A vehicle control method

By dynamically determining the near and far pre-aiming distances and combining the near-end heading curvature and the far-end position curvature to generate a feedforward control sequence, the problem of untimely response and insufficient foresight of the vehicle when the reference trajectory changes is solved, thus improving the trajectory following capability of intelligent driving vehicles.

CN123104690APending Publication Date: 2026-11-17上海易博图科技有限公司
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611276077.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-20
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In existing technologies, when a vehicle's geometric shape changes along a reference trajectory, the steering control quantity determined based on trajectory information at a single scale cannot simultaneously reflect the trajectory changes near the vehicle and the path direction further away. This results in the intelligent driving vehicle not responding to trajectory changes in a timely manner or failing to anticipate changes in the path ahead.

Method used

By acquiring the vehicle's current longitudinal speed and reference trajectory, the near and far aiming distances are dynamically determined, and the near and far aiming positions are established respectively. The feedforward control sequence is generated using the near-end heading curvature and the far-end position curvature. The feedforward control sequence is generated by combining the near-end weights and the far-end weights and performing weighted fusion to control the vehicle's driving trajectory.

Benefits of technology

This technology enables vehicles to respond promptly to local trajectory changes and anticipate distant paths under different speed conditions, reducing overshoot and the probability of continuous correction at curve entrances and improving the trajectory following performance of intelligent driving vehicles.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123104690A_ABST
    Figure CN123104690A_ABST
Patent Text Reader

Abstract

This disclosure provides a vehicle control method applicable to fields such as unmanned vehicles, autonomous driving, and autonomous driving. The method includes: acquiring the vehicle's current longitudinal speed, reference trajectory, and multiple predicted nodes, where each predicted node represents the vehicle's expected future spatial pose; determining near-pre-aiming distances and far-pre-aiming distances matching the current longitudinal speed; for each predicted node, determining its corresponding near-pre-aiming position and far-pre-aiming position along the reference trajectory, starting from that predicted node and according to the near-pre-aiming distance and far-pre-aiming distance; determining the near-end heading curvature based on the relative heading change relationship between the near-pre-aiming position and the predicted node; and determining the far-end position curvature based on the lateral relative position relationship between the far-pre-aiming position and the predicted node; generating a feedforward control sequence corresponding to each predicted node based on the near-end heading curvature and the far-end position curvature; and controlling the vehicle's trajectory during driving based on the feedforward control sequence.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This disclosure relates to the field of intelligent driving, and more specifically, to a vehicle control method. Background Technology

[0002] With the development of intelligent driving technology, vehicles typically perform lateral control based on a planned reference trajectory to follow the target driving path. During driving, the vehicle can determine the corresponding steering control amount based on the geometric information of the reference trajectory and control the vehicle's steering actuator accordingly.

[0003] However, when the geometry of the reference trajectory changes, the steering control quantity determined based on trajectory information of a single scale is difficult to simultaneously reflect the trajectory changes near the vehicle and the path direction further away. This may result in the vehicle's response to trajectory changes not being timely enough or insufficient foresight of changes in the path ahead, thus affecting the trajectory following effect of the intelligent driving vehicle. Summary of the Invention

[0004] In view of this, the present disclosure provides a vehicle control method.

[0005] One aspect of this disclosure provides a vehicle control method, comprising: acquiring the vehicle's current longitudinal speed, a reference trajectory, and multiple prediction nodes, wherein the prediction nodes represent the vehicle's expected future spatial pose; determining, based on the current longitudinal speed, near-pre-aiming distances and far-pre-aiming distances matching the vehicle speed; for each prediction node, determining, along the reference trajectory, a near-pre-aiming position and a far-pre-aiming position corresponding to the prediction node, based on the near-pre-aiming distance and the far-pre-aiming distance; determining a near-end heading curvature based on the relative heading change relationship between the near-pre-aiming position and the prediction node; and determining a far-end position curvature based on the lateral relative position relationship between the far-pre-aiming position and the prediction node; generating a feedforward control sequence corresponding to each prediction node based on the near-end heading curvature and the far-end position curvature; and controlling the vehicle's driving trajectory during driving based on the feedforward control sequence.

[0006] According to embodiments of this disclosure, based on the current longitudinal speed, a near-aiming distance and a far-aiming distance matching the vehicle speed are determined, including: determining the near-aiming distance based on the product of the current longitudinal speed and the near-aiming time scale and the minimum near-aiming distance, wherein the near-aiming distance is the larger of the two; determining the far-aiming distance based on the product of the current longitudinal speed and the far-aiming time scale and the minimum far-aiming distance, wherein the far-aiming distance is the larger of the two; wherein the far-aiming time scale is not less than the near-aiming time scale, and the minimum far-aiming distance is not less than the minimum near-aiming distance.

[0007] According to embodiments of this disclosure, determining the near-end heading curvature based on the relative heading change relationship between the near-preview position and the predicted node includes: determining the reference heading corresponding to the near-preview position based on the reference trajectory; determining the predicted heading corresponding to the predicted node based on the expected spatial pose represented by the predicted node; determining the heading difference between the reference heading and the predicted heading; periodically normalizing the heading difference according to the period of the heading angle to obtain the normalized heading difference; and determining the near-end heading curvature based on the normalized heading difference and the near-preview distance.

[0008] According to embodiments of this disclosure, determining the far-end position curvature based on the lateral relative positional relationship between the far-end aiming position and the predicted node includes: determining the predicted position and predicted heading corresponding to the predicted node based on the expected spatial pose represented by the predicted node; determining the planar displacement of the far-end aiming position relative to the predicted position and the Euclidean distance between them; transforming the planar displacement to a local coordinate system based on the predicted heading, and determining the lateral component of the planar displacement in the local coordinate system; and determining the far-end position curvature based on the ratio between twice the lateral component and the square of the Euclidean distance.

[0009] According to embodiments of this disclosure, a feedforward control sequence corresponding to each prediction node is generated based on the near-end heading curvature and the far-end position curvature, including: obtaining the vehicle speed-related vehicle yaw gain, which characterizes the steady-state mapping relationship between the steering angle on the vehicle steering actuator side and the vehicle yaw rate; converting the near-end heading curvature and the far-end position curvature into near-end feedforward angle and far-end feedforward angle respectively using the same vehicle speed-related vehicle yaw gain; weighting and fusing the near-end feedforward angle and the far-end feedforward angle according to the near-end weight and the far-end weight to obtain the fused feedforward angle corresponding to each prediction node, wherein the sum of the near-end weight and the far-end weight is one; and generating a feedforward control sequence based on the fused feedforward angle corresponding to each prediction node.

[0010] According to embodiments of this disclosure, obtaining vehicle speed-related vehicle yaw gain includes: obtaining vehicle wheelbase, steering system transmission ratio, and understeer characteristic coefficient, wherein the understeer characteristic coefficient characterizes the understeer characteristic of the vehicle; and determining vehicle speed-related vehicle yaw gain based on the current longitudinal speed, vehicle wheelbase, steering system transmission ratio, and understeer characteristic coefficient.

[0011] According to embodiments of this disclosure, for each predicted node, along the reference trajectory, starting from the predicted node, the near-pre-aiming position and far-pre-aiming position corresponding to the predicted node are determined according to the near-pre-aiming distance and the far-pre-aiming distance, respectively. This includes: obtaining the cumulative arc length sequence of the reference trajectory and the prediction time corresponding to each predicted node; determining the reference arc length position corresponding to each predicted node based on the current longitudinal velocity and the prediction time corresponding to each predicted node, wherein the reference arc length position represents the starting position of the pre-aiming query of the corresponding predicted node on the cumulative arc length axis of the reference trajectory; adding the near-pre-aiming distance and the far-pre-aiming distance to each reference arc length position to determine the corresponding near-pre-aiming arc length position and far-pre-aiming arc length position; and determining the near-pre-aiming position and far-pre-aiming position corresponding to the near-pre-aiming arc length position and far-pre-aiming arc length position respectively according to the reference trajectory.

[0012] According to embodiments of this disclosure, before determining the near-pre-aiming position and far-pre-aiming position of each predicted node along the reference trajectory, starting from the predicted node and according to the near-pre-aiming distance and far-pre-aiming distance, the method further includes: determining the endpoint of the effective domain of the planned path corresponding to the reference trajectory; if the reference arc length position corresponding to the predicted node exceeds the endpoint of the effective domain of the planned path, determining the curvature of the last reference node in the domain before the first reference arc length position exceeding the endpoint of the effective domain of the planned path as the tail-holding curvature; determining the curvature, heading, and position of the reference nodes exceeding the endpoint of the effective domain of the planned path node by node according to the tail-holding curvature, forming an extended reference sequence covering multiple predicted nodes.

[0013] According to embodiments of this disclosure, after determining the curvature, heading, and position of reference nodes exceeding the endpoint of the effective domain of the planned path node by node based on the tail-maintaining curvature to form an extended reference sequence covering multiple predicted nodes, the method further includes: when the far-pre-aiming arc length position exceeds the end of the extended reference sequence, determining the out-of-bounds arc length of the far-pre-aiming arc length position relative to the end of the extended reference sequence; determining the far-pre-aiming position based on the out-of-bounds arc length along the tangent corresponding to the end heading of the extended reference sequence; and when the near-pre-aiming arc length position exceeds the end of the extended reference sequence, determining the end heading of the extended reference sequence as the reference heading corresponding to the near-pre-aiming position.

[0014] According to embodiments of this disclosure, the method further includes: obtaining the remaining path length ahead of the vehicle; when the remaining path length is shorter than the far-field pre-aiming distance but longer than the near-field pre-aiming distance, linearly attenuating the feedforward control sequence based on the relative position of the remaining path length between the near-field pre-aiming distance and the far-field pre-aiming distance; and attenuating the feedforward control sequence to zero when the remaining path length is not longer than the near-field pre-aiming distance.

[0015] Another aspect of this disclosure provides a vehicle control device, comprising: a first acquisition module for acquiring the vehicle's current longitudinal speed, a reference trajectory, and multiple prediction nodes, wherein the prediction nodes represent the vehicle's expected future spatial pose; a first determination module for determining, based on the current longitudinal speed, a near-pre-aiming distance and a far-pre-aiming distance matching the vehicle speed; a second determination module for determining, for each prediction node, a near-pre-aiming position and a far-pre-aiming position corresponding to the prediction node along the reference trajectory, starting from the prediction node and according to the near-pre-aiming distance and the far-pre-aiming distance; a third determination module for determining the near-end heading curvature based on the relative heading change relationship between the near-pre-aiming position and the prediction node; a fourth determination module for determining the far-end position curvature based on the lateral relative position relationship between the far-pre-aiming position and the prediction node; a first generation module for generating a feedforward control sequence corresponding to each prediction node based on the near-end heading curvature and the far-end position curvature; and a first control module for controlling the vehicle's driving trajectory during driving based on the feedforward control sequence.

[0016] Another aspect of this disclosure provides an electronic device including: at least one processor; and a memory connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the vehicle control method of any of the foregoing embodiments.

[0017] Another aspect of this disclosure provides a computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to perform a vehicle control method according to any of the foregoing embodiments.

[0018] Another aspect of this disclosure provides a computer program product, including a computer program / instructions, characterized in that the computer program / instructions, when executed by a processor, implement the operation of the vehicle control method of any of the foregoing embodiments. Attached Figure Description

[0019] The above and other objects, features and advantages of this disclosure will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0020] Figure 1 A flowchart illustrating a vehicle control method according to an embodiment of the present disclosure is shown schematically.

[0021] Figure 2 A flowchart illustrating the determination of the curvature of a remote position in a vehicle control method according to an embodiment of the present disclosure is shown schematically.

[0022] Figure 3 A flowchart illustrating the generation of a feedforward control sequence in a vehicle control method according to an embodiment of the present disclosure is shown schematically.

[0023] Figure 4 A flowchart illustrating the determination of near and far aiming positions in a vehicle control method according to an embodiment of the present disclosure is shown.

[0024] Figure 5 A block diagram of a vehicle control device according to an embodiment of the present disclosure is schematically shown; and

[0025] Figure 6 A block diagram of an electronic device suitable for implementing the methods described above, according to embodiments of the present disclosure, is illustrated schematically. Detailed Implementation

[0026] The embodiments of the present disclosure will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the disclosure. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of the present disclosure for ease of explanation. However, it will be apparent that one or more embodiments may be practiced without these specific details. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concepts of the present disclosure.

[0027] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit this disclosure. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0028] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0029] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0030] In the embodiments disclosed herein, the collection, updating, analysis, processing, use, transmission, provision, disclosure, and storage of data (e.g., including but not limited to user personal information) comply with relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. In particular, necessary measures have been taken to prevent unauthorized access to user personal information data and to safeguard user personal information security, network security, and national security.

[0031] Embodiments of this disclosure provide a vehicle control method, including: acquiring the vehicle's current longitudinal speed, reference trajectory, and multiple prediction nodes, where each prediction node represents the vehicle's expected future spatial pose; determining near-pre-aiming distances and far-pre-aiming distances matching the vehicle speed based on the current longitudinal speed; for each prediction node, determining the near-pre-aiming position and far-pre-aiming position corresponding to the prediction node along the reference trajectory, starting from the prediction node and based on the near-pre-aiming distance and far-pre-aiming distance; determining the near-end heading curvature based on the relative heading change relationship between the near-pre-aiming position and the prediction node; and determining the far-end position curvature based on the lateral relative position relationship between the far-pre-aiming position and the prediction node; generating a feedforward control sequence corresponding to each prediction node based on the near-end heading curvature and the far-end position curvature; and controlling the vehicle's driving trajectory during driving based on the feedforward control sequence.

[0032] Figure 1 A flowchart illustrating a vehicle control method according to an embodiment of the present disclosure is shown schematically.

[0033] like Figure 1 As shown, the vehicle control method may include at least operations S110 to S170.

[0034] During operation S110, the vehicle's current longitudinal velocity, reference trajectory, and multiple prediction nodes are obtained. The prediction nodes represent the vehicle's expected spatial pose in the future.

[0035] In intelligent driving or autonomous driving scenarios, vehicles typically rely on perception and planning modules to generate future driving routes. The current longitudinal speed reflects the vehicle's current speed, and the reference trajectory usually contains a series of path points with geometric information. Predictive nodes can be discrete state points sampled by the predictive control algorithm within a future time window, used to describe the spatial position and attitude the vehicle expects to reach. Real-time vehicle speed can be read through onboard sensors (such as wheel speedometers or inertial measurement units), and the planned reference trajectory and predictive node sequence can be received from the upstream trajectory planning module. Taking an autonomous vehicle driving on urban roads as an example, the underlying control module continuously retrieves the current driving speed at fixed intervals, while simultaneously acquiring the expected driving trajectory for the next few seconds and a series of discrete predictive nodes issued by the navigation planning layer.

[0036] In S120 operation, based on the current longitudinal speed, the near and far preview distances are determined to match the vehicle speed. The preview distance determines the range the vehicle can "look" forward during control. The preview range is dynamically adjusted with vehicle speed, rather than using a fixed value. The near preview distance focuses on capturing upcoming local curves, while the far preview distance is used to grasp the overall path direction further away. Determining these two distances based on vehicle speed ensures the vehicle has appropriate reaction time at different speeds. Specifically, this can be achieved by establishing a mapping relationship between vehicle speed and preview distance, using a lookup table or a function model that varies with vehicle speed to calculate the spatial preview lengths at the near and far ends. For example, when an autonomous vehicle is speeding on a highway, the control logic calculates longer near and far preview distances based on the higher speed, allowing for a wider forward field of view; while when the vehicle is slowly crawling in congested traffic, these distances are shortened accordingly, ensuring the control focus is concentrated on the immediate local road conditions.

[0037] In operation S130, for each predicted node, the near and far pre-aiming positions of the predicted node are determined along the reference trajectory, starting from that predicted node, according to the near and far pre-aiming distances. For each predicted node in the sequence, it is used as a local observation starting point, and the specified near and far pre-aiming distances are extended forward along the reference trajectory to find the corresponding two future target points. By setting dual observation points, a dual-scale spatial observation benchmark is established for each predicted node. In specific implementation, the near and far pre-aiming distances can be superimposed on the position of each predicted node based on the geometric length of the reference trajectory, and then the corresponding near and far pre-aiming positions can be located on the trajectory through spatial geometric mapping. Assuming that there are ten predicted nodes in the prediction time domain, the control algorithm will traverse these nodes one by one, searching for the corresponding near and far endpoints along the trajectory for the first node, and then repeating this search process for the second node until all nodes have their own exclusive near and far pre-aiming positions.

[0038] In operation S140, the near-end heading curvature is determined based on the relative heading change between the near-preview position and the prediction node. The near-end heading curvature reflects the tangent change trend over a short distance. The relative heading change reflects the degree of deflection required in the vehicle's heading as it travels from the prediction node to the near-preview position. Utilizing the difference in heading to assess curvature reflects the geometric changes at the curve's entrance or in a localized area. Determining this curvature involves comparing the tangent direction at the near-preview position with the tangent direction at the prediction node, extracting the angle difference, and then combining this with the spatial span to convert it into a physical quantity characterizing the degree of local curvature. In a specific computational scenario, if the heading at the prediction node is due north, while the heading at the near-preview position is northeast, the algorithm extracts the angle change between these two directions and uses this shorter preview span to assess the upcoming local curve curvature.

[0039] In operation S150, the far-end position curvature is determined based on the lateral relative positional relationship between the far-end preview position and the prediction node. The far-end position curvature focuses on target point tracking over longer distances. The lateral relative positional relationship describes the degree of lateral offset of the far-end preview position from the current driving direction of the prediction node. Estimating curvature based on position offset can form a steering trend that closes towards the target point, providing forward-looking predictability. In practice, the lateral deviation of the far-end preview position in the coordinate space based on the prediction node can be analyzed, and the straight-line or curved span between the two points can be comprehensively considered to derive a geometric curvature guiding the vehicle towards the far target. For example, the algorithm projects the far-end preview position to the side of the prediction node, assesses the distance this far target point deviates to the left or right relative to the current vehicle heading, and then uses this lateral offset distance to calculate a curvature value guiding the vehicle closer.

[0040] In operation S160, based on the near-end heading curvature and the far-end position curvature, a feedforward control sequence corresponding to each prediction node is generated. The feedforward control sequence is a set of instructions guiding the actuators to act in advance. The near-end heading curvature provides local correction requirements, while the far-end position curvature provides global orientation guidance. Combining these two allows for both near-end response and far-end prediction. The generation of the sequence can be achieved by employing a fusion mechanism to integrate these two heterogeneous curvature information into a unified feedforward control quantity, and repeating this integration process for all nodes in the prediction time domain, thus forming a complete set of control command flows distributed over time or space. Taking the generation of the feedforward quantity for a single node as an example, specific algorithmic logic can be used to comprehensively consider the local curvature requirements extracted from the near end and the target approach requirements extracted from the far end, deriving a fused feedforward steering index. The indices of all nodes are then linked together to form a complete feedforward control sequence.

[0041] In operation S170, the vehicle's trajectory is controlled based on a feedforward control sequence. This feedforward control sequence, acting as an active steering intervention signal, establishes a steering state that matches the road curvature before lateral errors occur. Inputting the generated feedforward sequence into the underlying actuators or as a reference target for the predictive controller drives the steering mechanism to move according to the expected trajectory curvature. For example, when an autonomous vehicle enters a series of curves, the underlying steering controller applies a steering action based on the pre-generated feedforward sequence before the vehicle deviates from the reference trajectory, ensuring the vehicle conforms to the planned curve trajectory and reducing the lag in control response relative to changes in road curvature.

[0042] According to embodiments of this disclosure, by dynamically determining the near and far pre-aiming distances based on the current longitudinal speed, the problem of inconsistent reaction time of fixed spatial pre-aiming distances caused by changes in vehicle speed is solved, enabling the spatial forward-looking range to adapt to changes in vehicle speed. Simultaneously, dual-scale pre-aiming at both near and far ends is established for each prediction node. At the near end, relative heading changes are used to perceive local tangent changes to preserve response timeliness, while at the far end, lateral relative position is used to characterize the target's trajectory to provide forward-looking prediction. Combining two heterogeneous curvatures avoids the limitations of single pre-aiming geometry in terms of near-end noise amplification or far-end information loss. Furthermore, a feedforward control sequence is generated for each prediction node, enabling the control logic to know the steering requirements of the future trajectory and establish a matching steering state before lateral errors form. This reduces phase lag caused by correcting errors only after they have formed, lowering the probability of overshoot and continuous corrections at curve entrances or sudden curvature changes.

[0043] Based on the foregoing embodiments, operation S120 may include operations S210 to S220.

[0044] In operation S210, the near-lookout distance is determined based on the product of the current longitudinal speed and the near-lookout timescale, as well as the near-lookout minimum distance. The near-lookout distance is the larger of the two values. The near-lookout timescale reflects the expected buffer time for responding to local steering maneuvers, while the near-lookout minimum distance is the lower limit of the spatial observation baseline. By multiplying the current vehicle speed by the near-lookout timescale, a distance value that dynamically changes with vehicle speed is obtained. This dynamic distance value is compared with the pre-calibrated near-lookout minimum distance, and the larger of the two is taken as the final near-lookout distance.

[0045] Specifically, near-aiming distance It can be represented as:

[0046]

[0047] Where v is the current longitudinal velocity, T n For near-pre-aiming timescale, D n,min This is the minimum distance for near-targeting.

[0048] When an autonomous vehicle travels at a high speed on an urban expressway, the product of vehicle speed and time scale will exceed the minimum distance threshold. At this time, the near-look-ahead distance increases proportionally with vehicle speed, giving the vehicle an approximately constant reaction time. When the vehicle is slowly following another vehicle or even close to stopping in a congested area, the product term becomes smaller, and the near-look-ahead minimum distance will take over the calculation results, maintaining a non-zero geometric baseline to prevent the amplification of discrete errors caused by trajectory interpolation due to excessively short observation distance.

[0049] In operation S220, the far-seeking distance is determined based on the product of the current longitudinal velocity and the far-seeking timescale, as well as the minimum far-seeking distance. The far-seeking distance is the larger of the two. The far-seeking timescale defines the time span for observing the global path, and the minimum far-seeking distance sets the spatial lower limit for the far-end observation point. Corresponding to the near-end calculation logic, the current longitudinal velocity is multiplied by the far-seeking timescale to obtain the far-end dynamic distance, and then the maximum value between this dynamic distance and the minimum far-seeking distance is selected for output.

[0050] Specifically, long-range pre-aiming It can be represented as:

[0051]

[0052] Where v represents the current longitudinal velocity, T f D represents the time scale for long-range aiming. f,min This represents the minimum distance for long-range pre-aiming.

[0053] Taking a vehicle about to enter a high-curvature curve as an example, during the high-speed entry phase, the far-preview distance calculated based on the higher vehicle speed will push the observation point further into the curve, allowing the control logic to perceive the path direction in advance; while during the low-speed passage phase, the minimum far-preview distance ensures that the observation point will not shrink to the vicinity of the vehicle's current position, maintaining the effectiveness of the far-end observation.

[0054] According to embodiments of this disclosure, the far-field preview timescale is not less than the near-field preview timescale, and the minimum far-field preview distance is not less than the minimum near-field preview distance.

[0055] Numerical constraints on the time scale and minimum distance establish a spatial hierarchical relationship between the near-end and far-end pre-aiming mechanisms. The requirement that the far-end parameter be no less than the near-end parameter ensures that, under any vehicle speed condition, the far-end pre-aiming position is always in front of or coincides with the near-end pre-aiming position, forming a spatially distinct dual-scale observation structure. In a specific parameter configuration example, the near-end pre-aiming time scale can be set to 0.4 seconds and the minimum near-end pre-aiming distance to 2 meters; simultaneously, the far-end pre-aiming time scale can be set to 1.0 second and the minimum far-end pre-aiming distance to 5 meters. This ensures that the near-end pre-aiming distance always focuses on local road condition changes in front of the vehicle, while the far-end pre-aiming distance spans a longer spatial distance to capture the distant target position. The two mechanisms do not interfere with each other physically and complement each other.

[0056] According to embodiments of this disclosure, the near and far anti-aiming distances are determined by the larger of the product of vehicle speed and anti-aiming time scale and the minimum anti-aiming distance. This ensures that under medium-to-high speed conditions, the anti-aiming distance increases accordingly with the increase of vehicle longitudinal speed, maintaining an approximately constant anti-aiming reaction time. Under low-speed conditions, the minimum anti-aiming distance provides a baseline, maintaining a non-zero geometric baseline and preventing excessive sensitivity of curvature estimation to trajectory discrepancies caused by excessively short anti-aiming distances. Simultaneously, by limiting the far anti-aiming time scale and the corresponding parameters where the minimum distance is not less than that of the near end, a spatially differentiated dual-scale anti-aiming is stably formed over a wide speed range, enabling the near and far ends to respectively assume the roles of local tangential response and far-end spatial forward look.

[0057] Based on the foregoing embodiments, operation S140 may include operations S310 to S350.

[0058] In operation S310, the reference heading corresponding to the near-target position is determined based on the reference trajectory. The reference heading represents the tangential direction of the vehicle as it travels along the reference trajectory at the near-target position. The reference heading can be the tangential direction of the vehicle as it travels along the reference trajectory at the near-target position. The heading angle data for that position can be extracted by interpolating along the arc length on the reference trajectory. Assuming the reference trajectory consists of a series of discrete path points, when the near-target position falls between two known path points, the control unit calculates the corresponding reference heading using linear interpolation or spline interpolation algorithms.

[0059] In operation S320, the predicted heading corresponding to the predicted node is determined based on the expected spatial pose represented by the predicted node. The predicted heading is the expected frontal orientation of the vehicle at the current predicted node (or reference position). Similar to the processing of near-targeting positions, this heading can also be obtained by interpolation based on the reference trajectory using arc length. Taking the i-th predicted node in the prediction time domain as an example, the corresponding predicted heading θi (the unit can be radians rad) is extracted by parsing the geometric information of this node on the reference trajectory.

[0060] In operation S330, the heading difference between the reference heading and the predicted heading is determined. The heading difference reflects the deflection of the tangent direction of the reference trajectory over the spatial distance from the predicted node to the near-target position. The original heading difference is obtained by subtracting the reference heading at the near-target position from the predicted heading at the predicted node. Let the reference heading at the near-target position be θ. n,i The predicted heading of the predicted node is θ i Then calculate the difference θ between the two. n,i -θ i This quantifies the angle the vehicle needs to turn within this short distance.

[0061] In operation S340, the heading difference is periodically normalized according to the period of the heading angle to obtain the normalized heading difference. Because angles are periodic, the original heading difference may exhibit pseudo-jumps in value when crossing angle representation boundaries (such as -π and π). Periodic normalization maps the heading difference to a continuous principal value interval, eliminating non-physical numerical abrupt changes. In a specific implementation, normalization can be achieved using an arctangent function combined with sine and cosine functions, as expressed by the formula:

[0062]

[0063] Where, Δθ n,i To normalize the heading difference, θ n,i For reference heading, θ i To predict the course.

[0064] When a vehicle turns left by a small angle, but the value jumps from 3.14 to -3.14, the formula is able to calculate the actual small angle change, rather than producing a jump of about 2π.

[0065] When operating the S350, the near-end heading curvature is determined based on the normalized heading difference and the near-sight distance. The near-end heading curvature reflects the average degree of curvature within the near-sight interval. According to the geometric properties of plane curves, the heading change equals the integral of the curvature along the arc length. Dividing the normalized heading difference by the near-sight distance extracts the local curvature trend. The calculation formula can be expressed as:

[0066]

[0067] Where k h,i Near-end heading curvature (in meters) -1 ), Δθ n,i To normalize the heading difference, D n This is the close aiming distance.

[0068] Because the denominator is the first-order distance rather than the squared distance, this calculation method is insensitive to position interpolation disturbances. When a vehicle approaches the entrance to a curve, the curvature calculated using this formula is directly used as a quantitative indicator of the local tangent change.

[0069] According to embodiments of this disclosure, the curvature of the near-end trajectory is characterized by the change in heading between the near-preview position and the predicted node. Since the heading change is the integral of the curvature over the arc length, dividing it by the near-preview distance over a short distance directly extracts the local curvature trend. Compared to calculating curvature using position coordinates, the denominator is the first-order distance rather than the squared distance, reducing sensitivity to position interpolation disturbances and avoiding near-end noise amplification. Periodic normalization of the heading difference before curvature calculation eliminates spurious jumps when the heading angle crosses periodic boundaries, maintaining the continuity of near-end curvature calculation.

[0070] Figure 2 A flowchart illustrating the determination of the curvature of a remote position in a vehicle control method according to an embodiment of the present disclosure is shown schematically.

[0071] like Figure 2 As shown, based on the aforementioned embodiments, operation S150 may include operations S410 to S440.

[0072] In operation S410, the predicted position and predicted heading of the predicted node are determined based on the expected spatial pose represented by the predicted node. The predicted position and predicted heading can be the two-dimensional spatial coordinates of the vehicle at the predicted node and the vehicle's heading. The coordinate values ​​and angle values ​​corresponding to the node are extracted by interpolation calculations based on the arc length on the reference trajectory. Assuming that the predicted node is located on a curve segment of the known planned path, the x and y coordinates (x, y) of the node in the global coordinate system are calculated using a polynomial interpolation algorithm. i ,y i and the corresponding heading angle θ i .

[0073] In operation S420, the planar displacement of the far-preview position relative to the predicted position and the Euclidean distance between them are determined.

[0074] Planar displacement describes the coordinate difference between the far-seeking position and the predicted position in the global coordinate system, while Euclidean distance represents the straight-line length between these two points. The planar displacement is obtained by subtracting the coordinates of the predicted position from the coordinates of the far-seeking position, and the straight-line distance between the two points can be calculated using the Pythagorean theorem. The corresponding calculation formula can be expressed as:

[0075]

[0076] Among them, (x f,i ,y f,i (x) represents the coordinates of the far-field aiming position. i ,yi ) represents the coordinates of the predicted location, Δx i and Δy i L represents the planar displacement components in the global coordinate system. i The distance is Euclidean.

[0077] Taking the vehicle tracking a target point far ahead as an example, the coordinate differences between the target point and the current prediction node in the east-west and north-south directions are calculated respectively, and then the squared value of the straight-line distance between the two points is obtained.

[0078] In operation S430, the planar displacement is transformed to a local coordinate system based on the predicted heading, and the lateral component of the planar displacement in the local coordinate system is determined. The local coordinate system can be a relative coordinate system established with the predicted position as the origin and the predicted heading as the vertical axis. The lateral component reflects the lateral distance of the far-seeking position from the current vehicle heading. Using a rotation transformation matrix, the global planar displacement is projected onto the local coordinate system, and the coordinate component perpendicular to the predicted heading is extracted. The calculation formula for this process is:

[0079]

[0080] In the formula y l,i θ is the horizontal component in the local coordinate system. i To predict the course.

[0081] When the far-viewing position is located to the left front of the vehicle's expected driving direction, the lateral component obtained after coordinate rotation calculation is the vertical distance of the target point from the left of the vehicle's longitudinal axis.

[0082] In operation S440, the far-end position curvature is determined based on the ratio between twice the lateral component and the square of the Euclidean distance. The far-end position curvature corresponds to the geometry of a pure tracking arc from the predicted pose, targeting the far pre-aiming point. Multiplying the local lateral component by two and then dividing by the square of the Euclidean distance yields the geometric curvature guiding the vehicle toward the far target point.

[0083] The calculation formula can be written as:

[0084]

[0085] Where k p,i The curvature at the distal position (in meters) -1 ), y l,i For the lateral component, L i The distance is Euclidean.

[0086] When a vehicle needs to change lanes to the right or follow a road that curves to the right, by substituting the lateral offset to the right and the longer aiming distance, the curvature value calculated by this formula can characterize a smooth arc trajectory that fits the distant target.

[0087] According to embodiments of this disclosure, the positional relationship between the far-seeking position and the predicted node is transformed to the local heading coordinate system of the predicted node, and the curvature of the far-end position is determined using the lateral component and the distance between the two points. This preserves the lateral offset information of the target point relative to the current tangent, supplementing the path space orientation that cannot be expressed by the simple heading difference. Simultaneously, the far-seeking distance provides a larger geometric baseline, reducing the amplified effect of the position interpolation error caused by the distance squared denominator.

[0088] Figure 3 A flowchart illustrating the generation of a feedforward control sequence in a vehicle control method according to an embodiment of the present disclosure is shown.

[0089] like Figure 3 As shown, based on the aforementioned embodiments, operation S160 may include operations S510 to S540.

[0090] In operation S510, the vehicle speed-related yaw gain is obtained. This gain characterizes the steady-state mapping between the steering angle on the steering actuator side and the yaw rate. The vehicle speed-related yaw gain reflects the actual yaw rate generated by the vehicle when the steering wheel or underlying steering motor rotates a certain angle at a specific vehicle speed. Let the steering actuator side angle be δ (in radians, rad), and the yaw rate be ω (in rad / s). The steady-state gain from the steering angle on the actuator side to the yaw rate at the current speed is G(v) (the unit can be s). -1 If ω = G(v)δ, then ω = G(v)δ. This parameter can be obtained by consulting a vehicle dynamics calibration table or by calling a preset gain curve. For example, when an autonomous vehicle is traveling at a specific speed, the control system reads the yaw gain value corresponding to that speed, which is used to convert the geometric curvature into actual mechanical steering commands.

[0091] When operating the S520, the near-end heading curvature and far-end position curvature are converted into near-end feedforward angle and far-end feedforward angle respectively by using the vehicle yaw gain related to the same vehicle speed.

[0092] Since the yaw rate corresponding to the desired curvature is equal to the product of the vehicle speed and the curvature, and combining this with the aforementioned gain relationship, the geometric curvature can be uniformly mapped to the angle on the actuator side. The conversion between the proximal and distal curvatures can be expressed by the following formulas:

[0093]

[0094] Where, δ h,i For the near-end feedforward angle, δ p,i Where is the far-end feedforward angle, v is the effective longitudinal velocity, and k is the effective longitudinal velocity. h,i For the near-end heading curvature, k p,iLet G(v) be the curvature at the far end position, and G(v) be the yaw gain of the vehicle at the same speed. In a specific operation, the calculated local tangent curvature and the far-end approach curvature are substituted into the above formula to obtain two candidate values ​​for the steering wheel angle with the same physical dimensions and sign convention.

[0095] In operating S530, the near-end feedforward angle and the far-end feedforward angle are weighted and fused according to the near-end weight and the far-end weight to obtain the fused feedforward angle corresponding to each prediction node, wherein the sum of the near-end weight and the far-end weight is one.

[0096] Weighted fusion with complementary weights, i.e., convex combination, can balance the two signals without changing the overall output scale. Let the weight of the near-end heading component (i.e., near-end weight) be wh, satisfying 0 ≤ wh ≤ 1, and the weight of the far-end position component (i.e., far-end weight) be wp = 1 - wh. The formula for calculating the fusion feedforward angle is:

[0097]

[0098] Where, δ ff,i Let be the fusion feedforward angle corresponding to the i-th prediction node.

[0099] When w h As w increases, the output places greater emphasis on the timeliness of the proximal tangent change; when w h When the weight is reduced, the output places greater emphasis on the forward look-ahead of the distant target position. For example, in a low-speed kinematics predictive control branch, the proximal weight is fixed at 0.4 and the distal weight at 0.6. The system adds the two feedforward angles according to this ratio to obtain a comprehensive steering angle.

[0100] In operation S540, a feedforward control sequence is generated based on the fused feedforward angles corresponding to each prediction node. The fused feedforward angles calculated by each node in the prediction time domain are arranged in temporal or spatial order to form a complete feedforward control sequence. This sequence contains the steering execution commands for the vehicle at each discrete point in the future. The above conversion and fusion operations are performed on all prediction nodes, and the resulting δ is... ff,i The nodes are concatenated into an array or queue based on their index i, and then sent to the underlying steering controller for execution.

[0101] According to the embodiments of this disclosure, instead of directly weighting the two curvature paths at the curvature level, the two heterogeneous curvature paths are first converted to the actuator-side feedforward angle using the same vehicle yaw gain. This ensures that the two geometric curvature paths, after being mapped through the same vehicle response relationship, have the same physical dimensions, actuator mapping relationship, and sign reference. Then, complementary weights that sum to one are used for fusion. Without altering the overall feedforward scale reference due to the fusion process, this balances the timeliness of near-end heading changes and the forward visibility of far-end target positions, avoiding mutual antagonism caused by inconsistencies in the dimensions or signs of the two paths.

[0102] Based on the foregoing embodiments, operation S510 may include operations S610 to S620.

[0103] Operating S610, the vehicle's wheelbase, steering system ratio, and understeer coefficient are obtained. The understeer coefficient characterizes the vehicle's understeer characteristics. The vehicle's wheelbase, steering system ratio, and understeer coefficient are inherent mechanical structure and chassis dynamics properties of the vehicle itself. The wheelbase reflects the physical distance between the front and rear axles; the steering system ratio defines the proportional relationship between the steering actuator (such as the steering wheel or steering motor) angle and the actual wheel deflection angle; and the understeer coefficient quantifies the degree to which the vehicle deviates from neutral steering during steady-state steering motion.

[0104] When operating the S620, the vehicle speed-related yaw gain is determined based on the current longitudinal speed, vehicle wheelbase, steering system transmission ratio, and understeer characteristic coefficient. Combining the dynamic driving speed and inherent chassis parameters, a steering response ratio that matches the vehicle's current physical state is calculated. For example,

[0105]

[0106] Where G(v) is the vehicle speed-related yaw gain, v is the current longitudinal speed, L is the vehicle wheelbase (in meters), and i s K is the gear ratio of the steering system. u A coefficient characterizing understeering characteristics (unit: s) 2 / m 2 ).

[0107] When a vehicle accelerates on urban roads, the control algorithm substitutes the real-time vehicle speed, wheelbase, transmission ratio, and understeer coefficient into the above steady-state response formula to calculate a gain value that changes non-linearly with vehicle speed, which is used to perform the conversion from curvature to feedforward angle.

[0108] According to embodiments of this disclosure, the vehicle yaw gain is not a fixed ratio, but is determined jointly based on the current longitudinal speed, vehicle wheelbase, steering system transmission ratio, and understeer characteristics. This ensures that the conversion from curvature to actuator-side feedforward angle takes into account the vehicle's current speed and lateral steady-state response characteristics. The feedforward angle obtained from the two curvature conversions matches the vehicle's own steering response characteristics, thus physically compensating for the yaw characteristics of the feedforward command that are related to vehicle speed.

[0109] Figure 4 A flowchart illustrating the determination of near and far aiming positions in a vehicle control method according to an embodiment of the present disclosure is shown.

[0110] like Figure 4As shown, based on the aforementioned embodiments, operation S130 may include operations S710 to S740.

[0111] In operation of S710, the cumulative arc length sequence of the reference trajectory and the prediction time corresponding to each prediction node are obtained. The cumulative arc length sequence is the result of mapping the path points in the three-dimensional space of the reference trajectory to a one-dimensional distance scale, representing the actual distance traveled along the curve from the trajectory starting point to each path point. The prediction time represents the timestamp set by the predictive control algorithm for each discrete node within a future time window. The cumulative arc length sequence can be constructed by integrating or accumulating the chord length of the reference trajectory sent from upstream, and the time interval corresponding to each prediction node can be read from the configuration of the predictive controller. For example, for a planned 100-meter reference trajectory, the arc length value of each discrete point on the trajectory from the starting point is calculated; at the same time, if the prediction time domain is 2 seconds and the sampling period is 0.1 seconds, 20 prediction nodes are obtained, along with their corresponding prediction times of 0.1 seconds, 0.2 seconds, up to 2.0 seconds.

[0112] In operation S720, the reference arc length position of each prediction node is determined based on the current longitudinal velocity and the prediction time corresponding to each prediction node. The reference arc length position represents the starting position of the pre-aiming query of the corresponding prediction node on the cumulative arc length axis of the reference trajectory.

[0113] The reference arc length position is a one-dimensional coordinate projected onto a spatial reference trajectory from the prediction nodes in the time dimension. By multiplying the current longitudinal velocity by the prediction time of each node, the expected distance traveled by the vehicle at the corresponding time point can be estimated. This distance is the starting point for the pre-aiming query on the cumulative arc length axis. Let the effective longitudinal velocity be v, and the prediction time of the i-th node in the prediction time domain be t. i Then the reference position of the node on the reference arc length axis can be represented as s i =vt i When the vehicle is traveling at a speed of 10 meters per second, for the 5th prediction node with a prediction time of 0.5 seconds, its reference arc length position is calculated to be 5 meters, that is, the starting point for the pre-aiming query of the node is a place 5 meters away from the current position on the reference trajectory.

[0114] By operating S730, the near-pre-aiming distance and far-pre-aiming distance are added to each reference arc length position to determine the corresponding near-pre-aiming arc length position and far-pre-aiming arc length position.

[0115] On a one-dimensional cumulative arc length axis, the reference position of each node is extended forward by near and far anticipation distances respectively, yielding the arc length coordinates of two spatial observation points. The corresponding calculation formula can be expressed as:

[0116]

[0117] Among them, s n,i Let s be the near-targeting arc length position corresponding to the i-th node. f,i Let s be the far-preview arc length position corresponding to the i-th node. i D is the reference arc length position. n For close aiming distance, D f The target is the far aiming distance. For a node with a reference arc length of 5 meters, if the near aiming distance is 4 meters and the far aiming distance is 10 meters, a simple addition operation determines that the near aiming arc length of the node is 9 meters and the far aiming arc length is 15 meters.

[0118] When operating the S740, the near and far preview arc length positions are determined based on the reference trajectory. Using the calculated arc length coordinates, the corresponding two-dimensional or three-dimensional spatial coordinates are found in reverse on the reference trajectory. Interpolation addressing can be performed in the cumulative arc length sequence of the reference trajectory to extract the geometric points corresponding to the near and far preview arc length positions. For example, using arc length values ​​of 9 meters and 15 meters, linear interpolation is performed in the reference trajectory data table to find the specific latitude and longitude or local coordinate system coordinates 9 meters from the starting point as the near preview position, and the specific coordinates 15 meters from the starting point as the far preview position.

[0119] According to embodiments of this disclosure, each prediction node is mapped to a reference arc length position on the cumulative arc length axis of the reference trajectory, and its respective reference arc length position serves as the starting point for near and far anticipation queries. Dual-scale anticipation is established node-by-node throughout the entire prediction time domain. The feedforward calculation is not performed only once for the vehicle's current position, but rather for each node in the prediction time domain, calculating the corresponding future geometric requirements and generating feedforward information that corresponds one-to-one with each prediction node. This allows the predictive controller to consider future steering requirements in advance during the optimization process, reducing phase lag caused by relying solely on error correction after it has occurred.

[0120] Based on the aforementioned embodiments, operations S810 to S830 may be included before determining the near and far aiming positions corresponding to the prediction node.

[0121] In operation S810, the endpoint of the effective domain of the planned path corresponding to the reference trajectory is determined. The endpoint of the effective domain of the planned path represents the physical end position of the finite-length reference trajectory issued by the upstream planning module. By parsing the data structure of the reference trajectory, the cumulative arc length value corresponding to the last point of the trajectory is extracted as a threshold for judging whether subsequent predicted nodes have exceeded the boundary.

[0122] When operating S820, if the reference arc length position corresponding to the predicted node exceeds the end point of the effective domain of the planned path, the curvature of the last intra-domain reference node before the first reference arc length position exceeding the end point of the effective domain of the planned path is determined as the tail-preserving curvature. The tail-preserving curvature is a constant curvature value extracted at the end of the planned path for extrapolation. When the required aiming distance for the predicted node exceeds the length of the existing reference trajectory, straight-line extension is not used; instead, the last effective node before the boundary violation occurs is found, and its curvature information is extracted. Let the arc length of the end point of the effective domain of the planned path be s. end Traverse the predicted nodes and find the first position s that satisfies the baseline arc length. i >s end The predicted reference node is used to extract the curvature k of the last intradomain reference node preceding that node. end This is used as the tail-preserving curvature. For example, the reference trajectory is 60 meters long, the reference arc length of the 15th predicted node is 58 meters, and the reference arc length of the 16th predicted node is 62 meters. The 16th node exceeds the end of the effective domain, so the curvature value of the 15th node (or the end of the trajectory) is extracted and used as the tail-preserving curvature.

[0123] In operating S830, the curvature, heading, and position of reference nodes exceeding the effective domain endpoint of the planned path are determined node by node based on the tail-maintaining curvature, forming an extended reference sequence covering multiple predicted nodes. The extended reference sequence is a set of virtual trajectory points covering the entire prediction time domain, generated geometrically recursively based on the original planned path. Using the tail-maintaining curvature, constant curvature extrapolation is performed on predicted nodes that exceed the boundary. The curvature of nodes in the extended region directly inherits the tail-maintaining curvature, the heading is calculated by integrating the arc length increment according to this curvature, and the position is recursively derived by integrating the heading at the midpoint. Continuing with the previous example, for subsequent nodes exceeding the 60-meter endpoint, their curvature is uniformly set to 0.02m. -1 The difference in arc length between adjacent nodes is multiplied by 0.02m. -1 The course increment is calculated and accumulated to obtain the new course. Then, the difference between the new course and the arc length is used to calculate the horizontal and vertical coordinate increments through trigonometric functions, generating complete geometric information for the 16th, 17th, and other boundary-crossing nodes, and splicing them into an extended reference sequence covering the entire prediction time domain.

[0124] Based on the foregoing embodiments, operations S910 to S930 may be included after operation S830.

[0125] In operation S910, when the far-predicting arc length position exceeds the end of the extended reference sequence, the out-of-bounds arc length relative to the end of the extended reference sequence is determined. The out-of-bounds arc length represents the difference between the far-predicting arc length position and the cumulative arc length corresponding to the last node of the extended reference sequence. By comparing the far-predicting arc length position with the arc length of the last node of the extended reference sequence, if the far-predicting arc length position is greater than the arc length of the last node, the two are subtracted to calculate the length exceeding the fixed reference time domain. For example, if the total arc length of the extended reference sequence is 80 meters, and the far-predicting arc length position calculated by a certain prediction node is 85 meters, then it is determined that the far-predicting query exceeds the end of the extended sequence, and 80 meters are subtracted from 85 meters to obtain an out-of-bounds arc length of 5 meters.

[0126] When operating the S920, the far-seeking position is determined based on the overshoot arc length by following the tangent corresponding to the final heading of the extended reference sequence. The tangent corresponding to the final heading represents the straight line trajectory formed by the heading of the last node in the extended reference sequence. Starting from the last node of the extended reference sequence, extend forward along the straight line indicated by the heading angle of that node by a distance equal to the overshoot arc length. Calculate the coordinates of the endpoint of this extension using trigonometric functions, and use this coordinate as the far-seeking position after the overshoot. For example, if the coordinates of the final node of the extended reference sequence are (100, 50), the final heading is due east (i.e., 0 radians), and the overshoot arc length is 5 meters, extend forward 5 meters along the tangent in the due east direction to determine the coordinates of the far-seeking position as (105, 50).

[0127] When operating S930, if the near-aiming arc length position exceeds the end of the extended reference sequence, the heading at the end of the extended reference sequence is determined as the reference heading corresponding to the near-aiming position. The reference heading here refers to the fixed tangential direction assigned to the near-aiming position after exceeding the range of the extended reference sequence. When the near-aiming arc length position is also greater than the total arc length of the extended reference sequence, the geometric recursion of the heading is stopped, and the heading angle of the last node of the extended reference sequence is directly extracted and assigned to the near-aiming position that has exceeded the boundary. For example, if the near-aiming arc length position is 82 meters, exceeding the 80-meter end of the extended reference sequence, the heading angle of the end node at the 80-meter mark is read as 0.5 radians, and this 0.5 radians is directly used as the reference heading for the near-aiming position at 82 meters, without performing additional curvature integration.

[0128] According to embodiments of this disclosure, after forming the extended reference sequence, for cases where the pre-aiming query still exceeds the fixed end of the extended reference sequence, the far-pre-aiming position is determined using the end tangent, and the near-pre-aiming heading query that exceeds the boundary is handled using the end heading. Constant curvature recursion is responsible for maintaining the curvature continuity at the end of the planned path, and tangent extension further handles cases where the pre-aiming distance exceeds the extended prediction time domain. The two-level boundary processing works in sequence according to the data flow, avoiding the infinite extension of the constant curvature arc, providing stable geometric query results for out-of-bounds pre-aiming, and improving the continuity of the feedforward sequence under short-path conditions.

[0129] Based on the foregoing embodiments, the vehicle control method may further include operations S1010 to S1030.

[0130] In operation S1010, the remaining path length ahead of the vehicle is obtained. The remaining path length represents the arc distance along the reference trajectory from the vehicle's current position to the end of the effective domain of the planned path. This length value is obtained by calculating the difference between the cumulative arc length at the end of the planned path and the arc length corresponding to the vehicle's current position, or by directly reading the remaining path information from the upstream planning module. For example, if the total length of the reference trajectory issued by the upstream is 50 meters, and the vehicle has currently traveled to the 30th meter of the trajectory, then the remaining path length ahead of the vehicle is 20 meters.

[0131] In operation S1020, when the remaining path length is shorter than the far-pre-aiming distance but longer than the near-pre-aiming distance, the feedforward control sequence is linearly decayed according to the relative position of the remaining path length between the near-pre-aiming distance and the far-pre-aiming distance.

[0132] Linear attenuation refers to gradually reducing the magnitude of the feedforward control input according to a proportional relationship. When the remaining path length is between the near and far aiming distances, the far aiming can no longer obtain the true geometric information of the planned path. At this time, using the near and far aiming distances as boundaries, the relative proportion of the current remaining path length between these two boundaries is calculated, and this proportion is used as the attenuation coefficient multiplied by each feedforward angle in the feedforward control sequence. For example, if the near aiming distance is 5 meters, the far aiming distance is 15 meters, and the current remaining path length is 10 meters, since 10 meters is exactly in the middle of 5 meters and 15 meters, the attenuation ratio is determined to be 0.5. Subsequently, all feedforward angle values ​​in the feedforward control sequence are multiplied by 0.5 to achieve a halving of the control command output.

[0133] In operation S1030, when the remaining path length is not longer than the near-preview distance, the feedforward control sequence is decayed to zero. Decaying to zero means completely canceling feedforward control, relying solely on feedback control or other mechanisms to take over vehicle steering. When the remaining path length further shortens and cannot meet the near-preview distance requirement, it indicates that the actual planned path is about to end, and both near and far-end previews have left the effective planning domain. At this point, all values ​​in the feedforward control sequence are directly set to zero.

[0134] According to embodiments of this disclosure, the feedforward control sequence is attenuated in stages based on the relationship between the remaining path length ahead of the vehicle and the near and far aiming distances. When the vehicle approaches the end of the finite reference path and the remaining path is insufficient to support complete far or near aiming, the feedforward effect gradually weakens with the available path length, rather than suddenly disappearing at the end of the reference trajectory. This reduces the impact of the finite path length on the far aiming estimation and the continuity of the feedforward output, improving the continuity of the feedforward sequence under boundary conditions.

[0135] Figure 5 A block diagram of a vehicle control device according to an embodiment of the present disclosure is shown schematically.

[0136] like Figure 5 As shown, the vehicle control device 1100 may include a first acquisition module 1110, a first determination module 1120, a second determination module 1130, a third determination module 1140, a fourth determination module 1150, a first generation module 1160, and a first control module 1170.

[0137] The first acquisition module 1110 is used to acquire the vehicle's current longitudinal speed, reference trajectory, and multiple prediction nodes, where the prediction nodes represent the vehicle's expected future spatial pose. In some embodiments, the first acquisition module 1110 can be used to perform operation S110 in the vehicle control method described above, which will not be elaborated here.

[0138] The first determining module 1120 is used to determine the near pre-aiming distance and the far pre-aiming distance that match the vehicle speed based on the current longitudinal speed. In some embodiments, the first determining module 1120 can be used to perform operation S120 in the above-described vehicle control method, which will not be described in detail here.

[0139] The second determining module 1130 is used to determine the near-pre-aiming position and the far-pre-aiming position of each predicted node along the reference trajectory, taking the predicted node as the starting point and according to the near-pre-aiming distance and the far-pre-aiming distance, respectively. In some embodiments, the second determining module 1130 can be used to perform operation S130 in the above-described vehicle control method, which will not be elaborated here.

[0140] The third determining module 1140 is used to determine the near-end heading curvature based on the relative heading change relationship between the near-pre-aiming position and the predicted node. In some embodiments, the third determining module 1140 may be used to perform operation S140 in the above-described vehicle control method, which will not be elaborated here.

[0141] The fourth determining module 1150 is used to determine the far-end position curvature based on the lateral relative positional relationship between the far-preview position and the predicted node. In some embodiments, the fourth determining module 1150 may be used to perform operation S150 in the vehicle control method described above, which will not be elaborated here.

[0142] The first generation module 1160 is used to generate a feedforward control sequence corresponding to each prediction node based on the near-end heading curvature and the far-end position curvature. In some embodiments, the first generation module 1160 can be used to perform operation S170 in the above vehicle control method, which will not be described in detail here.

[0143] The first control module 1170 is used to control the vehicle's trajectory during driving based on a feedforward control sequence. In some embodiments, the first control module 1170 may be used to execute operation S170 in the vehicle control method described above, which will not be elaborated here.

[0144] Any one or more of the modules, submodules, units, and subunits according to embodiments of the present disclosure, or at least part of the functions of any one or more of them, can be implemented in one module. Any one or more of the modules, submodules, units, and subunits according to embodiments of the present disclosure can be implemented by dividing them into multiple modules. Any one or more of the modules, submodules, units, and subunits according to embodiments of the present disclosure can be at least partially implemented as hardware circuitry, such as a Field-Programmable Gate Array (FPGA), a Programmable Logic Array (PLA), a System-on-Chip, a System-on-a-Substrate, a System-on-Package, an Application-Specific Integrated Circuit (ASIC), or implemented in hardware or firmware by any other reasonable means of integrating or packaging circuitry, or implemented in software, hardware, or firmware, or in any suitable combination of any of these three implementation methods. Alternatively, one or more of the modules, submodules, units, and subunits according to embodiments of the present disclosure can be at least partially implemented as computer program modules, which, when run, can perform corresponding functions.

[0145] For example, any and more of the first acquisition module 1110, first determination module 1120, second determination module 1130, third determination module 1140, fourth determination module 1150, first generation module 1160, and first control module 1170 can be combined into one module / unit / subunit, or any one of these modules / units / subunits can be split into multiple modules / units / subunits. Alternatively, at least some of the functions of one or more of these modules / units / subunits can be combined with at least some of the functions of other modules / units / subunits and implemented in one module / unit / subunit. According to embodiments of this disclosure, at least one of the first acquisition module 1110, the first determination module 1120, the second determination module 1130, the third determination module 1140, the fourth determination module 1150, the first generation module 1160, and the first control module 1170 can be at least partially implemented as hardware circuits, such as field-programmable gate arrays (FPGAs), programmable logic arrays (PLAs), systems-on-a-chip, systems-on-a-substrate, systems-on-package, application-specific integrated circuits (ASICs), or any other reasonable means of integrating or packaging circuits, or implemented in hardware or firmware, or in any one of software, hardware, and firmware implementations, or in a suitable combination of any of these. Alternatively, at least one of the first acquisition module 1110, the first determination module 1120, the second determination module 1130, the third determination module 1140, the fourth determination module 1150, the first generation module 1160, and the first control module 1170 can be at least partially implemented as computer program modules, which can perform corresponding functions when the computer program module is run.

[0146] It should be noted that the data processing system part in the embodiments of this disclosure corresponds to the data processing method part in the embodiments of this disclosure. The specific description of the data processing system part is referred to in the data processing method part, and will not be repeated here.

[0147] Figure 6 A block diagram of an electronic device suitable for implementing the methods described above, according to embodiments of the present disclosure, is illustrated schematically. Figure 6 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0148] In autonomous or driverless scenarios, the electronic device in this embodiment can be configured as an on-board computing platform, autonomous driving domain controller, or vehicle electronic control unit (ECU) deployed on the driverless vehicle, for real-time processing of vehicle status and trajectory planning data, and for performing feedforward control calculations to control vehicle driving; or, the electronic device can also be a cloud server or remote computing center that communicates with the driverless vehicle, for calculating control commands in the cloud and sending them to the vehicle.

[0149] like Figure 6 As shown, an electronic device 1200 according to an embodiment of the present disclosure includes a processor 1201, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1202 or a program loaded from a storage portion 1208 into a random access memory (RAM) 1203. The processor 1201 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 1201 may also include onboard memory for caching purposes. The processor 1201 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0150] RAM 1203 stores various programs and data required for the operation of electronic device 1200. Processor 1201, ROM 1202, and RAM 1203 are interconnected via bus 1204. Processor 1201 performs various operations of the method flow according to embodiments of the present disclosure by executing programs in ROM 1202 and / or RAM 1203. It should be noted that the programs may also be stored in one or more memories other than ROM 1202 and RAM 1203. Processor 1201 may also perform various operations of the method flow according to embodiments of the present disclosure by executing programs stored in said one or more memories.

[0151] According to embodiments of this disclosure, the electronic device 1200 may further include an input / output (I / O) interface 1205, which is also connected to a bus 1204. The electronic device 1200 may also include one or more of the following components connected to the input / output (I / O) interface 1205: an input section 1206 including a keyboard, mouse, etc.; an output section 1207 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 1208 including a hard disk, etc.; and a communication section 1209 including a network interface card such as a LAN card, modem, etc. The communication section 1209 performs communication processing via a network such as the Internet. A drive 1210 is also connected to the input / output (I / O) interface 1205 as needed. A removable medium 1211, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 1210 as needed so that computer programs read from it can be installed into the storage section 1208 as needed.

[0152] According to embodiments of this disclosure, the method flow according to embodiments of this disclosure can be implemented as a computer software program. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable storage medium, the computer program containing program code for performing the methods shown in the flowchart. In such embodiments, the computer program can be downloaded and installed from a network via communication section 1209, and / or installed from removable medium 1211. When the computer program is executed by processor 1201, it performs the functions defined in the system of embodiments of this disclosure. According to embodiments of this disclosure, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0153] This disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs that, when executed, implement the method according to the embodiments of this disclosure.

[0154] According to embodiments of this disclosure, the computer-readable storage medium can be a non-volatile computer-readable storage medium. Examples include, but are not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this disclosure, the computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0155] For example, according to embodiments of this disclosure, a computer-readable storage medium may include the ROM 1202 and / or RAM 1203 described above and / or one or more memories other than ROM 1202 and RAM 1203.

[0156] Embodiments of this disclosure also include a computer program product comprising a computer program containing program code for performing the methods provided in the embodiments of this disclosure. When the computer program product is run on an electronic device, the program code is used to enable the electronic device to implement the vehicle control methods provided in the embodiments of this disclosure.

[0157] When the computer program is executed by the processor 1201, it performs the functions defined in the system / apparatus of this disclosure embodiments. According to embodiments of this disclosure, the systems, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0158] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices or magnetic storage devices. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and downloaded and installed via communication section 1209, and / or installed from removable medium 1211. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof. According to embodiments of this disclosure, program code for executing the computer programs provided in embodiments of this disclosure can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code may be executed entirely on a user computing device, partially on a user device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing devices can be connected to user computing devices via any type of network, including local area networks (LANs) or wide area networks (WANs), or they can be connected to external computing devices (e.g., via the Internet using an Internet service provider).

[0159] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions. Those skilled in the art will understand that the features described in the various embodiments of the present disclosure can be combined and / or combined in various ways, even if such combinations are not explicitly described in the present disclosure. In particular, the features described in the various embodiments of this disclosure may be combined and / or combined in various ways without departing from the spirit and teachings of this disclosure. All such combinations and / or combinations fall within the scope of this disclosure.

[0160] The embodiments of this disclosure have been described above. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of this disclosure. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of this disclosure, and all such substitutions and modifications should fall within the scope of this disclosure.

Claims

1. A vehicle control method, characterized in that, include: The vehicle's current longitudinal velocity, reference trajectory, and multiple prediction nodes are obtained, whereby the prediction nodes represent the vehicle's expected future spatial pose. Based on the current longitudinal speed, determine the near-aiming distance and far-aiming distance that match the vehicle speed, respectively; For each of the predicted nodes, along the reference trajectory, starting from the predicted node, the near-pre-aiming position and the far-pre-aiming position corresponding to the predicted node are determined according to the near-pre-aiming distance and the far-pre-aiming distance; The near-end heading curvature is determined based on the relative heading change relationship between the near-pre-aiming position and the predicted node; The curvature of the far end position is determined based on the lateral relative positional relationship between the far pre-aiming position and the predicted node; Based on the near-end heading curvature and the far-end position curvature, a feedforward control sequence corresponding to each prediction node is generated; The vehicle's trajectory during driving is controlled based on the feedforward control sequence.

2. The vehicle control method according to claim 1, characterized in that, The step of determining the near-aiming distance and far-aiming distance matching the vehicle speed based on the current longitudinal speed includes: The near-aiming distance is determined based on the product of the current longitudinal velocity and the near-aiming time scale and the near-aiming minimum distance, wherein the near-aiming distance is the larger of the two. The far-preview distance is determined based on the product of the current longitudinal velocity and the far-preview time scale and the far-preview minimum distance, wherein the far-preview distance is the larger of the two. Wherein, the far-viewing timescale is not less than the near-viewing timescale, and the far-viewing minimum distance is not less than the near-viewing minimum distance.

3. The vehicle control method according to claim 1, characterized in that, The step of determining the near-end heading curvature based on the relative heading change relationship between the near-pre-aiming position and the predicted node includes: Determine the reference heading corresponding to the near-targeting position based on the reference trajectory; The predicted heading corresponding to the predicted node is determined based on the expected spatial pose represented by the predicted node; Determine the heading difference between the reference heading and the predicted heading; The heading difference is periodically normalized according to the period of the heading angle to obtain the normalized heading difference; The near-end heading curvature is determined based on the normalized heading difference and the near-pre-aiming distance.

4. The vehicle control method according to claim 1, characterized in that, Determining the far-end position curvature based on the lateral relative positional relationship between the far-preview position and the predicted node includes: The predicted position and predicted heading of the predicted node are determined based on the expected spatial pose represented by the predicted node. Determine the planar displacement of the far-preview position relative to the predicted position and the Euclidean distance between them; The planar displacement is transformed to a local coordinate system based on the predicted heading, and the lateral component of the planar displacement in the local coordinate system is determined. The curvature at the distal position is determined by the ratio between twice the lateral component and the square of the Euclidean distance.

5. The vehicle control method according to claim 1, characterized in that, The step of generating a feedforward control sequence corresponding to each prediction node based on the near-end heading curvature and the far-end position curvature includes: Obtain the vehicle speed-related vehicle yaw gain, which characterizes the steady-state mapping relationship between the vehicle steering actuator side steering angle and the vehicle yaw rate. Using the same vehicle speed-related vehicle yaw gain, the near-end heading curvature and the far-end position curvature are respectively converted into near-end feedforward angle and far-end feedforward angle; The near-end feedforward angle and the far-end feedforward angle are weighted and fused according to the near-end weight and the far-end weight to obtain the fused feedforward angle corresponding to each prediction node, wherein the sum of the near-end weight and the far-end weight is one. The feedforward control sequence is generated based on the fusion feedforward angle corresponding to each prediction node.

6. The vehicle control method according to claim 5, characterized in that, The acquisition of vehicle speed-related vehicle yaw gain includes: The vehicle wheelbase, steering system transmission ratio, and understeer characteristic coefficient are obtained, wherein the understeer characteristic coefficient characterizes the understeer characteristics of the vehicle. The speed-related vehicle yaw gain is determined based on the current longitudinal speed, the vehicle wheelbase, the steering system transmission ratio, and the understeer characteristic coefficient.

7. The vehicle control method according to claim 1, characterized in that, For each predicted node, determining the near-aiming position and far-aiming position of the predicted node along the reference trajectory, starting from the predicted node and according to the near-aiming distance and far-aiming distance, includes: Obtain the cumulative arc length sequence of the reference trajectory and the prediction time corresponding to each prediction node; Based on the current longitudinal velocity and the prediction time corresponding to each prediction node, the reference arc length position corresponding to each prediction node is determined, wherein the reference arc length position represents the pre-aiming query start position of the corresponding prediction node on the cumulative arc length axis of the reference trajectory. Based on each of the aforementioned reference arc length positions, the near pre-aiming distance and the far pre-aiming distance are added respectively to determine the corresponding near pre-aiming arc length position and far pre-aiming arc length position; The near-pre-aiming position and the far-pre-aiming position are determined based on the reference trajectory.

8. The vehicle control method according to claim 7, characterized in that, Before determining the near-aiming position and far-aiming position corresponding to each predicted node along the reference trajectory, starting from the predicted node and according to the near-aiming distance and far-aiming distance, the method further includes: Determine the endpoint of the effective domain of the planned path corresponding to the reference trajectory; If the reference arc length position corresponding to the predicted node exceeds the end point of the effective domain of the planned path, the curvature of the last reference node in the domain before the first reference arc length position that exceeds the end point of the effective domain of the planned path is determined as the tail-preserving curvature. Based on the tail curvature, the curvature, heading, and position of reference nodes beyond the end of the effective domain of the planned path are determined node by node, forming an extended reference sequence covering the multiple predicted nodes.

9. The vehicle control method according to claim 8, characterized in that, After determining the curvature, heading, and position of reference nodes beyond the endpoint of the effective domain of the planned path node by node based on the tail-preserving curvature, forming an extended reference sequence covering the multiple predicted nodes, the method further includes: When the far-preview arc length position exceeds the end of the extended reference sequence, the out-of-bounds arc length of the far-preview arc length position relative to the end of the extended reference sequence is determined; The far-seeking position is determined based on the over-boundary arc length by tangent to the end heading of the extended reference sequence. When the near-pre-aiming arc length position exceeds the end of the extended reference sequence, the end heading of the extended reference sequence is determined as the reference heading corresponding to the near-pre-aiming position.

10. The vehicle control method according to any one of claims 1 to 9, characterized in that, Also includes: Get the remaining path length ahead of the vehicle; When the remaining path length is shorter than the far-pre-aiming distance but longer than the near-pre-aiming distance, the feedforward control sequence is linearly attenuated according to the relative position of the remaining path length between the near-pre-aiming distance and the far-pre-aiming distance. When the remaining path length is not longer than the near-pre-aiming distance, the feedforward control sequence is decayed to zero.