied by the vehicle and the obstacle vehicle in the pre-collision scenario, used to quantitatively characterize the spatial encroachment relationship;

[0064] Occupied polygon: An ordered sequence of vertices occupying vehicle space, constructed based on vehicle pose and geometric dimensions;

[0065] Three-state membership: fuzzy inference membership of avoidable state, transition state, and pre-collision state, used for dynamic scheduling control target weights;

[0066] MPC: Model Predictive Control, used for joint optimization of trajectory tracking, active collision avoidance, and collision mitigation;

[0067] Integrated control: Active collision avoidance and collision mitigation are incorporated into a unified control framework to achieve smooth switching of control objectives;

[0068] First contact characteristics: a collective term for the time of first contact, the point of first contact in the vehicle body coordinate system, and the angle of first contact;

[0069] Intelligent Driving Domain Controller: The physical carrier of this invention, the central decision-making and control unit of intelligent driving vehicles, outputs smooth control commands that can continuously evolve with the risk through online geometric analysis, risk state fuzzy decision-making and model predictive control weight dynamic scheduling.

[0070] In related technologies, active safety systems implemented by distributed or independent controllers still have shortcomings. In some extreme scenarios, active collision avoidance systems often fail to completely prevent collisions. At the same time, collision mitigation systems are usually designed and operated independently from active collision avoidance systems. When a vehicle transitions from an "avoidable" state to an "unavoidable" state, there is often a lack of a unified coordination mechanism between the two types of systems. This can easily lead to problems such as discontinuous switching of control objectives, mutual interference in decisions, and even control conflicts, affecting the overall safety control effect.

[0071] For example, one technical solution discloses a path planning method, device, electronic equipment, and autonomous vehicle to at least address the problem of low prediction accuracy in predicting vehicle trajectories in related technologies. The specific implementation scheme is as follows: While the main vehicle is traveling along its current trajectory, if an obstacle vehicle is detected, the system predicts the obstacle vehicle's entry point into the current trajectory, as well as the entry time at that point; based on the predicted entry point and time, it generates the main vehicle's current predicted travel information; it fuses the current predicted travel information with travel prediction information accumulated over a historical time range to obtain fused travel information; based on the fused travel information, it determines the target entry distance of the obstacle vehicle; and it updates the main vehicle's current trajectory using the target entry distance.

[0072] This scheme plans collision avoidance paths based on obstacle prediction trajectories and uses a safety and comfort-weighted trajectories for selection. However, it fails to calculate the pre-collision overlap region and first-contact geometric features, lacks collision mitigation guidance trajectory selection logic, and lacks fuzzy risk classification and MPC weight scheduling mechanisms. Furthermore, the trajectory planning in this scheme does not incorporate collision geometric constraints, making it unable to handle unavoidable scenarios. It lacks a three-state classification of pre-collision risk, and the control objective cannot be dynamically adjusted according to risk. It also fails to achieve integrated joint optimization of trajectory decision-making and tracking control.

[0073] To address at least one of the technical problems existing in the aforementioned related technologies, the present invention provides an integrated control method for active collision avoidance and collision mitigation. Figure 1 This is a flowchart illustrating an integrated active collision avoidance and collision mitigation control method provided in an embodiment of the present invention.

[0074] As one embodiment of the present invention, such as Figure 1As shown, the integrated active collision avoidance and collision mitigation control method includes:

[0075] Step S100: Obtain the multi-source sensor fusion state and prediction sequence of the vehicle and the obstacle vehicle to construct a vehicle pre-collision overlap domain analysis model;

[0076] Step S200: Based on the vehicle pre-collision overlap domain analysis model, analyze the overlap relationship between the candidate trajectory and the obstacle vehicle to obtain the completely avoidable set and the overlapping set;

[0077] Step S300: Determine the target trajectory based on the set state of the completely avoidable set and the set with overlapping elements;

[0078] Step S400: Perform integrated model predictive control based on the target trajectory, and output the vehicle's steering and longitudinal acceleration control sequence.

[0079] It should be noted that the execution subject in this embodiment can be an electronic device, which can be a computer device with data processing function, or other devices that can achieve the same or similar functions. This embodiment does not limit this. In this embodiment, the execution subject is a computer device, such as an intelligent driving domain controller, as an example for explanation.

[0080] Specifically, the integrated active collision avoidance and collision mitigation control method proposed in this embodiment can be integrated into the intelligent driving domain controller and the integrated safety control scheme that integrates active collision avoidance and collision mitigation functions. This enables the vehicle to coordinate and continuously adjust active collision avoidance and collision mitigation in a unified manner according to changes in the scenario and the degree of risk as the pre-collision risk continues to evolve. This allows the vehicle to avoid collisions as much as possible, and further mitigate the consequences of collisions when they are difficult to completely avoid, thereby improving the overall safety of intelligent driving vehicles in complex traffic scenarios.

[0081] In some embodiments, the multi-source sensor fusion states and prediction sequences of the vehicle and the obstacle vehicle are obtained to construct a vehicle pre-collision overlap domain analysis model, including: obtaining the state information of the vehicle and the obstacle vehicle output by multi-source sensor fusion and the prediction state sequence of the obstacle vehicle in the prediction time domain; constructing the occupancy polygons of the vehicle and the obstacle vehicle at each prediction step size based on the state information of the vehicle and the obstacle vehicle and the prediction state sequence; calculating the overlap domain polygon and the overlap domain area based on the occupancy polygon based on the polygon intersection clipping operation, and determining the time step of the first overlap of the occupancy domains based on the overlap domain area; calculating the first contact time, the first contact point and the first contact angle based on the time step of the first overlap of the occupancy domains; and constructing a vehicle pre-collision overlap domain analysis model based on the occupancy polygon, the overlap domain polygon, the overlap domain area, the time step of the first overlap, the first contact time, the first contact point and the first contact angle.

[0082] In some embodiments, calculating the first contact time, first contact point, and first contact angle based on the time step in which the occupancy domains first overlap includes: using the sampling interval corresponding to the time step in which the occupancy domains first overlap as a preset sampling interval; performing continuous-time interpolation on the poses of the vehicle and the obstacle vehicle within the preset sampling interval to obtain continuous-time occupancy domain polygons of the vehicle and the obstacle vehicle; defining a minimum boundary distance based on the continuous-time occupancy domain polygons of the vehicle and the obstacle vehicle; defining the minimum time in the preset sampling interval when the minimum boundary distance is first less than a preset threshold as the first contact time; and calculating the first contact point and the first contact angle based on the first contact time.

[0083] Specifically, such as Figure 2 As shown, step S1 is the pre-collision overlap domain analysis: obtain the state information of the self-vehicle and obstacle vehicle output by multi-source sensor fusion, as well as the predicted state sequence of the obstacle vehicle in the prediction time domain, and construct the occupancy polygons of the self-vehicle and obstacle vehicle at each prediction step; calculate the overlap domain polygons and overlap domain area based on polygon intersection clipping, and determine the time step at which the occupancy domains first overlap; further calculate the first contact time, the first contact point and the first contact angle in the vehicle coordinate system in the corresponding sampling interval, and construct the vehicle pre-collision overlap domain analysis model.

[0084] For example, in step S1, the occupied polygons of the vehicle and the obstacle vehicle at the k-th prediction step are represented by an ordered sequence of vertices:

[0085]

[0086] Among them, V e (k) represents the sequence of polygon vertices occupied by the vehicle, V o (k) represents the sequence of polygon vertices occupied by the obstacle vehicle. and Let n represent the coordinates of the corresponding vertices in a unified coordinate system. e and n o These represent the number of vertices of the polygon occupied by the vehicle and the obstacle vehicle, respectively.

[0087] For example, in step S1, the overlapping region polygons are obtained by polygon intersection clipping:

[0088]

[0089] The area of ​​the overlapping region is obtained by solving the following equation:

[0090]

[0091] Among them, V c(k) represents the sequence of vertices of the overlapping polygon of the vehicle and the obstacle vehicle at the k-th prediction step, and A(k) represents the area of ​​the corresponding overlapping region. c This represents the number of vertices in the overlapping polygon.

[0092] For example, in step S1, the time step in which the occupied domains first overlap is defined as:

[0093]

[0094] And in the interval The boundary of the occupied domain between the vehicle and the obstacle vehicle is obtained by constructing the continuous-time occupied domain constructor:

[0095]

[0096] Among them, Ω e (t) and Ω o (t) represent the constructors that map the vehicle pose parameters to the vehicle occupancy domain boundary, respectively. e and χ o These represent the geometric dimensions of the vehicle and the obstacle vehicle, respectively.

[0097] First contact moment The minimum interval distance d(t) between vehicle boundaries in continuous time is defined as:

[0098]

[0099] Where, ε d This is the threshold for determining the first contact.

[0100] For example, in step S1, the initial contact point and initial contact angle in the vehicle body coordinate system are determined as follows:

[0101] At the moment of first contact Establish a vehicle body coordinate system, and transform the boundaries of the occupied areas of the vehicle and the obstacle vehicle to the vehicle body coordinate system respectively, denoted as . and .

[0102] The first contact point from the vehicle side is defined as:

[0103]

[0104] The initial contact angle is defined as:

[0105]

[0106] in, θ represents the first point of contact from the vehicle side in the vehicle coordinate system. cThe first contact angle is represented by wrap(·), which represents the angle normalization function.

[0107] In this embodiment, by constructing and running a pre-collision overlap domain analysis model, the spatial encroachment relationship between the vehicle and the obstacle vehicle in the prediction time domain, the first overlap time, and the geometric features of the first contact are uniformly incorporated into the analysis framework. This achieves continuous characterization of key geometric information of the pre-collision process, overcoming the shortcomings of related technologies that rely solely on relative distance or a single risk indicator and are difficult to accurately characterize complex pre-collision relationships. This improves the accuracy and completeness of pre-collision risk analysis at the controller level.

[0108] In some embodiments, the overlap relationship between candidate trajectories and obstacle vehicles is analyzed based on the vehicle pre-collision overlap domain analysis model to obtain a completely avoidable set and a set with overlap, including: constructing a candidate trajectory set based on a lateral decision set, a longitudinal decision set, and a temporal set; generating a discrete state sequence of each candidate trajectory in the prediction time domain based on the candidate trajectory set; calculating the overlap relationship between each candidate trajectory and obstacle vehicles based on the vehicle pre-collision overlap domain analysis model and the discrete state sequence; and dividing the candidate trajectory set into a completely avoidable set and a set with overlap based on the overlap relationship.

[0109] In some embodiments, determining the target trajectory based on the set state of the completely avoidable set and the overlapping set includes: when the set state of the completely avoidable set is not empty, performing trajectory filtering within the completely avoidable set to determine the target trajectory; when the set state of the completely avoidable set is empty, constructing a comprehensive filtering index within the overlapping set, and selecting the target trajectory based on the comprehensive filtering index.

[0110] Specifically, such as Figure 2 As shown, the S2 target trajectory selection is as follows: a candidate trajectory set is constructed based on the horizontal decision set, the vertical decision set, and the time series set, generating a discrete state sequence of each candidate trajectory in the prediction time domain; combined with the vehicle pre-collision overlap domain analysis model constructed by the pre-collision overlap domain analysis operation, the overlap relationship between each candidate trajectory and the obstacle vehicle is calculated, and the set of completely avoidable obstacles and the set of overlapping obstacles are divided accordingly; according to the set state, the trajectory selection strategy of active collision avoidance guidance or collision mitigation guidance is selected respectively to determine the target trajectory.

[0111] For example, in step S2, the candidate trajectory set is defined as:

[0112]

[0113] in, For a horizontal decision set, For vertical decision sets, It is a time series set.

[0114] Any candidate trajectory τ j Discrete representation in the prediction time domain:

[0115]

[0116] Among them, (x ej (k),y ej (k) represents the vehicle position corresponding to the j-th candidate trajectory at the k-th prediction step, ψ ej (k) represents the corresponding heading angle, v ej (k) represents the corresponding velocity.

[0117] For example, in step S2, for any candidate trajectory τ j Calculate the sequence of vertices V of the polygon overlapping with the obstacle vehicle. cj (k) and the overlapping area A j (k), and divide the completely avoidable set and the existentially overlapping set according to the following formula:

[0118]

[0119] in, Represents a completely avoidable set. This indicates the existence of an overlapping set. For candidate trajectories belonging to the overlapping set, the first contact time, the first contact point in the vehicle coordinate system, and the first contact angle are further calculated within the corresponding sampling interval.

[0120] For example, in step S2, when the completely avoidable set is not empty, a comprehensive cost function is constructed based on the comfort cost and the stability cost:

[0121]

[0122] Among them, J comf (τ j J represents the cost of comfort. stab (τ j ) represents the stability cost, w comf and w stab These are the corresponding weight parameters.

[0123] The cost of comfort is defined as:

[0124]

[0125] in, This represents the longitudinal acceleration of the j-th candidate trajectory at the k-th prediction step. Let Δa represent the lateral acceleration of the j-th candidate trajectory at the k-th prediction step. j(k) represents the longitudinal acceleration change between adjacent prediction steps, and α1, α2 and α3 are the weight parameters of the longitudinal acceleration, lateral acceleration and longitudinal acceleration change cost terms, respectively.

[0126] The stability cost is defined as:

[0127]

[0128] in, This represents the heading angle deviation of the j-th candidate trajectory at the k-th prediction step. This represents the corresponding yaw rate. This represents the corresponding front wheel steering angle or equivalent steering control quantity; β1, β2, and β3 are the weight parameters for heading angle deviation, yaw rate, and steering input cost, respectively.

[0129] For example, in step S2, when the set of completely avoidable collisions is empty, a comprehensive cost function is constructed based on the collision probability cost and the collision severity cost:

[0130]

[0131] Among them, J col (τ j ) represents the comprehensive cost of the j-th candidate trajectory in the overlapping set; J prob (τ j ) represents the cost of a collision probability; J sev (τ j ) indicates the cost of collision severity; w prob and w sev These represent the weighting coefficients for the corresponding collision probability and collision severity cost terms, respectively.

[0132] The collision probability cost is defined as:

[0133]

[0134] Among them, g1, g2, g3, and g4 are weight parameters in the collision probability cost.

[0135] The collision severity cost is defined as:

[0136]

[0137] Where, ΔV j J represents the speed difference between the vehicle and the obstacle vehicle at the moment of the first overlap. p (τ j ) represents the collision point preference cost, J θ (τ j ) represents the collision angle preference cost term, v ej (kcj ), and θ cj This indicates that the j-th candidate trajectory occurs at consecutive time k. cj The velocity, point of impact, and angle of impact at time, v o (k cj ) represents the obstacle vehicle at continuous time k cj The speed of time, and θ ref The desired collision point and desired collision angle are preset respectively; h1, h2 and h3 represent the weight parameters of the cost terms of velocity difference, contact point preference and contact angle preference, respectively.

[0138] In this embodiment, by incorporating geometric features such as the overlap relationship, first contact point, and first contact angle obtained from the pre-collision overlap domain analysis into the candidate trajectory screening process, the target trajectory can characterize the potential collision relationship and collision contact features. Especially in scenarios where collisions are difficult to completely avoid, it can provide a basis for trajectory decision-making guided by collision damage mitigation, making up for the lack of collision contact geometric feature constraints in trajectory planning in related technologies.

[0139] In some embodiments, integrated model predictive control is performed based on the target trajectory to output the vehicle's steering and longitudinal acceleration control sequences, including: performing fuzzy inference based on the target trajectory to obtain three-state membership degrees; constructing an integrated control optimization model based on the target trajectory; scheduling the weights of each objective item in the integrated control optimization model according to the three-state membership degrees, and outputting the vehicle's steering control sequence and longitudinal acceleration control sequence.

[0140] In some embodiments, fuzzy reasoning is performed based on the target trajectory to obtain a three-state membership degree, including: constructing an input quantity based on the target trajectory and the overlap region area sequence of the target trajectory; wherein the input quantity includes a temporal proximity representation quantity, an overlap intensity representation quantity, and an overlap persistence representation quantity; fuzzifying the input quantity based on a preset membership function and a fuzzy rule base to obtain a fuzzy reasoning result; and normalizing the fuzzy reasoning result to obtain a three-state membership degree; wherein the three-state membership degree includes an avoidable state, a transitional state, and a pre-collision state.

[0141] Specifically, such as Figure 2As shown, step S3, integrated model predictive control, involves: extracting the evolution features of the overlap domain based on the target trajectory determined by the target trajectory screening operation, and constructing representations of time proximity, overlap intensity, and overlap persistence; performing fuzzy inference based on preset membership functions and fuzzy rule bases to obtain the three-state membership degrees of avoidable state, transition state, and pre-collision state; and constructing an integrated control optimization model including trajectory tracking, active collision avoidance, collision mitigation, and control regularization terms based on the target trajectory and the three-state membership degrees. The weights of each target term are then scheduled according to the three-state membership degrees, and the steering control sequence and longitudinal acceleration control sequence of the vehicle are output.

[0142] For example, in step S3, the time proximity, overlap intensity, and overlap duration characteristics are defined as follows:

[0143]

[0144] in, This indicates the first overlap step size corresponding to the target trajectory. Let ε represent the sequence of overlapping area regions corresponding to the target trajectory, where ε is a very small positive number and 1(·) is an indicator function.

[0145] The three-state membership degree is obtained by normalizing the fuzzy inference results:

[0146]

[0147] Where μ1, μ2, and μ3 represent the membership degrees of the avoidable state, the transition state, and the pre-collision state, respectively.

[0148] For example, the objective function of the integrated control optimization model is defined as:

[0149]

[0150] Among them, J trk (k) represents the trajectory tracking term, J avd (k) represents the active collision avoidance term, J mit (k) represents the collision mitigation term, J reg (k) represents the control regularization term, which is defined as follows:

[0151]

[0152] Among them, e y (k), e ψ (k) and e v (k) represent the lateral position error, heading error, and velocity error at the k-th prediction step, respectively, d eo (k) represents the minimum boundary distance between the occupancy region of the target vehicle and the occupancy region of the obstacle vehicle at the k-th prediction step, ε aLet ΔV(k) be a very small positive number, representing the speed difference between the vehicle and the obstacle vehicle at the k-th prediction step. p (k) represents the contact point preference term, J θ (k) represents the contact angle preference term, u(k) represents the control input at the k-th control step, and Δu(k) represents the change in control input between adjacent steps.

[0153] In this embodiment, by implementing three-state fuzzy membership degree and model predictive control optimization, trajectory tracking, active collision avoidance, and collision damage mitigation are incorporated into a unified control framework. The weights of each control objective are dynamically scheduled according to the pre-collision risk state, realizing continuous coordination of control objectives as the risk evolves and smooth switching of control output. This overcomes the problems of discontinuous control strategy switching and easy conflict or interference when implemented by discrete systems in related technologies, providing key hardware and algorithm integration support for active safety control of intelligent driving vehicles in complex pre-collision scenarios.

[0154] It should be noted that the integrated active collision avoidance and collision mitigation control method proposed in this embodiment can be integrated into the intelligent driving domain controller to achieve integrated active collision avoidance and collision mitigation control for pre-collision scenarios. This addresses the shortcomings of related technologies where active collision avoidance and collision mitigation systems implemented by separate controllers or systems are independent of each other, lack a unified coordination mechanism, and are difficult to achieve continuous control based on risk evolution in complex traffic scenarios. This intelligent driving domain controller, by integrating core functions such as pre-collision overlap domain analysis, candidate trajectory generation and target trajectory screening, and model predictive control optimization based on three-state membership scheduling at the hardware level, can prioritize active collision avoidance when collisions are avoidable, and further consider collision damage mitigation when collisions are difficult to completely avoid. This achieves integrated safety control of intelligent driving vehicles at the domain controller level in complex pre-collision scenarios, improving vehicle operation safety and control coordination. The intelligent driving domain controller is a software and hardware system integrating integrated control functions and is adaptable to intelligent driving systems in passenger cars, commercial vehicles, and other vehicles.

[0155] For example, using the method of this embodiment, in an intelligent driving domain controller, the following functions can be implemented: pre-collision overlap domain analysis and first contact feature calculation based on occupancy polygon intersection clipping; a two-layer filtering strategy for target trajectories by dividing trajectory sets according to overlap relationships and by scenario; a pre-collision three-state fuzzy membership degree calculation function based on overlap domain evolution features; and an MPC active collision avoidance and collision mitigation integrated control function based on three-state membership degree weight scheduling. The intelligent driving domain controller can realize the quantitative analysis function of vehicle occupancy polygon-overlap domain-first contact features for pre-collision scenarios; the scenario-based filtering function of pre-collision trajectories based on completely avoidable / overlapping sets; and the active collision avoidance and collision mitigation integrated model predictive control function based on three-state fuzzy membership degrees.

[0156] It is understood that alternative solutions can also be adopted in this embodiment, such as active collision avoidance and collision mitigation being independently controlled by separate controllers or systems and switched by fixed thresholds, passive control triggered by pure TTC / relative distance thresholds, and traditional MPC control with fixed weights.

[0157] It should be noted that the above method of this embodiment will be further explained below with reference to several specific embodiments.

[0158] Example 1

[0159] Specifically, this embodiment 1 aims to illustrate step S1 above, constructing a vehicle pre-collision overlap domain analysis model. First, the vehicle state information and obstacle vehicle prediction information output from multi-source sensor fusion are acquired and expressed. Considering that intelligent driving vehicles are typically equipped with various multimodal perception devices such as LiDAR, optical sensors, millimeter-wave radar, and inertial navigation units, and can combine high-precision map information and target motion prediction modules, the current position, motion state, and future motion trend of surrounding traffic participants can be obtained. Through the above perception, fusion, and prediction process, the current position, heading angle, speed, and other state information of the vehicle and obstacle vehicles can be obtained, as well as the predicted state sequence of the obstacle vehicles at each prediction step size within the prediction time domain.

[0160] For example, this embodiment, based on the obtained state information and prediction information, analyzes the spatial encroachment relationship between the vehicle and the obstacle vehicle in the prediction time domain and the pre-collision contact characteristics, including the geometric features of the overlapping domain of the pre-collision, the time of first contact, the point of first contact, and the angle of first contact. Through the above analysis, a vehicle pre-collision overlapping domain analysis model is constructed, providing a unified overlapping domain analysis method for subsequent candidate trajectory evaluation and screening and collision mitigation control. Figure 3 As shown, it specifically includes:

[0161] (1) First, the spatial occupancy area of ​​the autonomous vehicle and the obstacle vehicle in the prediction time domain is geometrically represented, and then the overlapping polygon of the two and its area are calculated at each prediction step to characterize the spatial encroachment relationship between the autonomous vehicle and the obstacle vehicle in the prediction time domain.

[0162] The discrete time interval for prediction in the time domain is:

[0163]

[0164] Based on the state information of the autonomous vehicle and the obstacle vehicle from multi-source sensor fusion, and the predicted state sequence of the obstacle vehicle in the prediction time domain, the vertex sequence of the occupied polygons of the autonomous vehicle and the obstacle vehicle is constructed at each prediction step. The occupied polygons of the autonomous vehicle and the obstacle vehicle at the k-th prediction step are represented as follows:

[0165]

[0166] Among them, V e (k) represents the sequence of polygon vertices occupied by the vehicle at the k-th prediction step, V o (k) represents the sequence of polygon vertices occupied by the obstacle vehicle at the kth prediction step; and Let n represent the coordinates of the corresponding vertices in a unified coordinate system. e and n o These represent the number of vertices of the polygon occupied by the vehicle and the obstacle vehicle, respectively. The vertices are arranged in a clockwise or counterclockwise direction to ensure the stability of subsequent polygon intersection operations.

[0167] For example, the overlapping region of the autonomous vehicle and the obstacle vehicle under the predicted step size is calculated, and the vertex sequence of the overlapping region polygon is obtained by polygon intersection clipping:

[0168]

[0169] Among them, V c (k) represents the sequence of vertices of the polygon overlapping the vehicle and the obstacle vehicle at the k-th prediction step. Clip(·) represents the polygon intersection clipping operation, and Sutherland-Hodgman, Weiler-Atherton, or other methods suitable for solving polygon intersections can be selected according to actual needs. When V e (k) and V o (k) When there is no intersection, V c (k) outputs an empty sequence.

[0170] For the overlapping polygon at the k-th prediction step, its vertex sequence can be written as:

[0171]

[0172] in, n represents the coordinates of the vertices of the overlapping domain in the global coordinate system. c This represents the number of vertices of the overlapping polygon.

[0173] The overlapping area A(k) represents the degree of spatial encroachment between the vehicle and the obstacle vehicle at the k-th prediction step, and is determined by the following equation:

[0174]

[0175] When A(k) = 0 throughout the entire prediction time domain, it means that there is no overlap between the two domains in the current prediction time domain; when A(k) > 0, it means that the two domains have overlapped at the corresponding step size, and the larger A(k) is, the higher the degree of spatial encroachment at the corresponding time.

[0176] (2) Then, based on the time distribution of the overlapping area, determine the time step when the vehicle and the obstacle vehicle first overlap, and calculate the continuous time when the boundary first contacts within the corresponding sampling interval.

[0177] The overlapping area sequence obtained according to step (1) of Example 1 The minimum step size that satisfies the condition that the area of ​​the overlapping region is greater than zero for the first time is defined as the first overlap step size:

[0178]

[0179] Where, k c It represents the time step in which the occupants of the vehicle and the obstacle vehicle first overlap in the discrete time domain, and only indicates the position where the occupants of the vehicle and the obstacle vehicle first have a non-empty intersection under the discrete prediction step.

[0180] Due to the sampling interval, the actual moment of first contact with the boundary may not fall exactly at the discrete time. It is more likely to be located in the sampling interval. A specific point in a continuous time period.

[0181] Therefore, in the interval The poses of the vehicle and the obstacle vehicles are interpolated continuously over time, and the occupied region polygon is constructed accordingly.

[0182]

[0183] Among them, Ω e (t) and Ω o (t) represent the continuous time-occupied boundary regions of the vehicle and the obstacle vehicle at continuous time t, respectively. G e (·) and G o (·) represents the constructor that maps the vehicle pose parameters to the geometric boundary of the vehicle's occupied domain. Its inputs are the vehicle's position ((x(t),y(t)), heading angle ψ(t), and vehicle geometric dimension parameter χ) at the corresponding time. The output is the polygon of the vehicle's occupied domain in the global coordinate system at the corresponding time. The continuous time pose can be obtained by linear interpolation, spline interpolation, or other continuous interpolation methods according to actual needs.

[0184] The minimum distance function between the boundaries of two polygons is defined as:

[0185]

[0186] in, Ω e (t) and Ω o(t) represents the boundary of the domain occupied by the vehicle and the obstacle vehicle at consecutive time t; p and q represent the candidate points on the corresponding boundary; Let represent the Euclidean distance norm. The function d(t) characterizes the minimum interval distance between vehicle boundaries over continuous time.

[0187] interval The minimum distance between the inner boundary and the inner boundary is less than the preset threshold ε for the first time. d The minimum time is defined as the time of first contact, and is determined by the following equation:

[0188]

[0189] in, ε represents the moment when the boundary between the vehicle and the obstacle vehicle first actually comes into contact in a continuous time sense; d The initial contact threshold ε can be preset according to actual needs. d The value.

[0190] (3) When the time of first contact is determined Based on this, the contact position and relative motion direction relationship at the time of the first contact are further calculated to characterize the geometric and motion features of the pre-collision contact state.

[0191] At the moment of first contact Establish a vehicle body coordinate system with the vehicle's center of gravity as the origin and the vehicle's longitudinal direction of travel as the vertical axis, and set the time... The boundaries of the occupied domains of the vehicle and the obstacle vehicle are transformed from the global coordinate system to the vehicle coordinate system, denoted as:

[0192]

[0193] in, and Representing time respectively The boundary of the domain occupied by the vehicle and the obstacle vehicle in the vehicle body coordinate system; and This represents a two-dimensional rigid body coordinate transformation function from the global coordinate system to the vehicle body coordinate system. Its inputs are the vehicle's occupied domain boundary in the global coordinate system at the corresponding time, as well as the vehicle's position and heading information at that time. The output is the representation of the occupied domain boundary in the vehicle body coordinate system.

[0194] In the vehicle coordinate system, define the first contact point from the side of the vehicle. for:

[0195]

[0196] in, Indicates candidate points on the vehicle boundary. Indicates candidate points on the boundary of the obstacle vehicle. This indicates the boundary point where the vehicle first comes into contact with the obstacle vehicle from its side. Correspondingly, the corresponding contact point on the obstacle vehicle's side can also be determined. This provides a geometric basis for constructing contact point preferences in subsequent collision mitigation control.

[0197] To reflect the difference in the relative motion direction between the vehicle and the obstacle vehicle at the moment of collision, the first contact angle is further calculated to characterize the relative velocity direction relationship between the vehicle and the obstacle vehicle at the moment of first contact.

[0198] Let the moment of first contact between the vehicle and the obstacle vehicle be... The velocity direction angles are respectively and First contact angle θ c Defined as the angle between the directions of their velocities:

[0199]

[0200] Here, wrap(·) represents the angle normalization function, which is used to normalize the angle to the interval (-π, π].

[0201] Based on the above content in Example 1, the vehicle pre-collision overlap domain analysis model was finally constructed. The vehicle pre-collision overlap domain analysis model can characterize the spatial encroachment relationship between the vehicle and the obstacle vehicle in the prediction time domain, and depict the key geometric and motion features at the time of the first contact, providing a unified analytical basis for subsequent candidate trajectory screening, collision risk identification and collision mitigation control.

[0202] Example 2

[0203] Specifically, this embodiment 2 aims to illustrate step S2 above, which involves constructing candidate trajectory generation and target trajectory selection. Based on the pre-collision overlap domain analysis results, the overlap relationship between the candidate trajectory and the obstacle vehicle is analyzed, and the target trajectory is determined by combining active collision avoidance decision or collision mitigation decision. The target trajectory comprehensively considers the vehicle's dynamic performance, the pre-collision relationship between the trajectory and the obstacle vehicle, and the corresponding safety decisions, providing trajectory input for subsequent integrated control optimization. The specific process is as follows: Figure 4 As shown, it includes:

[0204] (1) First, a candidate trajectory set is constructed, and the motion state of each candidate trajectory in the prediction time domain is discretized. The candidate trajectory set T is defined as:

[0205]

[0206] in, This is a set of lateral decision points, used to characterize different lateral avoidance decisions or lateral offset schemes, generated through vehicle dynamics sampling; This is a set of longitudinal decision points, used to characterize different longitudinal speed planning or braking schemes, generated through vehicle dynamics sampling; This is a time series set used to characterize the duration or timing parameters of different trajectory actions, generated through a preset sampling interval; function Indicates the lateral decision parameters A candidate trajectory is jointly determined by the longitudinal decision parameter s and the time sequence parameter T. Depending on the actual needs, the fifth-order polynomial method, the Bézier curve method, or other trajectory generation methods that meet the initial and terminal boundary conditions can be adopted.

[0207] For any candidate trajectory τ j ∈ The trajectory can be discretized in the prediction time domain according to the sampling period, and its discrete representation can be expressed as a state sequence:

[0208]

[0209] Among them, (x ej (k),y ej (k) represents the vehicle position corresponding to the j-th candidate trajectory at the k-th prediction step; ψ ej (k) represents the heading angle at the corresponding moment; v ej (k) represents the velocity at the corresponding moment.

[0210] (2) Trajectory Generation Method. Based on the above-mentioned construction of the candidate trajectory set, and combined with the vehicle pre-collision overlap domain analysis model constructed in Example 1, the overlap relationship between each candidate trajectory and the obstacle vehicle is calculated step by step to determine the sequence of vertex V of the overlap domain polygon between the j-th candidate trajectory and the obstacle vehicle at the k-th prediction step. cj (k), the corresponding overlapping area A j (k) and the time step k in which the occupied regions corresponding to the j-th candidate trajectory first overlap. cj .

[0211] If A satisfies all prediction step sizes j If (k)=0, it means that the candidate trajectory does not overlap with the obstacle vehicle's occupied domain in the current prediction time domain. Based on this determination, the candidate trajectory set is divided into a completely avoidable set. Overlapping sets :

[0212]

[0213] For the existence of overlapping sets For each candidate trajectory, its first contact information in the continuous time sense needs to be further determined. Based on the first contact time determination method in Example 1 above, the first contact time of the trajectory is determined. The first contact point in the vehicle body coordinate system is then determined. Angle θ of first contact cj .

[0214] (3) After dividing the candidate trajectory set, select the corresponding trajectory screening strategy based on whether there is a feasible trajectory in the completely avoidable set.

[0215] When the completely avoidable set When the path is not empty, it means that there is at least one candidate trajectory in the current prediction time domain that can completely avoid overlapping with the obstacle vehicle's occupied domain. Trajectory selection is only performed within the set of completely avoidable paths, and a selection criterion is constructed based on the trade-off between comfort and stability.

[0216]

[0217] Among them, J safe (τ j ) represents the comprehensive cost of the j-th candidate trajectory in the completely avoidable set; J comf Indicates the cost of comfort, used to characterize the smoothness of acceleration, jerk, or steering changes corresponding to the trajectory; J stab (τ j ) represents the stability cost, used to characterize the vehicle's dynamic stability corresponding to the trajectory; w comf and w stab These represent the weight coefficients of the corresponding cost terms.

[0218] The cost of comfort is defined as:

[0219]

[0220] in, This represents the longitudinal acceleration of the j-th candidate trajectory at the k-th prediction step. Let Δa represent the lateral acceleration of the j-th candidate trajectory at the k-th prediction step. j (k) represents the longitudinal acceleration change between adjacent prediction steps, and α1, α2 and α3 are the weight parameters of the longitudinal acceleration, lateral acceleration and longitudinal acceleration change cost terms, respectively.

[0221] The stability cost is defined as:

[0222]

[0223] in, This represents the heading angle deviation of the j-th candidate trajectory at the k-th prediction step. This represents the corresponding yaw rate. This represents the corresponding front wheel steering angle or equivalent steering control quantity; β1, β2, and β3 are the weight parameters for heading angle deviation, yaw rate, and steering input cost, respectively.

[0224] For example, when the set is completely avoidable When empty, it indicates that there are no candidate trajectories that completely avoid overlapping occupied domains within the current prediction time domain. Further comprehensive screening criteria are constructed based on collision probability and collision severity only within the overlapping set.

[0225]

[0226] Among them, J col (τ j ) represents the comprehensive cost of the j-th candidate trajectory in the overlapping set; J prob (τ j ) represents the cost of a collision probability; J sev (τ j ) indicates the severity of the collision cost; w prob and w sev These represent the weighting coefficients for the corresponding collision probability and collision severity cost terms, respectively.

[0227] The collision probability cost is defined as:

[0228]

[0229] Where g1, g2, g3 and g4 are weight parameters in the collision probability cost; 1(·) is the indicator function, used to statistically predict the proportion of overlapping steps in the time domain.

[0230] D j The number of overlapping step sizes is defined as:

[0231]

[0232] The cost of collision severity is:

[0233]

[0234] Where, ΔV j The speed difference between the vehicle and the obstacle vehicle is represented when the first overlap occurs; h1, h2 and h3 represent the weight parameters of the speed difference, contact point preference and contact angle preference cost terms, respectively.

[0235] J p (τ j ) represents the collision point preference cost, defined as:

[0236]

[0237] in, This indicates that the j-th candidate trajectory occurs at consecutive time k. cj The point of collision at time θ ref This indicates the preset expected collision point.

[0238] J θ (τ j The term ) represents the collision angle preference cost, defined as:

[0239]

[0240] Where, θ cj This indicates that the j-th candidate trajectory occurs at consecutive time k. cj The collision angle at time, θ ref This indicates the preset desired collision angle.

[0241] After completing the above comprehensive cost construction, traversal search, gradient-based nonlinear optimization algorithm, heuristic search method, or other numerical optimization methods suitable for screening discrete candidate trajectories can be selected according to the implementation requirements to evaluate and compare each candidate trajectory in the candidate trajectory set, and select the candidate trajectory with the minimum corresponding comprehensive cost value as the target trajectory.

[0242] When the completely avoidable set is not empty, the candidate trajectory that satisfies the following equation is selected as the target trajectory:

[0243]

[0244] When the set of completely avoidable trajectories is empty, the candidate trajectory that satisfies the following formula is selected as the target trajectory:

[0245]

[0246] in, The final determined target trajectory is represented by the discrete state sequence corresponding to the target trajectory in the prediction time domain, denoted as:

[0247]

[0248] in, , and These represent the vehicle's position, heading angle, and velocity at the k-th prediction step of the target trajectory, respectively.

[0249] Example 3

[0250] Specifically, this embodiment 3 aims to illustrate step S3 above, constructing a three-state membership calculation and integrated control optimization model based on the target trajectory. This process takes the target trajectory determined in embodiment 2 and its corresponding overlapping domain features as input. On one hand, it calculates the three-state membership of the avoidable state, transition state, and pre-collision state through fuzzy inference, and uses the three-state membership to schedule the control target weights. On the other hand, it constructs a model prediction control trajectory tracking model based on the target trajectory, thereby outputting the vehicle's steering control sequence and longitudinal acceleration control sequence, realizing the integration of active collision avoidance and collision mitigation. The specific process is as follows: Figure 5 As shown, it includes:

[0251] (1) Based on the target trajectory and its overlapping domain area sequence determined in Example 2, construct the three-state fuzzy inference input quantity, including the temporal proximity, overlap intensity and overlap persistence representation quantity.

[0252] Time proximity s t Defined as:

[0253]

[0254] When the first overlap occurs earlier, s t The larger the value, the earlier the pre-collision risk will arrive.

[0255] Overlap strength s a Defined as:

[0256]

[0257] Here, ε is a very small positive number set to prevent the denominator from being zero. This quantity is used to characterize the maximum spatial encroachment intensity of the target trajectory in the prediction time domain.

[0258] Overlapping persistence s d Defined as:

[0259]

[0260] Where 1(·) represents the characteristic function. d The larger the value, the stronger the continuity of the overlap between the target trajectory and the obstacle vehicle.

[0261] After obtaining the above input quantities, each input quantity is fuzzified. The membership function can be a trapezoidal membership function, triangular membership function, Gaussian membership function, or other membership function forms suitable for fuzzy inference, depending on the implementation requirements. For each input quantity, three classes of fuzzy sets (low, medium, and high) are defined, and a fuzzy rule base is constructed using a min-max inference approach. The fuzzy rules are defined as follows:

[0262]

[0263] Where L(·), M(·), and H(·) represent low, medium, and high membership functions, respectively; m1, m2, and m3 represent the fuzzy inference outputs corresponding to the avoidable state, the transition state, and the pre-collision state, respectively.

[0264] Further normalization of the fuzzy inference results yields the three-state membership degree:

[0265]

[0266] Where μ1, μ2 and μ3 represent the membership degree of the avoidable state, the membership degree of the transition state and the membership degree of the pre-collision state corresponding to the target trajectory, respectively.

[0267] Based on the obtained three-state membership degrees, the weights of each objective term in the model's predictive control objective function are further scheduled according to the three-state membership degrees:

[0268]

[0269] Among them, W T1 W T2 W A1 W A2 W M2 and W M3 For preset weight parameters; w trk w avd and w mit These represent the weighting coefficients for the trajectory error term, collision avoidance term, and collision mitigation term, respectively. Through this weighting method, the importance of trajectory tracking and active collision avoidance can be enhanced when the avoidability level is high, and the importance of collision mitigation can be enhanced when the pre-collision level is high, thereby achieving continuous adjustment of the control target as the pre-collision state evolves.

[0270] (2) Based on the obtained control weights and the target trajectory determined in Example 2, an MPC trajectory tracking model is constructed, and the vehicle control sequence is defined as follows in the prediction time domain:

[0271]

[0272] Where c represents the control sequence in the prediction time domain; u(k) represents the control input at the k-th control step; δ(k) represents the steering control quantity; and a(k) represents the longitudinal acceleration control quantity.

[0273] The objective function for model predictive control is:

[0274]

[0275] Among them, J trk (k) represents the trajectory tracking term, J avd (k) represents the active collision avoidance term, Jmit (k) represents the collision mitigation term, J reg (k) indicates the control regularization term.

[0276] The trajectory tracking term is used to constrain the tracking error of the vehicle on the target trajectory, and is defined as:

[0277]

[0278] Among them, e y (k), e ψ (k) and e v (k) represent the lateral position error, heading error, and velocity error at the k-th prediction step, respectively.

[0279] The active collision avoidance term characterizes the relative proximity between the target trajectory and the obstacle vehicle at the k-th prediction step, and is defined as:

[0280]

[0281] Where, d eo (k) represents the minimum boundary distance between the occupancy region of the target vehicle and the occupancy region of the obstacle vehicle at the k-th prediction step, ε a A very small positive number set to prevent the denominator from being zero. The minimum boundary distance can be expressed as:

[0282]

[0283] in, This indicates the boundary of the vehicle's occupied domain corresponding to the target trajectory. This indicates the boundary of the area occupied by the obstacle vehicle. The closer the two vehicles are, the more pronounced the boundary becomes (J). avd The larger (k) is, the stronger the corresponding active collision avoidance requirement.

[0284] The collision mitigation term is used to characterize the effects of relative velocity, contact position, and contact angle on the consequences of a collision when complete avoidance of collision is unavoidable. It is defined as follows:

[0285]

[0286] Where ΔV(k) represents the speed difference between the vehicle and the obstacle vehicle at the k-th prediction step, and r1, r2 and r3 represent the weight coefficients of the corresponding target items.

[0287] J p (k) represents the contact point preference term, defined as:

[0288]

[0289] in, This represents the contact point in the vehicle coordinate system corresponding to the target trajectory at the k-th prediction step. This indicates the preset desired contact point.

[0290] J θ (k) represents the contact angle preference term, defined as:

[0291]

[0292] in, θ represents the contact angle of the target trajectory at the k-th prediction step. ref This indicates the preset desired contact angle.

[0293] By introducing collision point preference and collision angle preference terms, in scenarios where a collision is unavoidable, the vehicle can be guided to control in a relative contact direction that is more conducive to damage mitigation.

[0294] The control regularization term is used to suppress excessive control inputs and their rates of change, and is defined as follows:

[0295]

[0296] Where Δu(k) represents the change in control input between adjacent control steps.

[0297] By constructing the above objective function, trajectory tracking, active collision avoidance, collision mitigation, and control smoothness can be uniformly optimized, enabling intelligent driving vehicles to achieve dynamic coordination of control objectives and smooth switching of control outputs in pre-collision scenarios as pre-collision risks continue to evolve.

[0298] In this embodiment, the aforementioned intelligent driving domain controller for integrated active collision avoidance and collision mitigation control in pre-collision scenarios can be adopted to address the shortcomings of related technologies, such as the lack of unified analysis of key geometric features of the pre-collision process and the difficulty in effectively integrating collision contact geometry information into trajectory decision-making and control optimization. By constructing a pre-collision overlap domain analysis model, target trajectory screening, three-state fuzzy inference, and model predictive control optimization, the spatial encroachment relationship, initial contact geometry, and risk evolution degree during the pre-collision process can be accurately characterized. Furthermore, the trajectory tracking, active collision avoidance, and collision mitigation targets can be dynamically coordinated according to different risk states, thereby achieving continuous switching of control strategies and stable output of control quantities in pre-collision scenarios, providing key technical support for active safety protection of intelligent driving vehicles.

[0299] The active collision avoidance and collision mitigation integrated control method provided in this embodiment includes: acquiring the multi-source sensor fusion state and prediction sequence of the vehicle and the obstacle vehicle to construct a vehicle pre-collision overlap domain analysis model; analyzing the overlap relationship between candidate trajectories and obstacle vehicles based on the vehicle pre-collision overlap domain analysis model to obtain a fully avoidable set and a set with overlap; determining the target trajectory based on the set state of the fully avoidable set and the set with overlap; and performing integrated model predictive control based on the target trajectory to output the steering and longitudinal acceleration control sequences of the vehicle. This embodiment, by constructing a pre-collision overlap domain analysis model, integrates the vehicle and obstacle vehicles into a unified analysis framework, improving the accuracy and completeness of pre-collision risk analysis and providing a unified geometric analysis basis for subsequent trajectory decision-making and control. Introducing the features obtained from the pre-collision overlap domain analysis into the candidate trajectory screening process enables the target trajectory to characterize potential collision relationships and collision contact features, compensating for the lack of collision contact geometric feature constraints in trajectory planning.

[0300] Reference Figure 6 , Figure 6 This is a structural block diagram of an embodiment of the integrated active collision avoidance and collision mitigation control system of the present invention. Figure 6 As shown, the integrated active collision avoidance and collision mitigation control system includes:

[0301] The pre-collision overlap domain analysis module 10 is used to acquire the multi-source sensor fusion state and prediction sequence of the vehicle and the obstacle vehicle to construct a vehicle pre-collision overlap domain analysis model.

[0302] The set partitioning module 20 is used to analyze the overlap relationship between candidate trajectories and obstacle vehicles based on the vehicle pre-collision overlap domain analysis model, and obtain a completely avoidable set and a set with overlap.

[0303] The target trajectory filtering module 30 is used to determine the target trajectory based on the set state of the completely avoidable set and the set with overlapping elements;

[0304] The integrated predictive control module 40 is used to perform integrated model predictive control based on the target trajectory and output the vehicle's steering and longitudinal acceleration control sequence.

[0305] Specifically, the active collision avoidance and collision mitigation integrated control system in this embodiment can be configured in the intelligent driving domain controller. By constructing and running a pre-collision overlap domain analysis model, the spatial encroachment relationship between the vehicle and the obstacle vehicle in the prediction time domain, the first overlap time, and the geometric features of the first contact are uniformly incorporated into the analysis framework. This achieves continuous representation of key geometric information of the pre-collision process, overcoming the shortcomings of related technologies that rely only on relative distance or a single risk indicator and are difficult to accurately characterize complex pre-collision relationships. Thus, the accuracy and completeness of pre-collision risk analysis are improved at the controller level.

[0306] For example, by incorporating geometric features such as the overlap relationship, first contact point, and first contact angle obtained from the pre-collision overlap domain analysis into the candidate trajectory screening process, the target trajectory can characterize the potential collision relationship and collision contact features. Especially in scenarios where collisions are difficult to completely avoid, it can provide a basis for trajectory decision-making guided by collision damage mitigation, making up for the lack of collision contact geometric feature constraints in trajectory planning in related technologies.

[0307] For example, by implementing three-state fuzzy membership degree and model predictive control optimization in the intelligent driving domain controller, trajectory tracking, active collision avoidance, and collision damage mitigation are incorporated into a unified control framework. The weights of each control objective are dynamically scheduled according to the pre-collision risk state, realizing continuous coordination of control objectives as the risk evolves and smooth switching of control output. This overcomes the problems of discontinuous control strategy switching and easy conflict or interference when implemented by discrete systems in related technologies, providing key hardware and algorithm integration support for active safety control of intelligent driving vehicles in complex pre-collision scenarios.

[0308] The integrated active collision avoidance and collision mitigation control system provided in this embodiment improves the accuracy and completeness of pre-collision risk analysis by constructing a pre-collision overlap domain analysis model, which unifies the analysis framework for both the vehicle and obstacle vehicles, providing a unified geometric analysis basis for subsequent trajectory decision-making and control. The features obtained from the pre-collision overlap domain analysis are introduced into the candidate trajectory selection process, enabling the target trajectory to characterize potential collision relationships and collision contact features, thus compensating for the lack of collision contact geometric feature constraints in trajectory planning.

[0309] Furthermore, for technical details not described in detail in this embodiment of the integrated active collision avoidance and collision mitigation control system, please refer to the integrated active collision avoidance and collision mitigation control method provided in any embodiment of the present invention, which will not be repeated here.

[0310] Based on the same inventive concept, embodiments of the present invention also provide an electronic device. Figure 7 This is a structural block diagram of an electronic device provided in an embodiment of the present invention. Figure 7 As shown, an embodiment of the present invention provides an electronic device including: one or more processors 101, a memory 102, and one or more I / O interfaces 103. The memory 102 stores one or more programs, which, when executed by the one or more processors, enable the one or more processors to implement any of the active collision avoidance and collision mitigation integrated control methods described in the above embodiments; the one or more I / O interfaces 103 are connected between the processor and the memory, configured to enable information interaction between the processor and the memory.

[0311] The processor 101 is a device with data processing capabilities, including but not limited to a central processing unit (CPU); the memory 102 is a device with data storage capabilities, including but not limited to random access memory (RAM, more specifically SDRAM, DDR, etc.), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), and flash memory (FLASH); the I / O interface (read / write interface) 103 is connected between the processor 101 and the memory 102, and can realize information interaction between the processor 101 and the memory 102, including but not limited to a data bus (Bus).

[0312] In some embodiments, the processor 101, memory 102, and I / O interface 103 are interconnected via bus 104, and thus connected to other components of the computing device.

[0313] In some embodiments, the one or more processors 101 include a field-programmable gate array.

[0314] This invention also provides a computer-readable medium. The computer-readable medium stores a computer program, which, when executed by a processor, implements the steps of any of the integrated active collision avoidance and collision mitigation control methods described in the above embodiments. The computer-readable storage medium can be volatile or non-volatile.

[0315] This invention also provides a computer program product, including computer-readable code, or a non-volatile computer-readable storage medium carrying computer-readable code. When the computer-readable code is run in the processor of an electronic device, the processor in the electronic device executes the above-described integrated active collision avoidance and collision mitigation control method.

[0316] Those skilled in the art will understand that all or some of the steps, systems, and apparatuses disclosed above, and their functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof. In hardware implementations, the division between functional modules / units mentioned above does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed collaboratively by several physical components. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software can be distributed on a computer-readable storage medium, which may include computer storage media (or non-transitory media) and communication media (or transient media).

[0317] As is known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information, such as computer-readable program instructions, data structures, program modules, or other data. Computer storage media includes, but is not limited to, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), static random access memory (SRAM), flash memory or other memory technologies, portable compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, it is known to those skilled in the art that communication media typically contain computer-readable program instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0318] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0319] The computer program instructions used to perform the operations of this invention may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may 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 a remote computer, the remote computer may 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 may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing state information from the computer-readable program instructions. This electronic circuitry can execute the computer-readable program instructions to implement various aspects of the invention.

[0320] The computer program product described herein can be implemented specifically through hardware, software, or a combination thereof. In one alternative embodiment, the computer program product is specifically embodied in a computer storage medium; in another alternative embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0321] Various aspects of the present invention are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer-readable program instructions.

[0322] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that, when executed by the processor of the computer or other programmable data processing apparatus, they create means for implementing the functions / actions specified in one or more blocks of the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium that causes a computer, programmable data processing apparatus, and / or other device to operate in a particular manner; thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing aspects of the functions / actions specified in one or more blocks of the flowchart and / or block diagram.

[0323] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to perform the functions / actions specified in one or more boxes of a flowchart and / or block diagram.

[0324] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of an instruction, which contains one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than those shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0325] Example embodiments have been disclosed herein, and while specific terminology has been used, it is for illustrative purposes only and should be construed as such, and is not intended to be limiting. In some instances, it will be apparent to those skilled in the art that features, characteristics, and / or elements described in conjunction with particular embodiments may be used alone, or in combination with features, characteristics, and / or elements described in conjunction with other embodiments, unless otherwise expressly indicated. Therefore, those skilled in the art will understand that various changes in form and detail may be made without departing from the scope of the invention as set forth in the appended claims.

Claims

1. A method for integrated active collision avoidance and collision mitigation control, characterized in that, include: The multi-source sensor fusion states and prediction sequences of the vehicle and the obstacle vehicle are obtained to construct a vehicle pre-collision overlap domain analysis model; Based on the vehicle pre-collision overlap domain analysis model, the overlap relationship between candidate trajectories and obstacle vehicles is analyzed to obtain a completely avoidable set and a set with overlap. The target trajectory is determined based on the set state of the completely avoidable set and the set that overlaps with it; Based on the target trajectory, an integrated model predictive control is performed, outputting the vehicle's steering and longitudinal acceleration control sequences.

2. The method according to claim 1, characterized in that, The acquisition of multi-source sensor fusion states and prediction sequences from the vehicle and obstacle vehicles to construct a vehicle pre-collision overlap domain analysis model includes: Acquire the state information of the vehicle and the obstacle vehicle from the multi-source sensor fusion output, as well as the predicted state sequence of the obstacle vehicle in the prediction time domain; Based on the state information of the autonomous vehicle and the obstacle vehicle and the predicted state sequence, the occupancy polygons of the autonomous vehicle and the obstacle vehicle at each prediction step length are constructed respectively. Based on the polygon intersection clipping operation, the overlapping region polygons and the area of ​​the overlapping region are calculated according to the occupied polygons, and the time step when the occupied regions first overlap is determined according to the area of ​​the overlapping region. The first contact time, first contact point, and first contact angle are calculated based on the time step in which the occupied domains first overlap. Based on the occupied polygon, overlapping region polygon, overlapping region area, time step of the first overlap, first contact time, first contact point, and first contact angle, a vehicle pre-collision overlapping region analysis model is constructed.

3. The method according to claim 2, characterized in that, The calculation of the first contact time, first contact point, and first contact angle based on the time step of the first overlap of the occupied domains includes: The sampling interval corresponding to the time step in which the occupied domains first overlap is taken as the preset sampling interval; Within the preset sampling interval, continuous-time interpolation is performed on the poses of the vehicle and the obstacle vehicle to obtain the continuous-time occupied region polygon of the vehicle and the obstacle vehicle. The minimum distance between the boundaries is defined based on the polygonal domain occupied by the autonomous vehicle and the obstacle vehicle during continuous time. The minimum moment within the preset sampling interval when the minimum distance at the boundary first falls below the preset threshold is defined as the first contact moment. The first contact point and the first contact angle are calculated based on the first contact time.

4. The method according to claim 1, characterized in that, The vehicle pre-collision overlap domain analysis model is used to analyze the overlap relationship between candidate trajectories and obstacle vehicles, resulting in a completely avoidable set and a set with overlap, including: A candidate trajectory set is constructed based on the horizontal decision set, the vertical decision set, and the time series set. Generate a discrete state sequence of each candidate trajectory in the prediction time domain based on the candidate trajectory set; Based on the vehicle pre-collision overlap domain analysis model and discrete state sequence, the overlap relationship between each candidate trajectory and the obstacle vehicle is calculated. Based on the overlap relationship, the candidate trajectory set is divided into a completely avoidable set and a set with overlap.

5. The method according to claim 1, characterized in that, The step of determining the target trajectory based on the set state of the completely avoidable set and the set with overlapping elements includes: When the set state of the completely avoidable set is not empty, trajectory filtering is performed within the completely avoidable set to determine the target trajectory; When the set of completely avoidable sets is empty, a comprehensive screening index is constructed within the set of overlapping sets, and a target trajectory is selected based on the comprehensive screening index.

6. The method according to claim 1, characterized in that, The integrated model predictive control based on the target trajectory outputs a steering and longitudinal acceleration control sequence for the vehicle, including: Fuzzy reasoning is performed based on the target trajectory to obtain the three-state membership degree; An integrated control optimization model is constructed based on the target trajectory; The weights of each objective item in the integrated control optimization model are scheduled according to the three-state membership degree, and the steering control sequence and longitudinal acceleration control sequence of the vehicle are output.

7. The method according to claim 6, characterized in that, The fuzzy reasoning based on the target trajectory to obtain the three-state membership degree includes: Based on the target trajectory and the overlap region area sequence of the target trajectory, input quantities are constructed; wherein, the input quantities include a temporal proximity representation quantity, an overlap intensity representation quantity, and an overlap persistence representation quantity; The input quantity is fuzzified based on a preset membership function and a fuzzy rule base to obtain the fuzzy inference result. The fuzzy reasoning results are normalized to obtain a three-state membership degree; wherein the three-state membership degree includes an avoidable state, a transitional state, and a pre-collision state.

8. An integrated active collision avoidance and collision mitigation control system, characterized in that, include: The pre-collision overlap domain analysis module is used to obtain the multi-source sensor fusion state and prediction sequence of the vehicle and the obstacle vehicle in order to build a vehicle pre-collision overlap domain analysis model. The set partitioning module is used to analyze the overlap relationship between candidate trajectories and obstacle vehicles based on the vehicle pre-collision overlap domain analysis model, and obtain the completely avoidable set and the overlapping set. The target trajectory filtering module is used to determine the target trajectory based on the set state of the completely avoidable set and the set with overlapping elements; An integrated predictive control module is used to perform integrated model predictive control based on the target trajectory, and output the vehicle's steering and longitudinal acceleration control sequences.

9. An electronic device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1 to 7.

10. A computer-readable medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 7.

Images