An automatic driving heavy truck safety evaluation method based on multi-dimensional driving parameters

Abstract

The application discloses a kind of automatic driving heavy truck safety evaluation methods based on multidimensional driving parameter, it is related to automatic driving technical field, including, based on combination risk transmission record, with current control instruction as expansion starting point, in longitudinal acceleration execution boundary and steering angular velocity execution boundary according to corresponding execution resolution expansion control value, form candidate control quantity, the combination vehicle motion state corresponding to candidate control quantity is deduced hour by hour, respectively form current joint safety control domain and next joint safety control domain, and according to control quantity change ability match corresponding candidate control quantity, form stage safety undertaking record;According to stage safety undertaking record, verify whether current joint safety control domain exists, the position and continuous undertaking state of current control instruction, determine real-time safety evaluation result, and select corresponding candidate control quantity to generate control correction instruction.The application improves the real-time, continuity and accuracy of lateral risk identification, lane-changing safety evaluation and control correction.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, and in particular to a safety evaluation method for autonomous heavy-duty trucks based on multi-dimensional driving parameters. Background Technology

[0002] Autonomous driving heavy-duty truck control technology is evolving from single-vehicle perception and rule-based control towards multi-source parameter fusion, vehicle combination motion simulation, and real-time risk assessment. Existing technologies typically utilize information such as vehicle speed, lateral acceleration, yaw rate, vehicle position, and lane boundaries to monitor vehicle operating status and mitigate collision and lane departure risks through braking control, steering control, and lane-keeping strategies. With the increasing application of tractor-trailer combinations in trunk line transportation, port transportation, and closed roads, the controlled object has expanded from a single vehicle to articulated combinations, placing higher demands on lateral response transmission, vehicle outline constraints, and continuous control capabilities during lane changes.

[0003] Existing safety evaluation technologies for autonomous heavy-duty trucks still have certain shortcomings. Multidimensional driving parameters are usually analyzed based on independent sampled values, lacking temporal correlation with the control cycle. It is difficult to distinguish between real sudden responses and isolated sensor jumps, and it is also difficult to accurately describe the time delay and response amplification phenomenon caused by the transmission of lateral response from the tractor to the trailer. Existing safety evaluations usually judge the safety of control commands at a single moment or a single lane-changing phase, lacking continuous verification of the ability to convert candidate control quantities between the current lane-changing phase and the next lane-changing phase. This can easily lead to situations where the current phase meets safety constraints, but the control cannot be converted in time when entering the next phase. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a safety evaluation method for autonomous driving heavy trucks based on multi-dimensional driving parameters to solve the problems of inaccurate identification of risk transmission between tractor and trailer and difficulty in continuously verifying the safety handover of control quantities during adjacent lane change phases.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a safety evaluation method for autonomous driving heavy-duty trucks based on multi-dimensional driving parameters. The method includes: acquiring multi-dimensional driving parameters, road environment information, and current control commands of the autonomous driving heavy-duty truck; performing spatiotemporal correlation and risk preservation processing on the multi-dimensional driving parameters and road environment information to form a multi-dimensional driving state record; locating the lateral response start time of the tractor and the peak response time of the trailer in the multi-dimensional driving state record, determining the risk transmission time difference and the trailer response amplification state, and determining the lane-changing stage corresponding to the response peak according to the positional relationship between the vehicle outline and the lane boundary to form a combined risk transmission record; based on the combined risk transmission record, ... Starting from the current control command, the control values ​​are expanded within the longitudinal acceleration execution boundary and the steering angular velocity execution boundary according to the corresponding execution resolution to form candidate control quantities. The combined vehicle motion state corresponding to the candidate control quantities is deduced hourly to form the current joint safety control domain and the next joint safety control domain. The corresponding candidate control quantities are matched according to the control quantity change capability to form a stage safety transition record. Based on the stage safety transition record, the existence of the current joint safety control domain, the position of the current control command, and the continuous transition status are verified to determine the real-time safety evaluation result and select the corresponding candidate control quantities to generate control correction commands.

[0007] As a preferred embodiment of the autonomous driving heavy truck safety evaluation method based on multi-dimensional driving parameters described in this invention, the step of forming a multi-dimensional driving state record includes: Vehicle speed, vehicle position, vehicle attitude, steering angle, tractor lateral acceleration, tractor yaw rate and trailer yaw rate are obtained from vehicle bus and inertial measurement unit. Vehicle spacing, relative speed, current lane boundary position and target lane boundary position are obtained from environmental perception unit. The sampling time is associated with the control cycle in which the current control command is located, and the time relationship between parameter changes and the current control command within the same control cycle is maintained. Based on the vehicle speed and steering angle at the start of the control cycle, the longitudinal acceleration and steering angular velocity in the current control command, and the duration of the control cycle, the range of speed change and the range of lateral position change at the end of the control cycle are deduced to form the achievable range of vehicle movement. Continuous verification is performed on the changes in vehicle spacing, relative speed, and vehicle position relative to the current lane boundary and the target lane boundary to form a multi-dimensional driving state record.

[0008] As a preferred embodiment of the safety evaluation method for autonomous heavy-duty trucks based on multi-dimensional driving parameters described in this invention, the step of forming a combined risk transmission record includes: According to the sampling time sequence of the multi-dimensional driving state record, the sampling time when the lateral acceleration and yaw rate of the tractor are synchronously enhanced is identified and determined as the start time of the lateral response of the tractor. After the start time of the lateral response of the tractor, the sampling time when the yaw rate of the trailer changes from enhancement to decay is identified and determined as the peak time of the trailer response. The time interval between the start time of the lateral response of the tractor and the peak time of the trailer response is calculated. The peak yaw rates of the tractor and the trailer are compared to determine the risk transmission time difference and the trailer response amplification state, respectively. Based on the position of the tractor and trailer outlines relative to the current lane boundary and the target lane boundary at the peak of the trailer response, and combined with the sequential relationship and lateral position changes of the tractor and trailer outlines crossing lane boundaries, the lane-changing phase corresponding to the peak of the trailer response is determined, forming a combined risk transmission record.

[0009] As a preferred embodiment of the autonomous driving heavy-duty truck safety evaluation method based on multi-dimensional driving parameters described in this invention, the step of determining the risk transmission time difference and trailer response amplification state includes: The risk transmission process that has appeared at the start time of the lateral response of the tractor and the peak time of the response of the trailer is extracted from the multi-dimensional driving status record. The process is then filtered in the order of consistent steering direction, decreasing speed difference at the start time of the lateral response of the tractor, and decreasing steering angle change difference. The remaining risk transmission process after filtering is determined as the reference response process. Align the lateral response start time of the tractor in the reference response process with the lateral response start time of the current tractor. Based on the time interval from the lateral response start time of the tractor to the peak response time of the trailer in the reference response process, determine the expected peak response time of the current trailer in the current sampling time series, and determine the time interval from the lateral response start time of the current tractor to the expected peak response time of the current trailer as the risk transmission time difference. Based on the ratio of the change in the peak yaw rate of the tractor relative to the peak yaw rate of the tractor during the reference response process, the peak yaw rate of the trailer during the reference response process is adjusted to determine the expected response amplitude of the trailer. The expected response amplitude of the trailer is then compared with the peak yaw rate of the tractor to determine the amplification state of the trailer response.

[0010] As a preferred embodiment of the safety evaluation method for autonomous heavy-duty trucks based on multi-dimensional driving parameters described in this invention, the safety acceptance record in the formation stage includes: Read the longitudinal acceleration execution boundary, steering angular velocity execution boundary, longitudinal acceleration rate of change, steering angular velocity rate of change and corresponding execution resolution given by the vehicle controller. Based on the longitudinal acceleration and steering angular velocity in the current control command, select longitudinal acceleration control values ​​and steering angular velocity control values ​​step by step and combine them to form candidate control quantities according to the corresponding execution resolution along the direction of numerical increase and the direction of numerical decrease. Starting with the current vehicle state in the multi-dimensional driving state record, the combined vehicle motion state corresponding to each candidate control quantity is deduced sequentially. The trailer motion state is adjusted according to the combined risk transmission record to form a candidate state sequence. The braking distance, vehicle outline position and relative motion limit between the tractor and trailer within the candidate state sequence form the dynamic safety boundary of the heavy truck. The candidate control quantities that have not exceeded the dynamic safety boundary of the heavy truck are retained to form the current joint safety control domain and the next joint safety control domain. Based on the moment when the combined vehicle outline reaches the position of the lane boundary corresponding to the next lane-changing stage, the remaining time of the stage is determined. According to the longitudinal acceleration change rate and steering angular velocity change rate, the control transition time required to transition from the candidate control quantity in the current joint safety control domain to the candidate control quantity in the next joint safety control domain is obtained. The continuous handover state and the handover interruption stage are determined, and a stage safety handover record is formed.

[0011] As a preferred embodiment of the autonomous driving heavy truck safety evaluation method based on multi-dimensional driving parameters described in this invention, the step of adjusting the trailer's motion state according to the combined risk transmission record includes: The trailer response peak moment in the reference response process is time-aligned with the current expected peak moment of the trailer response. Combined with the current expected response amplitude of the trailer and the peak relationship between the yaw rate of the tractor and the trailer in the reference response process, the yaw rate of the trailer is corrected to form a trailer response correction sequence. The trailer response correction sequence is written into the candidate state sequence. The articulation angle between the tractor and the trailer is updated by accumulating the difference between the yaw rate of the tractor and the yaw rate in the trailer response correction sequence. The yaw rate in the trailer response correction sequence is accumulated hourly. The trailer position and trailer attitude in the candidate state sequence are updated with the articulation position of the tractor and the trailer stored in the vehicle controller as the connection reference.

[0012] As a preferred embodiment of the autonomous driving heavy truck safety evaluation method based on multi-dimensional driving parameters described in this invention, the step of forming the current joint safety control domain and the next joint safety control domain includes: The time-by-time changes of the combined vehicle speed, longitudinal position and vehicle spacing are tracked along the candidate state sequence. Based on the longitudinal acceleration, the longitudinal motion of the combined vehicle during braking is extrapolated from the boundary. The range of longitudinal positions in which the combined vehicle maintains a distance from adjacent vehicles during braking is determined as the braking distance limit. According to the vehicle controller's stored dimensions of the tractor, trailer, and articulated motion range, the vehicle position and attitude in the candidate state sequence are converted into the tractor's outer profile position, trailer's outer profile position, and relative motion state between the tractor and trailer. Combined with the current lane boundary and the target lane boundary, vehicle outline passage restrictions and relative motion restrictions between the tractor and trailer are formed. The combined range of vehicle motion states, defined by braking distance limits, vehicle outline passage limits, and relative motion limits between the tractor and trailer, is determined as the dynamic safety boundary of the heavy-duty truck. The candidate state sequence is verified hourly, and the candidate control quantities that are continuously within the dynamic safety boundary of the heavy-duty truck are retained. They are then assigned to the current joint safety control domain and the next joint safety control domain according to the lane-changing stage in the combined risk transmission record.

[0013] As a preferred embodiment of the safety evaluation method for autonomous heavy-duty trucks based on multi-dimensional driving parameters described in this invention, the step of determining the continuous carrying state and the carrying interruption stage includes: The starting sampling time of the candidate state sequence is determined as the current evaluation time. Along the candidate state sequence corresponding to the candidate control quantity in the current joint safety control domain, the sampling time when the combined vehicle outline reaches the lane boundary position relationship corresponding to the next lane change stage is located. The remaining time of the stage is determined. The candidate control quantities in the current joint safety control domain and the next joint safety control domain are matched. The control transition time is calculated based on the changes in the longitudinal acceleration control value and the steering angular velocity control value and the corresponding rate of change. Candidate control quantity pairs whose control transition time does not exceed the remaining time of the corresponding stage are selected to form a control quantity correspondence. The control quantity correspondence is tracked along the sequence of track-changing stages. Track-changing stages where the control quantity correspondence continues to exist are identified as continuous acceptance states, and track-changing stages where the control quantity correspondence stops are identified as acceptance interruption stages. The continuous acceptance states and acceptance interruption stages are written into the stage safety acceptance record.

[0014] As a preferred embodiment of the autonomous driving heavy truck safety evaluation method based on multi-dimensional driving parameters described in this invention, the step of determining the real-time safety evaluation result includes: Using longitudinal acceleration and steering angular velocity as control dimensions, the current control command is mapped to the candidate control quantities in the current joint safety control domain. The position of the current control command relative to the boundary of the current joint safety control domain is located along each control dimension, and the candidate control quantity with the smallest difference from the current control command is retained to determine the position of the current control command. The verification is performed in a progressive manner according to the following order: the existence of the current joint security control domain, the current control command being within the current joint security control domain, and the continuous acceptance status. The progressive verification results are associated with the acceptance interruption stage in the stage security acceptance record to determine the range of lane-changing stages where the current control command can maintain safety and the source of risk limitations, thus forming a real-time security evaluation result.

[0015] As a preferred embodiment of the safety evaluation method for autonomous driving heavy trucks based on multi-dimensional driving parameters described in this invention, the step of selecting corresponding candidate control quantities to generate control correction instructions includes: Based on the sources of risk limitations in the real-time security assessment results, the candidate control quantities in the current joint security control domain are screened step by step in the order of reducing the difference in control quantities between the current control command and the current joint security control domain, maintaining the correspondence of control quantities, and shortening the control transition time. The candidate control quantities retained after step-by-step screening are determined as the target candidate control quantities. Using the current control command and the target candidate control quantity as the start and end positions, the control values ​​are arranged step by step between the start and end positions according to the longitudinal acceleration change rate, the steering angular velocity change rate and the corresponding execution resolution to form a control correction sequence. The control value to be executed first in the control correction sequence is selected to generate a control correction command, and the vehicle state after the control correction command is executed is written into the multi-dimensional driving state record of the next control cycle.

[0016] The beneficial effects of this invention are as follows: By identifying the start time of the tractor's lateral response and the peak time of the trailer's response according to the sampling time sequence in the multi-dimensional driving state record, the risk transmission time difference and the peak relationship of the yaw rate between the tractor and the trailer are extracted. Based on the steering direction, speed difference, and steering angle change difference, the reference response process is screened, which can predict the expected peak time of the current trailer response and the expected amplitude of the current trailer response, forming a combined risk transmission record that reflects the lateral risk transmission characteristics of articulated vehicles. Starting from the current control command, candidate control quantities are expanded to deduce the motion state of the combined vehicles. Combined with braking distance limits, vehicle outline passage limits, and relative motion limits between the tractor and the trailer, a joint safety control domain for adjacent lane-changing stages is formed. By comparing the remaining time of the stage with the control transition time, the continuous acceptance state and the acceptance interruption stage are determined, thereby improving the real-time performance and continuity of trailer response risk identification, lane-changing process safety evaluation, and control correction. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a safety evaluation method for autonomous heavy-duty trucks based on multi-dimensional driving parameters.

[0019] Figure 2 This is a schematic diagram of the formation of a combined risk transmission record.

[0020] Figure 3 A schematic diagram of the safety handover record for each stage is generated.

[0021] Figure 4 This is a schematic diagram for real-time safety evaluation and control correction. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figures 1-4 This is one embodiment of the present invention, which provides a method for safety evaluation of autonomous driving heavy trucks based on multi-dimensional driving parameters, including the following steps: S1. Obtain multi-dimensional driving parameters, road environment information and current control commands of the autonomous driving heavy truck, perform spatiotemporal correlation and risk maintenance processing on the multi-dimensional driving parameters and road environment information, and form a multi-dimensional driving status record.

[0026] S1.1 Obtain vehicle speed, vehicle position, vehicle attitude, steering angle, tractor lateral acceleration, tractor yaw rate and trailer yaw rate from vehicle bus and inertial measurement device; obtain vehicle spacing, relative speed, current lane boundary position and target lane boundary position from environmental sensing device; and associate the sampling time with the control cycle in which the current control command is located, maintaining the time relationship between parameter changes and the current control command within the same control cycle.

[0027] Specifically, vehicle speed, vehicle position, and steering angle are sent to the vehicle bus by the vehicle controller, vehicle positioning device, and angle detector of the steering actuator, respectively; vehicle attitude, tractor lateral acceleration, and tractor yaw rate are obtained by the inertial measurement unit of the tractor, and trailer yaw rate is obtained by the inertial measurement unit of the trailer; the vehicle speed, vehicle position, vehicle attitude, steering angle, tractor lateral acceleration, tractor yaw rate, and trailer yaw rate at the corresponding sampling time are aggregated into multi-dimensional driving parameters.

[0028] Vehicle spacing and relative speed are determined based on the perception results of adjacent vehicles by the environmental sensing device. The current lane boundary position and the target lane boundary position are determined based on the lane boundaries identified by the onboard camera. The acquired data is converted to a coordinate reference consistent with the vehicle position and aggregated into road environment information. The current control command, including longitudinal acceleration and steering angular velocity, and its transmission time are obtained from the vehicle controller. The control cycle is defined by the transmission times of the current control command and the next control command. The multi-dimensional driving parameters and road environment information within the control cycle are associated with the current control command, and the original sampling time of each data point is retained.

[0029] S1.2 Based on the vehicle speed and steering angle at the start of the control cycle, the longitudinal acceleration and steering angular velocity in the current control command, and the duration of the control cycle, the range of speed change and the range of lateral position change at the end of the control cycle are deduced to form the achievable range of vehicle movement. Continuous corresponding verification is performed on the changes in vehicle spacing, relative speed, and vehicle position relative to the current lane boundary and the target lane boundary to form a multi-dimensional driving state record.

[0030] Specifically, the control cycle duration, tractor wheelbase, longitudinal acceleration execution resolution, and steering angular velocity execution resolution are read from the vehicle controller. The vehicle speed, lateral position, heading angle, and steering angle at the start of the control cycle are extracted from the multi-dimensional driving parameters. The vehicle motion state is deduced by sampling interval by using the current control command and its adjacent executable control values ​​formed according to the corresponding execution resolution.

[0031] The expression for the deduced vehicle motion state is: ; in, Indicates the sampling time The vehicle speed at that time Indicates the sampling time Vehicle speed at that time Indicates the sampling time The longitudinal acceleration control value at that time, Indicates the sampling time Steering angle at that time Indicates the sampling time The turning angle at that time Indicates the sampling time The value of the steering angular velocity control at that time, Indicates the sampling time The yaw rate of the tractor at that time, subscript Indicates a tractor unit; Indicates the sampling time The heading angle at that time Indicates the sampling time The heading angle at that time Indicates the sampling time The horizontal position at that time Indicates the sampling time The horizontal position at that time Indicates the sampling time The longitudinal position at that time Indicates the sampling time The longitudinal position at that time Indicates the wheelbase of the tractor unit. Indicates the sampling interval.

[0032] By extrapolating each set of control values ​​to the end of the control cycle, the minimum and maximum values ​​of vehicle speed and lateral position are extracted to form the speed variation range and lateral position variation range, respectively. The correlation between vehicle speed and lateral position corresponding to the same set of control values ​​is maintained, and the range of motion state jointly defined by the speed variation range and the lateral position variation range is determined as the range of motion that the vehicle can achieve.

[0033] The vehicle speed sampling values ​​and speed change ranges within the control cycle are verified separately, as are the vehicle lateral position sampling values ​​and lateral position change ranges. Taking the direction of relative speed that causes the vehicle spacing to decrease as the positive direction, the sampling time to be verified and its adjacent sampling times before and after it are determined as three consecutive sampling times. When at least one of the following conditions is met—the vehicle spacing continuously decreases, the relative speed remains positive, and the lateral distance between the vehicle position and either the current lane boundary or the target lane boundary continuously decreases—it is determined that the corresponding road environment information and the sampling values ​​exceeding the corresponding change range maintain a continuous correspondence.

[0034] When either the vehicle speed sample value or the vehicle lateral position sample value exceeds the corresponding range of change and maintains a continuous correspondence, the corresponding sample value is marked as risk change data, and the original value, the original sampling time, and the corresponding current control command are retained, so that the continuous occurrence of real risk changes is not corrected because they exceed the range of movement that the vehicle can achieve.

[0035] If either the vehicle speed sample value or the vehicle lateral position sample value exceeds the corresponding range of change, and the continuous correspondence is not established within three consecutive sampling times, and the sample value deviates from the direction of change maintained by the preceding and following valid sample values ​​only needs to be verified, the corresponding sample value is marked as abnormal data. When valid sample values ​​exist before and after the abnormal data, linear interpolation is performed between the preceding and following valid sample values ​​according to the sampling time interval; when the abnormal data is located at the end of the control cycle and lacks a valid sample value on one side, the two valid sample values ​​closest to that sampling time are linearly extended in the time direction; the current control command, the associated multidimensional driving parameters and road environment information, the risk change data that maintains the original values, and the corrected abnormal data are arranged according to the sampling time to form a multidimensional driving state record.

[0036] S2. Locate the start time of the lateral response of the tractor and the peak time of the trailer response in the multi-dimensional driving status record, determine the risk transmission time difference and the trailer response amplification state, and determine the lane-changing stage corresponding to the response peak according to the positional relationship between the vehicle outline and the lane boundary, forming a combined risk transmission record.

[0037] S2.1. Based on the sampling time sequence of the multi-dimensional driving state record, identify the sampling moment when the lateral acceleration and yaw rate of the tractor vehicle increase synchronously, and determine it as the start time of the lateral response of the tractor vehicle. After the start time of the lateral response of the tractor vehicle, identify the sampling moment when the yaw rate of the trailer changes from increase to decrease, and determine it as the peak time of the trailer response. Calculate the time interval between the start time of the lateral response of the tractor vehicle and the peak time of the trailer response, compare the peak yaw rates of the tractor vehicle and the trailer, and determine the risk transmission time difference and the amplification state of the trailer response, respectively.

[0038] Specifically, the lateral acceleration of the tractor, the yaw rate of the tractor, and the yaw rate of the trailer are read from the multi-dimensional driving status record. Among them, the original values ​​of risk change data are retained, and the corrected sampled values ​​are used for abnormal data. The direction and absolute value changes of the lateral acceleration and yaw rate of the tractor are tracked along the sampling time sequence. The starting sampling time when the two parameters point to the same lateral direction and the absolute value increases synchronously and continuously is determined as the starting time of the lateral response of the tractor.

[0039] Starting from the moment the tractor's lateral response begins, the absolute value of the trailer's yaw rate is tracked. The maximum sampled value formed when the continuous increase turns into a continuous decrease is determined as the peak value of the trailer's yaw rate, and the corresponding sampling moment is determined as the peak moment of the trailer's response. The sampling interval from the moment the tractor's lateral response begins to the peak moment of the trailer's response is determined as the completed risk transmission process, and the corresponding time interval is determined as the historical risk transmission time difference.

[0040] Extract the peak yaw rate of the tractor and the trailer during the completed risk transfer process. When the absolute value of the peak yaw rate of the tractor is zero, mark the corresponding completed risk transfer process as invalid and exclude it from the reference response process screening. When the absolute value of the peak yaw rate of the tractor is not zero, determine the ratio between the absolute value of the peak yaw rate of the trailer and the absolute value of the peak yaw rate of the tractor as the historical trailer response amplification factor. When the historical trailer response amplification factor is greater than one, determine the historical trailer response amplification state as response amplification. When the historical trailer response amplification factor is not greater than one, determine the historical trailer response amplification state as response non-amplification.

[0041] S2.2 Extract the risk transmission process that has appeared at the start time of the tractor's lateral response and the peak time of the trailer's response from the multi-dimensional driving status record. Filter the process in the order of consistent steering direction, decreasing speed difference at the start time of the tractor's lateral response, and decreasing steering angle change difference. The remaining risk transmission process after filtering is determined as the reference response process.

[0042] Specifically, the completed risk transfer process, which has determined the start time of the tractor's lateral response and the peak time of the trailer's response, is read from the multidimensional driving status record. The completed risk transfer processes participating in the screening all include vehicle speed, steering angle, tractor yaw rate, trailer yaw rate, historical risk transfer time difference, historical trailer response amplification coefficient, and historical trailer response amplification status within the corresponding sampling interval, and the absolute value of the peak value of the tractor yaw rate is not zero. If there is a completed risk transfer process in the multidimensional driving status record that has the same steering direction as the current tractor's lateral response and a valid peak relationship, the reference response process screening is performed; if not, the current trailer response prediction based on the reference response process is not performed, and the trailer response peak value is identified along the current sampling time sequence. After the trailer response peak value is formed, the corresponding process is written into the completed risk transfer process.

[0043] Based on the direction of the yaw rate and the direction of the steering angle change of the tractor at the start of the lateral response, the steering direction of each risk transmission process is determined, and the steering direction of the current lateral response of the tractor is determined in the same way, retaining risk transmission processes with consistent steering directions.

[0044] The absolute difference between the vehicle speed at the start of the tractor's lateral response and the vehicle speed at the current start of the tractor's lateral response in each risk transmission process is defined as the vehicle speed difference; the absolute difference between the change in the steering angle at the start of the tractor's lateral response relative to the steering angle at the start of the control cycle and the change in the steering angle corresponding to the current tractor's lateral response is defined as the steering angle change difference.

[0045] The risk transmission process with the smallest speed difference is retained. If there are multiple risk transmission processes with the same speed difference, the risk transmission process with the smallest steering angle change difference is retained. If both the speed difference and the steering angle change difference are the same, the risk transmission process with the trailer response peak time closest to the current evaluation time is retained, and the finally retained risk transmission process is determined as the reference response process.

[0046] S2.3 Align the start time of the lateral response of the tractor in the reference response process with the start time of the lateral response of the current tractor. Based on the time interval from the start time of the lateral response of the tractor to the peak time of the trailer response in the reference response process, determine the expected peak time of the current trailer response in the current sampling time series, and determine the time interval from the start time of the lateral response of the current tractor to the expected peak time of the current trailer response as the risk transmission time difference.

[0047] Specifically, the starting time of the lateral response of the tractor, the peak time of the trailer response, and the historical risk transmission time difference are read from the reference response process, and the starting time of the current lateral response of the tractor and the current sampling time sequence are read from the multi-dimensional driving status record.

[0048] Keeping the time interval between each sampling moment and the starting moment of the tractor's lateral response in the reference response process unchanged, the starting moment of the tractor's lateral response in the reference response process is made to coincide with the starting moment of the current tractor's lateral response. Taking the starting moment of the current tractor's lateral response as the starting point, the historical risk transmission time difference is extended along the current sampling time sequence. When the end point of the extension coincides with the sampling moment, the corresponding sampling moment is determined as the expected peak moment of the current trailer response. When the end point of the extension is located between adjacent sampling moments, the first sampling moment no earlier than the end point of the extension is determined as the expected peak moment of the current trailer response.

[0049] The time interval between the start of the lateral response of the current tractor and the expected peak time of the current trailer response is determined as the current expected risk transmission time difference; for the risk transmission process in which the current trailer response has not yet reached its peak, the risk transmission time difference specifically represents the current expected risk transmission time difference.

[0050] S2.4. According to the change ratio of the current peak yaw rate of the tractor relative to the peak yaw rate of the tractor during the reference response process, adjust the peak yaw rate of the trailer during the reference response process, determine the expected response amplitude of the trailer, compare the expected response amplitude of the trailer with the current peak yaw rate of the tractor, and determine the amplification state of the trailer response.

[0051] Specifically, when the yaw rate of the tractor unit has changed from increasing to decreasing and the yaw rate of the trailer unit has not yet reached its peak, the peak value of the current yaw rate of the tractor unit is extracted from the multi-dimensional driving state record, and the peak values ​​of the yaw rates of the tractor unit and the trailer unit are extracted from the reference response process; the absolute value of the peak value of the yaw rate of the tractor unit in the reference response process is not zero; when the absolute value of the peak value of the current yaw rate of the tractor unit is zero, the current identification result is marked as an invalid lateral response, and the calculation of the expected response amplitude of the current trailer unit is not performed.

[0052] When the absolute value of the current yaw rate peak value of the tractor is not zero, adjust the peak value of the trailer yaw rate in the reference response process according to the change ratio of the current peak value of the tractor's yaw rate peak value relative to the peak value of the tractor's yaw rate peak value in the reference response process, and determine the expected response amplitude of the current trailer.

[0053] The expression for determining the expected response amplitude of the current trailer is: ; In the formula, This indicates the expected response amplitude of the trailer at present. This indicates the current peak yaw rate of the tractor unit. This indicates the peak value of the trailer yaw rate during the reference response process. This indicates the peak yaw rate of the tractor during the reference response process, where the subscript... Indicates trailer, subscript Indicates tractor unit, subscript Indicates the current lateral response of the tractor unit, subscript This indicates a reference response process.

[0054] The ratio between the current expected response amplitude of the trailer and the absolute value of the peak value of the current yaw rate of the tractor is determined as the current trailer response amplification factor. When the current trailer response amplification factor is greater than one, the current expected trailer response amplification state is determined as response amplification. When the current trailer response amplification factor is not greater than one, the current expected trailer response amplification state is determined as response not amplified.

[0055] S2.5. Based on the position of the tractor and trailer outlines relative to the current lane boundary and the target lane boundary at the peak of the trailer response, and combined with the sequential relationship and lateral position changes of the tractor and trailer outlines crossing the lane boundaries, determine the lane-changing stage corresponding to the peak of the trailer response, and form a combined risk transmission record.

[0056] Specifically, a risk transmission process has been established at the peak moment of trailer response. The vehicle position, vehicle attitude, current lane boundary position and target lane boundary position corresponding to the peak moment of trailer response are read from the multi-dimensional driving status record. Based on the outer dimensions of the tractor and trailer, the articulation position of the tractor and trailer and the corresponding articulation angle, the outer dimensions of the tractor and trailer at the peak moment of trailer response are determined.

[0057] When the trailer response has not yet reached its peak, the current vehicle status and current control commands are read from the multi-dimensional driving status record. According to the vehicle motion state deduction process used when the vehicle's achievable motion range is formed, the tractor position and tractor attitude are deduced to the expected peak moment of the current trailer response. According to the relationship between the trailer yaw rate and the peak value of the trailer yaw rate during the reference response process, the trailer yaw change is adjusted with the expected response amplitude of the current trailer, and the articulation angle, trailer position and trailer attitude are updated hourly to determine the tractor outline and trailer outline at the expected peak moment of the current trailer response.

[0058] By comparing the positions of the tractor and trailer outlines relative to the current lane boundary and the target lane boundary at the peak of the trailer response and the expected peak of the current trailer response, and combining the sequential relationship and lateral position changes of the tractor and trailer outlines crossing the lane boundaries, the lane-changing phases corresponding to the peak of the trailer response and the expected peak of the current trailer response are determined.

[0059] The lateral response start time of the tractor unit, the peak response time of the trailer unit, the historical risk transmission time difference, the historical trailer response amplification factor, the historical trailer response amplification status and the corresponding lane-changing stage of the completed risk transmission process are associated with the reference response process of the current incomplete risk transmission process, the current lateral response start time of the tractor unit, the current expected peak response time of the trailer unit, the current expected risk transmission time difference, the current expected response amplitude of the trailer unit, the current trailer response amplification factor, the current expected trailer response amplification status and the lane-changing stage corresponding to the current expected peak response time of the trailer unit, according to the sampling time, to form a combined risk transmission record.

[0060] S3. Based on the combined risk transmission record, taking the current control command as the starting point, the control values ​​are expanded within the longitudinal acceleration execution boundary and the steering angular velocity execution boundary according to the corresponding execution resolution to form candidate control quantities. The combined vehicle motion state corresponding to the candidate control quantities is deduced hourly to form the current joint safety control domain and the next joint safety control domain respectively. The corresponding candidate control quantities are matched according to the control quantity change capability to form a stage safety handover record.

[0061] S3.1 Read the longitudinal acceleration execution boundary, steering angular velocity execution boundary, longitudinal acceleration rate of change, steering angular velocity rate of change and corresponding execution resolution given by the vehicle controller. Based on the longitudinal acceleration and steering angular velocity in the current control command, select longitudinal acceleration control values ​​and steering angular velocity control values ​​step by step in the direction of numerical increase and numerical decrease according to the corresponding execution resolution, and combine them to form candidate control quantities.

[0062] Specifically, the longitudinal acceleration execution boundary, steering angular velocity execution boundary, longitudinal acceleration rate of change, steering angular velocity rate of change, longitudinal acceleration execution resolution, and steering angular velocity execution resolution corresponding to the current vehicle state are read from the vehicle controller. The longitudinal acceleration execution boundary and steering angular velocity execution boundary respectively limit the value range of the corresponding control value, the longitudinal acceleration rate of change and steering angular velocity rate of change limit the change amplitude of the corresponding control value per unit time, and the longitudinal acceleration execution resolution and steering angular velocity execution resolution limit the value interval between adjacent executable control values.

[0063] The longitudinal acceleration and steering angular velocity in the current control command are used as the starting points for the expansion of the corresponding control dimensions. Control values ​​are selected step by step along the direction of numerical increase and numerical decrease according to the corresponding execution resolution until the corresponding execution boundary is reached, forming the longitudinal acceleration control value sequence and the steering angular velocity control value sequence respectively. The control values ​​in the two control value sequences are combined respectively, and the control value combination corresponding to the current control command is retained together to form candidate control quantities. The longitudinal acceleration rate of change, the steering angular velocity rate of change, and the corresponding execution resolution are associated with the candidate control quantities.

[0064] S3.2. Starting with the current vehicle state in the multi-dimensional driving state record, the combined vehicle motion state corresponding to each candidate control quantity is deduced sequentially. The trailer motion state is adjusted according to the combined risk transmission record to form a candidate state sequence. The braking distance, vehicle outline position and relative motion limit between the tractor and trailer within the candidate state sequence are used to form the dynamic safety boundary of the heavy truck. Candidate control quantities that have not exceeded the dynamic safety boundary of the heavy truck are retained to form the current joint safety control domain and the next joint safety control domain.

[0065] Specifically, the vehicle speed, vehicle position, vehicle attitude, steering angle, tractor lateral acceleration, tractor yaw rate, and trailer yaw rate at the current evaluation moment are determined from the multi-dimensional driving state record as the current vehicle state. The vehicle spacing, relative speed, current lane boundary position, and target lane boundary position are also read at the same sampling moment. Each candidate control variable is derived from the current vehicle state, ensuring that the derived results reflect the differences in control values ​​between the candidate control variables.

[0066] The longitudinal acceleration control value and steering angular velocity control value in each candidate control variable are applied to the current vehicle state. According to the sampling interval used in the multi-dimensional driving state record, the vehicle speed, steering angle, tractor yaw rate, heading angle, longitudinal position and lateral position are updated hourly using the vehicle motion state expression to form the tractor motion state corresponding to each candidate control variable. The time domain of the simulation covers the current lane change stage and the next lane change stage.

[0067] Within the current evaluation time domain, based on the vehicle spacing and relative speed at the current evaluation moment, the longitudinal position and vehicle spacing of adjacent vehicles are updated hourly according to the approach direction and amplitude represented by the relative speed. The current lane boundary position and the target lane boundary position remain unchanged under the road coordinate reference. After entering the next control cycle, the vehicle spacing, relative speed, current lane boundary position, and target lane boundary position are reread using the multi-dimensional driving state record of the next control cycle, and the candidate states are re-evaluated. The hourly updated tractor motion state, adjacent vehicle longitudinal position, vehicle spacing, current lane boundary position, and target lane boundary position are associated according to the sampling time to establish the candidate state sequence sampling time sequence corresponding to each candidate control quantity. The trailer motion state is adjusted according to the combined risk transmission record, and the trailer motion state is written into the corresponding sampling time to form the candidate state sequence.

[0068] S3.3. Align the peak moment of the trailer response in the reference response process with the expected peak moment of the current trailer response. Combine the expected amplitude of the current trailer response with the peak relationship between the yaw rate of the tractor and the trailer in the reference response process to correct the amplitude of the trailer yaw rate, thus forming a trailer response correction sequence.

[0069] Specifically, the reference response process, the expected peak time of the current trailer response, the expected amplitude of the current trailer response, the amplification factor of the current trailer response, and the amplification state of the current trailer response are read from the combined risk transmission record. The trailer yaw rate, the corresponding sampling time, and the peak time of the trailer response are extracted from the reference response process.

[0070] For each candidate control quantity, the yaw rate and its peak value of the tractor are extracted from the corresponding tractor motion state. According to the change ratio of the peak value of the tractor yaw rate corresponding to the candidate control quantity to the current peak value of the tractor yaw rate, the current trailer's expected response amplitude is adjusted to form the trailer's expected response amplitude of the corresponding candidate control quantity. The amplitude relationship between the trailer's expected response amplitude of the corresponding candidate control quantity and the peak value of the tractor yaw rate is verified using the current trailer response amplification factor.

[0071] Based on the time interval between the peak moment of the trailer response during the reference response process and the expected peak moment of the current trailer response, the sampling time of the trailer yaw rate during the reference response process is shifted as a whole to make the peak moment of the trailer response coincide with the expected peak moment of the current trailer response, while maintaining the interval and order between each sampling moment; the expected amplitude of the trailer response corresponding to each candidate control variable is used as the target peak value, and each sampling value is synchronously corrected according to the amplitude relationship between each trailer yaw rate and the peak value of the trailer yaw rate during the reference response process, forming a trailer response correction sequence corresponding to each candidate control variable.

[0072] When the trailer response is expected to be amplified, verify that the peak value of the trailer yaw rate in the trailer response correction sequence is greater than the peak value of the corresponding tractor yaw rate; when the trailer response is expected to be unamplified, verify that the peak value of the trailer yaw rate is not greater than the peak value of the corresponding tractor yaw rate; if the corresponding amplitude relationship is not maintained, readjust the expected response amplitude of the trailer for the corresponding candidate control quantity according to the current trailer response amplification coefficient.

[0073] S3.4 Write the trailer response correction sequence into the candidate state sequence. Update the articulation angle between the tractor and the trailer by accumulating the difference between the yaw rate of the tractor and the yaw rate in the trailer response correction sequence. Accumulate the yaw rate in the trailer response correction sequence hourly. Use the articulation position of the tractor and the trailer stored in the vehicle controller as the connection reference to update the trailer position and trailer attitude in the candidate state sequence.

[0074] Specifically, according to the candidate control quantity and the sampling time, the response correction sequence of each trailer is written into the candidate state sequence of the corresponding candidate control quantity, so that the yaw rate of the tractor and the yaw rate of the trailer in the same candidate state sequence maintain the corresponding relationship at the sampling time.

[0075] For each candidate state sequence, starting from the current articulation angle, the difference between the tractor's yaw rate and the trailer's yaw rate in the corresponding trailer response correction sequence is accumulated hourly according to the sampling interval to form the articulation angle at each sampling time; starting from the current trailer attitude, the trailer attitude at each sampling time is updated by accumulating the trailer's yaw rate in the trailer response correction sequence according to the sampling interval.

[0076] Using the articulated positions of the tractor and trailer stored in the vehicle controller as the connection reference, and combining the tractor position, tractor attitude, articulation angle and trailer connection dimensions at each sampling time, the corresponding trailer position is determined; the articulation angle, trailer position and trailer attitude are written into the corresponding candidate state sequence, so that each candidate state sequence contains the tractor motion state and trailer motion state corresponding to the same candidate control quantity.

[0077] S3.5. Track the hourly changes in the combined vehicle speed, longitudinal position and vehicle spacing along the candidate state sequence, perform boundary extrapolation of the longitudinal motion of the combined vehicle during braking based on the longitudinal acceleration, and determine the range of longitudinal positions in which the combined vehicle maintains a distance from adjacent vehicles during braking as the braking distance limit.

[0078] Specifically, the combined vehicle speed, longitudinal position, vehicle spacing, and relative speed are read along each candidate state sequence. The longitudinal position and vehicle spacing of adjacent vehicles are updated hourly according to the relative speed at the current evaluation time and the sampling interval of the candidate state sequence. The original values ​​of risk change data are retained, and the sampled values ​​of abnormal data are corrected. The combined vehicle speed and longitudinal position at each sampling time in the candidate state sequence are used as the starting point for braking simulation. The longitudinal acceleration control value in the candidate control quantity is adjusted to the braking direction boundary value in the longitudinal acceleration execution boundary according to the longitudinal acceleration change rate. The combined vehicle speed and longitudinal position are updated according to the sampling interval until the combined vehicle stops, forming the braking motion process of the corresponding candidate control quantity.

[0079] The longitudinal distance between the front end of the tractor and the adjacent vehicle in front, and the longitudinal distance between the rear end of the trailer and the adjacent vehicle behind are tracked along the braking process. Based on the condition that both longitudinal distances remain positive, the upper and lower limits of the longitudinal position that the combined vehicle can reach at each sampling time are determined. The range between the upper and lower limits of the longitudinal position is associated according to the sampling time to form a braking distance limit, which is used to verify whether the combined vehicle in the candidate state sequence has longitudinal collision avoidance space.

[0080] S3.6. Based on the tractor's outer dimensions, trailer's outer dimensions, and articulated motion range stored in the vehicle controller, convert the vehicle position and vehicle attitude in the candidate state sequence into the tractor's outer dimensions, trailer's outer dimensions, and the relative motion state between the tractor and trailer. Combine the current lane boundary and the target lane boundary to form vehicle outer dimensions passage restrictions and relative motion restrictions between the tractor and trailer.

[0081] Specifically, the vehicle controller reads the tractor's outer dimensions, trailer's outer dimensions, and articulation range. The tractor's outer dimensions and trailer's outer dimensions include the vehicle's length, width, and the position of the vehicle reference point relative to the outer edge, respectively. The articulation range represents the range of articulation angles that can be formed between the longitudinal direction of the tractor and the longitudinal direction of the trailer.

[0082] The position, attitude, position, attitude, and articulation angle of the tractor and trailer are read at each sampling time along the candidate state sequence. Based on the outer dimensions of the tractor and trailer, the positions of the tractor and trailer are converted into the outer dimensions of the tractor and trailer, respectively. The current lane boundary position and the target lane boundary position are converted to the same coordinate reference. Based on the lane change stage in the combined risk transmission record, the lane boundaries that can be crossed and the outer lane boundaries that cannot be crossed are determined.

[0083] The positional relationship between the outer contours of the tractor and trailer and the lane boundary is compared at each sampling time. The range of vehicle movement states where neither the outer contour of the tractor nor the outer contour of the trailer exceeds the common coverage of the current lane and the target lane, and the order of the outer contour crossing is consistent with the lane change phase, is determined as the vehicle outer contour passage restriction.

[0084] The articulation angle between the tractor and trailer is determined based on the tractor's and trailer's postures at the same sampling time, and the direction of change of the articulation angle is tracked to form the relative motion state between the tractor and trailer. The range of combined vehicle motion states where the articulation angle is within the range of articulation motion and will not exceed the range of articulation motion at adjacent sampling times according to the current direction of change is determined as the relative motion limit between the tractor and trailer.

[0085] S3.7. The combined range of vehicle motion states defined by braking distance limit, vehicle outline passage limit and relative motion limit between tractor and trailer is determined as the dynamic safety boundary of heavy truck. The candidate state sequence is verified hourly, the candidate control quantity that is continuously within the dynamic safety boundary of heavy truck is retained, and the lane change stage in the combined risk transmission record is assigned to the current joint safety control domain and the next joint safety control domain.

[0086] Specifically, the sampling time is associated with braking distance limits, vehicle outline passage limits, and relative motion limits between the tractor and trailer. The braking distance limit defines the range of longitudinal positions that the combined vehicle can reach during braking. The vehicle outline passage limit defines the range of vehicle positions and vehicle attitudes that the outlines of the tractor and trailer can pass through. The relative motion limit between the tractor and trailer defines the range of states that can be reached by the articulation angle and the changes in the articulation angle.

[0087] At each sampling time, the range of motion of the combined vehicle that simultaneously meets the braking distance limit, vehicle outline passage limit, and relative motion limit between the tractor and trailer is determined as the dynamic safety boundary of the heavy truck at the corresponding sampling time. The dynamic safety boundary of the heavy truck is updated hourly with the speed of the combined vehicle, the position of adjacent vehicles, the positional relationship between the vehicle outline and the lane boundary, and the change of the articulation angle. It is used to represent the range of motion of the combined vehicle that can maintain longitudinal spacing, lane passage, and articulation stability at the corresponding sampling time.

[0088] Read the lane-changing phase corresponding to the trailer response peak from the combined risk transmission record, and determine the current lane-changing phase corresponding to the current evaluation moment and the next lane-changing phase adjacent to the current lane-changing phase according to the order of changes in the positional relationship between the combined vehicle outline and the lane boundary. According to the lane-changing phase, divide each candidate state sequence into the candidate state sequence segment corresponding to the current lane-changing phase and the candidate state sequence segment corresponding to the next lane-changing phase.

[0089] The motion state of the combined vehicle is verified by sampling each time step along the candidate state sequence segment corresponding to the current lane-changing stage. When the motion state of the combined vehicle within the candidate state sequence segment is within the corresponding dynamic safety boundary of the heavy truck at all sampling times, the corresponding candidate control quantity is retained and included in the current joint safety control domain. When the motion state of the combined vehicle exceeds the dynamic safety boundary of the heavy truck for the first time, the corresponding candidate control quantity is not included in the current joint safety control domain, and the type of restriction breached is recorded.

[0090] All candidate control quantities are identified as candidate control quantities to be verified in the next joint safety control domain. The next joint safety control domain is not determined by directly executing the candidate state sequence formed by the candidate control quantities to be verified from the current vehicle state. Instead, it forms candidate control quantity pairs with the candidate control quantities to be verified in the current joint safety control domain, and deduces the actual control transition process according to the control transition time. It also continues to deduce the next lane change stage from the combined vehicle motion state after the control transition. Only the candidate control quantities to be verified that are continuously within the dynamic safety boundary of the heavy truck during the control transition process and the next lane change stage are included in the next joint safety control domain.

[0091] The current joint safety control domain consists of candidate control quantities and their candidate state sequence segments in the current lane change phase; the next joint safety control domain consists of candidate control quantities verified through control transition and the candidate state sequence segments formed by the corresponding candidate control quantity pairs in the next lane change phase.

[0092] S3.8. Determine the starting sampling time of the candidate state sequence as the current evaluation time. Along the candidate state sequence corresponding to the candidate control quantity in the current joint safety control domain, locate the sampling time when the combined vehicle outline reaches the lane boundary position relationship corresponding to the next lane change stage. Determine the remaining time of the stage. Match the candidate control quantities in the current joint safety control domain and the next joint safety control domain. Calculate the control transition time based on the changes in longitudinal acceleration control value and steering angular velocity control value and the corresponding rate of change.

[0093] Specifically, candidate control quantities and their corresponding candidate state sequences are read from the current joint safety control domain. Since each candidate state sequence starts with the current vehicle state in the multi-dimensional driving state record, the starting sampling time of the candidate state sequence is uniformly determined as the current evaluation time, so that the remaining time of the stage corresponding to different candidate control quantities adopts the same time starting point.

[0094] The current lane-changing phase is read from the combined risk transmission record. Following the order of changes in the positional relationship between the combined vehicle outline and the lane boundary, the positional relationship of the lane boundary in the adjacent lane-changing phase after the current lane-changing phase is extracted to determine the positional relationship of the lane boundary corresponding to the next lane-changing phase. The positional relationship between the outline of the tractor and the outline of the trailer and the current lane boundary and the target lane boundary is compared at each sampling time along each candidate state sequence. The sampling time when the outline of the combined vehicle first reaches the positional relationship of the lane boundary corresponding to the next lane-changing phase is determined as the phase transition time.

[0095] The time interval between the current evaluation time and the stage transition time is calculated, and this time interval is determined as the stage remaining time of the corresponding candidate control quantity. The correlation between the stage remaining time and the corresponding candidate control quantity is maintained. When the lane boundary position relationship corresponding to the next lane change stage does not appear in the candidate state sequence, the stage transition time of the corresponding candidate control quantity is not formed, and the corresponding candidate control quantity is excluded from the control transition matching between the current lane change stage and the next lane change stage.

[0096] Each candidate control quantity in the current joint safety control domain is combined with the candidate control quantity to be verified in the next joint safety control domain to form a candidate control quantity pair from the candidate control quantity of the current lane change phase to the candidate control quantity to be verified in the next lane change phase; the longitudinal acceleration control value, steering angular velocity control value, corresponding execution resolution, corresponding rate of change, and the phase remaining time of the candidate control quantity of the current lane change phase are read respectively, and the corresponding control transition time and control carryover margin are calculated.

[0097] The expressions for calculating the control transition time and control margin are as follows: ; in, This indicates the first [number] in the current joint security control domain. The candidate control variable is transferred to the next joint security control domain. The control transition time required for each candidate control variable Indicates the longitudinal acceleration execution resolution. Indicates the resolution of the steering angular velocity. Represents the rate of change of longitudinal acceleration amplitude, Represents the rate of change of steering angular velocity amplitude, Indicates the first The remaining time for each candidate control quantity in the stage, indicated by the superscript. Indicates the remaining time for the stage; Indicates the first in the current joint security control domain The candidate control variable is transferred to the next joint security control domain. The control margin corresponding to each candidate control quantity; Indicates the first The longitudinal acceleration control value of the candidate control variable Indicates the first The longitudinal acceleration control value of the candidate control variable Indicates the first Steering angular velocity control values ​​of candidate control quantities Indicates the first The steering angular velocity control value of the candidate control quantity.

[0098] Both longitudinal acceleration execution resolution and steering angular velocity execution resolution are positive values; when there is no change in control value in a control dimension for a candidate control quantity pair, the adjustment time of that control dimension is set to zero; when the corresponding control dimension needs to change the control value but the corresponding rate of change is zero, the candidate control quantity pair is determined to be non-convertible and the control margin is no longer calculated; when the control margin is not less than zero, the candidate control quantity pair can complete the control value conversion before the combined vehicle enters the next lane-changing phase.

[0099] The control value change is converted into the number of control execution levels to be crossed according to the corresponding execution resolution. Then, the longitudinal acceleration adjustment time and steering angular velocity adjustment time are determined according to the corresponding rate of change. The maximum value of the longitudinal acceleration adjustment time and steering angular velocity adjustment time is taken as the control transition time to ensure that both control dimensions have been adjusted. The control transition time is deducted from the remaining time of the stage to form a control margin. A control margin of not less than zero indicates that the corresponding candidate control quantity pair can be converted before the combined vehicle enters the next lane-changing stage.

[0100] S3.9. Screen candidate control quantity pairs whose control transition time does not exceed the remaining time of the corresponding stage, form a control quantity correspondence, trace the control quantity correspondence along the sequence of track-changing stages, determine the track-changing stage where the control quantity correspondence continues to exist as the continuous acceptance state, and determine the track-changing stage where the control quantity correspondence stops as the acceptance interruption stage, and write the continuous acceptance state and the acceptance interruption stage into the stage safety acceptance record.

[0101] Specifically, candidate control quantity pairs with a control carryover margin of not less than zero are retained, and the stage transition time is located along the candidate state sequence corresponding to the candidate control quantity of the current lane-changing stage. From the stage transition time, the sampling time that is closest to the current evaluation time and whose time interval between the stage transition time is not less than the control transition time is determined as the control transition start time.

[0102] Extract the vehicle speed, vehicle position, vehicle attitude, steering angle, tractor yaw rate, trailer yaw rate, and articulation angle corresponding to the control transition start time from the candidate state sequence, and determine the extracted combined vehicle motion state as the transition start point of the corresponding candidate control quantity pair.

[0103] Starting from the transition point, the candidate control quantities for the current lane-changing stage are adjusted hourly according to the longitudinal acceleration change rate, steering angular velocity change rate, and corresponding execution resolution to the candidate control quantities to be verified in the next lane-changing stage. At each sampling moment, the tractor motion state is updated using the candidate vehicle motion state deduction process, and the trailer yaw rate, articulation angle, trailer position, and trailer attitude are updated using the trailer motion state adjustment process. At the same time, the actual stage transition time is repositioned based on the updated combined vehicle outline.

[0104] If the candidate control quantities to be verified for the next lane change phase are adjusted before the actual phase transition time, and the combined vehicle motion state during the control transition process remains within the dynamic safety boundary of the heavy truck, the combined vehicle motion state at the actual phase transition time is determined as the starting point for the deduction of the next lane change phase; the combined vehicle motion state is further deduced according to the candidate control quantities to be verified for the next lane change phase, and the braking distance limit, vehicle outline passage limit, and relative motion limit between the tractor and trailer are verified hourly.

[0105] When the combined vehicle motion state during the control transition process and the next lane change phase remains within the dynamic safety boundary of the heavy truck, the corresponding candidate control quantity to be verified is assigned to the next joint safety control domain, and the transition start point, transition end point, phase remaining time, control transition time, and control carry-over margin of the candidate control quantity pair are associated to form a control quantity correspondence relationship.

[0106] When there is at least one control quantity correspondence between the current joint security control domain and the next joint security control domain, the continuous acceptance state is determined to continue; when there is no control quantity correspondence, the next lane change phase is determined to be the acceptance interruption phase; the current joint security control domain, the next joint security control domain, the control quantity correspondence, the remaining phase time, the control transition time, the control acceptance margin, the continuous acceptance state, and the acceptance interruption phase are written into the phase security acceptance record.

[0107] S4. Based on the phased security handover record, verify the existence of the current joint security control domain, the location of the current control command, and the continuous handover status, determine the real-time security evaluation result, and select the corresponding candidate control quantity to generate control correction commands.

[0108] S4.1 Using longitudinal acceleration and steering angular velocity as control dimensions respectively, perform position mapping between the current control command and the candidate control quantities in the current joint safety control domain. Locate the position of the current control command relative to the boundary of the current joint safety control domain along each control dimension, and retain the candidate control quantity with the smallest difference from the current control command to determine the position of the current control command.

[0109] Specifically, the existence status of the current joint safety control domain is read from the phase safety handover record; if there is no candidate control quantity in the current joint safety control domain, the position mapping between the current control command and the current joint safety control domain is not executed, the position of the current control command is recorded as an unmappable position, and the state verification without a safety candidate control quantity is entered; if there is a candidate control quantity in the current joint safety control domain, the longitudinal acceleration is used as one control dimension and the steering angular velocity is used as another control dimension, and the longitudinal acceleration control value and steering angular velocity control value in each candidate control quantity are mapped to the corresponding control position, and the longitudinal acceleration and steering angular velocity in the current control command are mapped to the same control value space.

[0110] When there are candidate control quantities in the current joint safety control domain, the connection relationship between adjacent candidate control quantities in the current joint safety control domain is checked according to the longitudinal acceleration execution resolution and the steering angular velocity execution resolution. When there are no adjacent candidate control quantities in at least one of the following directions: longitudinal acceleration increasing direction, longitudinal acceleration decreasing direction, steering angular velocity increasing direction, and steering angular velocity decreasing direction, the control position of the corresponding candidate control quantity in that direction is determined as the boundary of the current joint safety control domain.

[0111] Along both the longitudinal acceleration control dimension and the steering angular velocity control dimension, the control position of the current control command is compared with the boundary of the current joint safety control domain. If the candidate control quantity corresponding to the current control command exists in the current joint safety control domain, the position of the current control command is determined as either an in-domain position or a boundary position, depending on whether the current control command is located at the boundary of the current joint safety control domain. If the candidate control quantity corresponding to the current control command does not exist in the current joint safety control domain, the direction of increase or decrease of the value of the current control command relative to the boundary of the current joint safety control domain is determined along each control dimension, and the position of the current control command is determined as an out-of-domain position.

[0112] For each candidate control quantity in the current joint safety control domain, the number of longitudinal acceleration execution resolutions and the number of steering angular velocity execution resolutions that differ between the candidate control quantity and the current control command are counted respectively. The sum of the corresponding execution resolutions is determined as the control quantity difference number. The candidate control quantity that reaches the minimum control quantity difference number is determined as the candidate control quantity with the smallest difference from the current control command.

[0113] The location of the current control command is determined by associating the corresponding position status in the domain, boundary position, and external position, the direction of increase or decrease of the value of the current control command relative to the boundary of the current joint security control domain, the number of control quantity differences of each candidate control quantity, and the candidate control quantity with the smallest difference from the current control command.

[0114] S4.2. Perform progressive verification in the order of the existence of the current joint security control domain, the current control command being within the current joint security control domain, and the continuous acceptance status. Associate the progressive verification results with the acceptance interruption stage in the stage security acceptance record to determine the range of lane-changing stages where the current control command can maintain safety and the source of risk limitations, and form a real-time security evaluation result.

[0115] Specifically, progressive verification means verifying each item in the order of whether the current joint security control domain exists, whether the current control command is within the current joint security control domain, and whether the continuous acceptance status is maintained. If the current verification item is satisfied, the next item is verified. If the current verification item is not satisfied, the scope of the safe lane-changing phase is not extended to the next item.

[0116] Verify whether there are candidate control quantities in the current joint safety control domain. If there are no candidate control quantities in the current joint safety control domain, it indicates that the candidate state sequence corresponding to each candidate control quantity has exceeded the dynamic safety boundary of the heavy truck in the current lane change phase. It is determined that the current control command cannot maintain safety in the current lane change phase and forms a state of no safe candidate control quantity. Read the braking distance limit, vehicle outline passage limit and relative motion limit between tractor and trailer corresponding to each candidate control quantity being excluded. The restriction type that causes the current joint safety control domain to be empty is identified as the source of risk restriction. Write the state of no safe candidate control quantity and the source of risk restriction into the real-time safety evaluation result as the basis for screening the target candidate control quantity from all candidate control quantities.

[0117] When candidate control variables exist in the current joint safety control domain, the position of the current control command is further verified: if the current control command is located within or at the boundary of the current joint safety control domain, it indicates that the current control command can maintain safety during the current lane-changing phase, and the continuous acceptance status is further verified; if the current control command is located outside the domain, it indicates that a safety control value exists within the current joint safety control domain, but the combined vehicle motion state corresponding to the current control command cannot continuously meet the dynamic safety boundary of heavy trucks, and the direction of the current control command relative to the boundary of the current joint safety control domain and the corresponding type of limitation breach are identified as the source of risk limitation.

[0118] When the current control command is within the current joint safety control domain, read the control quantity correspondence with the candidate control quantity corresponding to the current control command as the starting point of the transition from the phase safety handover record; if there is at least one control quantity correspondence with the candidate control quantity corresponding to the current control command as the starting point of the transition, and the continuous handover state continues, determine the current lane change phase and the next lane change phase as the lane change phase range in which the current control command can maintain safety.

[0119] If there is a control quantity correspondence between the current joint security control domain and the next joint security control domain, but there is no control quantity correspondence with the candidate control quantity corresponding to the current control instruction as the starting point of the transition, then the current lane-changing stage is determined as the lane-changing stage range in which the current control instruction can maintain safety, and the next lane-changing stage is determined as the interruption stage corresponding to the current control instruction.

[0120] For candidate control quantity pairs that take the candidate control quantity corresponding to the current control command as the starting point for conversion but have not formed a corresponding relationship, the corresponding control conversion time, the remaining time of the stage, and the dynamic safety boundary of the heavy truck that is breached during the control conversion process are read. When the control conversion time exceeds the remaining time of the stage, the insufficient control quantity change capability is identified as the source of risk limitation. When the combined vehicle motion state breaks through the dynamic safety boundary of the heavy truck during the control conversion process, the corresponding breached limitation is identified as the source of risk limitation. When there is no candidate control quantity to be verified that can be classified into the next joint safety control domain, the limitation type corresponding to the candidate control quantity to be verified being excluded is identified as the source of risk limitation.

[0121] The current state of the joint security control domain, the location of the current control command, the continuous acceptance status, the range of lane-changing phases where the current control command can maintain safety, the acceptance interruption phase, and the source of risk limitation are associated with the current evaluation time to form a real-time security evaluation result. The real-time security evaluation result is used to determine whether the current control command needs to be modified and to screen target candidate control quantities according to the source of risk limitation.

[0122] S4.3. Based on the sources of risk limitations in the real-time security evaluation results, the candidate control quantities in the current joint security control domain are screened step by step in the order of reducing the difference in control quantities between the current control command and the current joint security control domain, maintaining the correspondence of control quantities, and shortening the control transition time. The candidate control quantities retained after step-by-step screening are determined as the target candidate control quantities.

[0123] Specifically, the current joint security control domain status, the status of no security candidate control quantity, and the source of risk limitation are read from the real-time security evaluation results, and the current joint security control domain, the next joint security control domain, the correspondence of control quantities, and the control transition time are read from the phase security handover record.

[0124] When a joint safety control domain exists, candidate control quantities and corresponding candidate state sequences are read from the current joint safety control domain, and the candidate state sequence corresponding to the current control command is also read. When the source of risk limitation is braking distance limitation, the forward longitudinal interval and backward longitudinal interval of each candidate state sequence are compared with the candidate state sequence corresponding to the current control command at the same sampling time, and candidate control quantities whose minimum longitudinal interval is not less than the minimum longitudinal interval corresponding to the current control command are retained. When the source of risk limitation is vehicle outline passage limitation, candidate control quantities whose minimum distance between the outline of the tractor and trailer and the boundary of the lane that cannot be crossed is not less than the minimum distance corresponding to the current control command are retained. When the source of risk limitation is relative motion limitation between the tractor and trailer, candidate control quantities whose minimum distance between the articulation angle and the boundary of the articulation range is not less than the minimum distance corresponding to the current control command are retained. When the source of risk limitation is insufficient control quantity change capability, candidate control quantities that have already established a control quantity correspondence with the next joint safety control domain are retained.

[0125] For candidate control quantities that meet the risk limitation source screening criteria, first retain the candidate control quantity whose control quantity difference with the current control instruction reaches the minimum value, then retain the candidate control quantity that can establish a control quantity correspondence with the next joint security control domain, and finally retain the candidate control quantity whose control transition time reaches the minimum value. The final retained candidate control quantity is determined as the target candidate control quantity.

[0126] When there is no safe candidate control quantity in the real-time safety evaluation result, the candidate state sequence corresponding to each candidate control quantity is read from all candidate control quantities that form the candidate state sequence. The sampling time when the risk limit source is first broken is located along the candidate state sequence. The time interval from the current evaluation time to the located sampling time is counted to form the boundary holding time of the corresponding candidate control quantity.

[0127] Candidate control quantities whose boundary holding time reaches the maximum value are retained. When there are multiple candidate control quantities with the same boundary holding time, the number of restriction types that are broken when each candidate control quantity first breaks through the dynamic safety boundary of the heavy truck is counted, and the candidate control quantity whose restriction type number reaches the minimum value is retained. When there are still multiple candidate control quantities, the number of longitudinal acceleration execution resolutions and the number of steering angular velocity execution resolutions that differ between each candidate control quantity and the current control command are counted respectively. The sum of the corresponding execution resolutions is determined as the control quantity difference number, and the candidate control quantity whose control quantity difference number reaches the minimum value is retained. The finally retained candidate control quantity is determined as the target candidate control quantity.

[0128] The target candidate control quantity formed under the state of no safety candidate control quantity is used to postpone the sampling time when the corresponding limit of the risk limit source is broken, and does not indicate that the corresponding candidate state sequence is already within the dynamic safety boundary of the heavy truck; after the control correction command is executed, the multi-dimensional driving state record and the current joint safety control domain are re-formed in the next control cycle.

[0129] S4.4. Using the current control command and the target candidate control quantity as the start and end positions, arrange the control values ​​step by step between the start and end positions according to the longitudinal acceleration change rate, steering angular velocity change rate and corresponding execution resolution to form a control correction sequence. Select the control value to be executed first in the control correction sequence to generate a control correction command, and write the vehicle state after the control correction command is executed into the multi-dimensional driving state record of the next control cycle.

[0130] Specifically, the longitudinal acceleration and steering angular velocity are read from the current control command, and the longitudinal acceleration control value and steering angular velocity control value are read from the target candidate control quantity. The adjustment direction of each control dimension from the current control command to the target candidate control quantity is determined, and the longitudinal acceleration change rate, steering angular velocity change rate, corresponding execution resolution and control cycle duration are called.

[0131] Based on the corresponding rate of change and control cycle duration, determine the allowable change in control values ​​for each control dimension within a single control cycle. Starting from the current control instruction, within the allowable change range for a single control cycle, select the farthest reachable execution resolution position along the adjustment direction, and determine the corresponding control value as the control value to be executed in an execution sequence. Using the determined control value to be executed as the starting point for the next execution sequence, repeatedly select the farthest reachable execution resolution position until the target candidate control value is reached, forming a control correction sequence.

[0132] The longitudinal acceleration control value and steering angular velocity control value with the same execution order are combined; when one control dimension reaches the corresponding control value in the target candidate control quantity first, the already reached control value is maintained and the control value of another control dimension is arranged; the control correction instruction is generated from the control value to be executed first in the control correction sequence, and the vehicle controller is continuously adjusted to the control value to be executed according to the corresponding rate of change.

[0133] After the control correction command is executed, the vehicle status after execution is read from the spatiotemporal correlation data of the next control cycle, and the vehicle status after execution is written into the multidimensional driving status record of the next control cycle as input for the next candidate control quantity expansion and real-time safety evaluation.

[0134] In summary, this invention improves the real-time performance and continuity of trailer response risk identification, lane-changing process safety evaluation, and control correction by: identifying the start time of the tractor's lateral response and the peak time of the trailer's response according to the sampling time sequence in multi-dimensional driving state records; extracting the risk transmission time difference and the peak relationship of the yaw rate between the tractor and trailer; and selecting reference response processes based on steering direction, speed difference, and steering angle change difference. This allows for the prediction of the current trailer's expected peak response time and expected response amplitude, forming a combined risk transmission record that reflects the lateral risk transmission characteristics of articulated vehicles. Furthermore, it expands candidate control quantities starting from the current control command, deduces the combined vehicle motion state, and combines braking distance limits, vehicle outline passage limits, and relative motion limits between the tractor and trailer to form a joint safety control domain for adjacent lane-changing phases. By comparing the remaining time of the phase with the control transition time, it determines the continuous connection state and the connection interruption phase, thereby improving the real-time performance and continuity of trailer response risk identification, lane-changing process safety evaluation, and control correction.

[0135] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A safety evaluation method for autonomous heavy-duty trucks based on multi-dimensional driving parameters, characterized in that, include: The system acquires multi-dimensional driving parameters, road environment information, and current control commands of autonomous driving heavy trucks, performs spatiotemporal correlation and risk maintenance processing on the multi-dimensional driving parameters and road environment information, and forms a multi-dimensional driving status record. In the multi-dimensional driving status record, locate the start time of the lateral response of the tractor and the peak time of the trailer response, determine the risk transmission time difference and the trailer response amplification state, and determine the lane-changing stage corresponding to the response peak according to the positional relationship between the vehicle outline and the lane boundary, thus forming a combined risk transmission record. Based on the combined risk transmission record, starting from the current control command, the control values ​​are expanded within the longitudinal acceleration execution boundary and the steering angular velocity execution boundary according to the corresponding execution resolution to form candidate control quantities. The combined vehicle motion state corresponding to the candidate control quantities is deduced hourly to form the current joint safety control domain and the next joint safety control domain respectively. The corresponding candidate control quantities are matched according to the control quantity change capability to form a stage safety handover record. Based on the phased security handover record, verify the existence of the current joint security control domain, the location of the current control command, and the continuous handover status, determine the real-time security evaluation result, and select the corresponding candidate control quantity to generate control correction commands.

2. The method for safety evaluation of autonomous heavy-duty trucks based on multi-dimensional driving parameters as described in claim 1, characterized in that, The formation of the multi-dimensional driving status record includes: Vehicle speed, vehicle position, vehicle attitude, steering angle, tractor lateral acceleration, tractor yaw rate and trailer yaw rate are obtained from vehicle bus and inertial measurement unit. Vehicle spacing, relative speed, current lane boundary position and target lane boundary position are obtained from environmental perception unit. The sampling time is associated with the control cycle in which the current control command is located, and the time relationship between parameter changes and the current control command within the same control cycle is maintained. Based on the vehicle speed and steering angle at the start of the control cycle, the longitudinal acceleration and steering angular velocity in the current control command, and the duration of the control cycle, the range of speed change and the range of lateral position change at the end of the control cycle are deduced to form the achievable range of vehicle movement. Continuous verification is performed on the changes in vehicle spacing, relative speed, and vehicle position relative to the current lane boundary and the target lane boundary to form a multi-dimensional driving state record.

3. The method for safety evaluation of autonomous driving heavy trucks based on multi-dimensional driving parameters as described in claim 2, characterized in that, The formation of the combined risk transmission record includes: According to the sampling time sequence of the multi-dimensional driving state record, the sampling time when the lateral acceleration and yaw rate of the tractor are synchronously enhanced is identified and determined as the start time of the lateral response of the tractor. After the start time of the lateral response of the tractor, the sampling time when the yaw rate of the trailer changes from enhancement to decay is identified and determined as the peak time of the trailer response. The time interval between the start time of the lateral response of the tractor and the peak time of the trailer response is calculated. The peak yaw rates of the tractor and the trailer are compared to determine the risk transmission time difference and the trailer response amplification state, respectively. Based on the position of the tractor and trailer outlines relative to the current lane boundary and the target lane boundary at the peak of the trailer response, and combined with the sequential relationship and lateral position changes of the tractor and trailer outlines crossing lane boundaries, the lane-changing phase corresponding to the peak of the trailer response is determined, forming a combined risk transmission record.

4. The method for safety evaluation of autonomous driving heavy trucks based on multi-dimensional driving parameters as described in claim 3, characterized in that, The determination of the risk transmission time difference and trailer response amplification status includes: The risk transmission process that has appeared at the start time of the lateral response of the tractor and the peak time of the response of the trailer is extracted from the multi-dimensional driving status record. The process is then filtered in the order of consistent steering direction, decreasing speed difference at the start time of the lateral response of the tractor, and decreasing steering angle change difference. The remaining risk transmission process after filtering is determined as the reference response process. Align the lateral response start time of the tractor in the reference response process with the lateral response start time of the current tractor. Based on the time interval from the lateral response start time of the tractor to the peak response time of the trailer in the reference response process, determine the expected peak response time of the current trailer in the current sampling time series, and determine the time interval from the lateral response start time of the current tractor to the expected peak response time of the current trailer as the risk transmission time difference. Based on the ratio of the change in the peak yaw rate of the tractor relative to the peak yaw rate of the tractor during the reference response process, the peak yaw rate of the trailer during the reference response process is adjusted to determine the expected response amplitude of the trailer. The expected response amplitude of the trailer is then compared with the peak yaw rate of the tractor to determine the amplification state of the trailer response.

5. The method for safety evaluation of autonomous driving heavy trucks based on multi-dimensional driving parameters as described in claim 4, characterized in that, The security takeover record during the formation phase includes: Read the longitudinal acceleration execution boundary, steering angular velocity execution boundary, longitudinal acceleration rate of change, steering angular velocity rate of change and corresponding execution resolution given by the vehicle controller. Based on the longitudinal acceleration and steering angular velocity in the current control command, select longitudinal acceleration control values ​​and steering angular velocity control values ​​step by step and combine them to form candidate control quantities according to the corresponding execution resolution along the direction of numerical increase and the direction of numerical decrease. Starting with the current vehicle state in the multi-dimensional driving state record, the combined vehicle motion state corresponding to each candidate control quantity is deduced sequentially. The trailer motion state is adjusted according to the combined risk transmission record to form a candidate state sequence. The braking distance, vehicle outline position and relative motion limit between the tractor and trailer within the candidate state sequence form the dynamic safety boundary of the heavy truck. The candidate control quantities that have not exceeded the dynamic safety boundary of the heavy truck are retained to form the current joint safety control domain and the next joint safety control domain. Based on the moment when the combined vehicle outline reaches the position of the lane boundary corresponding to the next lane-changing stage, the remaining time of the stage is determined. According to the longitudinal acceleration change rate and steering angular velocity change rate, the control transition time required to transition from the candidate control quantity in the current joint safety control domain to the candidate control quantity in the next joint safety control domain is obtained. The continuous handover state and the handover interruption stage are determined, and a stage safety handover record is formed.

6. The method for safety evaluation of autonomous driving heavy trucks based on multi-dimensional driving parameters as described in claim 5, characterized in that, The adjustment of trailer movement status according to the combined risk transmission record includes: The trailer response peak moment in the reference response process is time-aligned with the current expected peak moment of the trailer response. Combined with the current expected response amplitude of the trailer and the peak relationship between the yaw rate of the tractor and the trailer in the reference response process, the yaw rate of the trailer is corrected to form a trailer response correction sequence. The trailer response correction sequence is written into the candidate state sequence. The articulation angle between the tractor and the trailer is updated by accumulating the difference between the yaw rate of the tractor and the yaw rate in the trailer response correction sequence. The yaw rate in the trailer response correction sequence is accumulated hourly. The trailer position and trailer attitude in the candidate state sequence are updated with the articulation position of the tractor and the trailer stored in the vehicle controller as the connection reference.

7. The method for safety evaluation of autonomous driving heavy trucks based on multi-dimensional driving parameters as described in claim 6, characterized in that, The formation of the current joint security control domain and the next joint security control domain includes: The time-by-time changes of the combined vehicle speed, longitudinal position and vehicle spacing are tracked along the candidate state sequence. Based on the longitudinal acceleration, the longitudinal motion of the combined vehicle during braking is extrapolated from the boundary. The range of longitudinal positions in which the combined vehicle maintains a distance from adjacent vehicles during braking is determined as the braking distance limit. According to the vehicle controller's stored dimensions of the tractor, trailer, and articulated motion range, the vehicle position and attitude in the candidate state sequence are converted into the tractor's outer profile position, trailer's outer profile position, and relative motion state between the tractor and trailer. Combined with the current lane boundary and the target lane boundary, vehicle outline passage restrictions and relative motion restrictions between the tractor and trailer are formed. The combined range of vehicle motion states, defined by braking distance limits, vehicle outline passage limits, and relative motion limits between the tractor and trailer, is determined as the dynamic safety boundary of the heavy-duty truck. The candidate state sequence is verified hourly, and the candidate control quantities that are continuously within the dynamic safety boundary of the heavy-duty truck are retained. They are then assigned to the current joint safety control domain and the next joint safety control domain according to the lane-changing stage in the combined risk transmission record.

8. The method for safety evaluation of autonomous heavy-duty trucks based on multi-dimensional driving parameters as described in claim 7, characterized in that, The determination of the continuous acceptance status and the acceptance interruption stage includes: The starting sampling time of the candidate state sequence is determined as the current evaluation time. Along the candidate state sequence corresponding to the candidate control quantity in the current joint safety control domain, the sampling time when the combined vehicle outline reaches the lane boundary position relationship corresponding to the next lane change stage is located. The remaining time of the stage is determined. The candidate control quantities in the current joint safety control domain and the next joint safety control domain are matched. The control transition time is calculated based on the changes in the longitudinal acceleration control value and the steering angular velocity control value and the corresponding rate of change. Candidate control quantity pairs whose control transition time does not exceed the remaining time of the corresponding stage are selected to form a control quantity correspondence. The control quantity correspondence is tracked along the sequence of track-changing stages. Track-changing stages where the control quantity correspondence continues to exist are identified as continuous acceptance states, and track-changing stages where the control quantity correspondence stops are identified as acceptance interruption stages. The continuous acceptance states and acceptance interruption stages are written into the stage safety acceptance record.

9. The method for safety evaluation of autonomous driving heavy trucks based on multi-dimensional driving parameters as described in claim 8, characterized in that, The determination of real-time security evaluation results includes: Using longitudinal acceleration and steering angular velocity as control dimensions, the current control command is mapped to the candidate control quantities in the current joint safety control domain. The position of the current control command relative to the boundary of the current joint safety control domain is located along each control dimension, and the candidate control quantity with the smallest difference from the current control command is retained to determine the position of the current control command. The verification is performed in a progressive manner according to the following order: the existence of the current joint security control domain, the current control command being within the current joint security control domain, and the continuous acceptance status. The progressive verification results are associated with the acceptance interruption stage in the stage security acceptance record to determine the range of lane-changing stages where the current control command can maintain safety and the source of risk limitations, thus forming a real-time security evaluation result.

10. The method for safety evaluation of autonomous driving heavy trucks based on multi-dimensional driving parameters as described in claim 9, characterized in that, The step of selecting the corresponding candidate control quantity to generate the control correction instruction includes: Based on the sources of risk limitations in the real-time security assessment results, the candidate control quantities in the current joint security control domain are screened step by step in the order of reducing the difference in control quantities between the current control command and the current joint security control domain, maintaining the correspondence of control quantities, and shortening the control transition time. The candidate control quantities retained after step-by-step screening are determined as the target candidate control quantities. Using the current control command and the target candidate control quantity as the start and end positions, the control values ​​are arranged step by step between the start and end positions according to the longitudinal acceleration change rate, the steering angular velocity change rate and the corresponding execution resolution to form a control correction sequence. The control value to be executed first in the control correction sequence is selected to generate a control correction command, and the vehicle state after the control correction command is executed is written into the multi-dimensional driving state record of the next control cycle.

Images