An unmanned vehicle road surface disturbance prediction and active stability control method

Abstract

The present application belongs to the technical field of vehicle stability control, and in particular to a method for predicting road surface disturbance and actively stabilizing unmanned vehicles. The method comprises the following steps: S1: collecting road environment information and vehicle motion state; S2: time synchronization and space unification of multi-source information; S3: local three-dimensional road surface geometry modeling; S4: road surface disturbance prediction modeling; S5: disturbance feedforward compensation control quantity calculation; S6: attitude feedback stabilizing control calculation; S7: feedforward control and feedback control fusion: the disturbance feedforward compensation control quantity and the attitude feedback control quantity are fused to obtain the final stabilizing control input of the vehicle; S8: actuator attitude adjustment: according to the stabilizing control input, the vehicle actuator is controlled and adjusted. The present application can improve the attitude stability performance and anti-disturbance ability of unmanned vehicles under complex road conditions, and enhance the running safety and adaptability of vehicles in actual application scenarios.

Description

Technical Field

[0001] This invention relates to the field of vehicle stability control technology, specifically to a method for predicting road disturbances and actively stabilizing and controlling unmanned vehicles. Background Technology

[0002] With the development of autonomous driving technology, intelligent sensing technology, and vehicle motion control technology, unmanned vehicles have been widely used in scenarios such as agricultural operations, logistics transportation, mine inspection, and field exploration. In these application environments, unmanned vehicles typically need to operate on unstructured or semi-structured road conditions, such as gravel roads, muddy roads, sloping roads, and temporary construction roads. These types of road surfaces often have significant irregularities and randomness, characterized by large elevation fluctuations, frequent slope changes, uneven local distribution of bumps and depressions, and unstable adhesion conditions.

[0003] When autonomous vehicles travel on such complex road surfaces, the contact state between the tires and the road surface is constantly changing. The resulting external stimuli are transmitted to the vehicle body through the chassis structure and suspension system, which can easily cause fluctuations in the vehicle's pitch, roll, and yaw attitude. Instability in vehicle attitude not only reduces ride comfort and safety but may also affect the measurement accuracy of onboard sensors, thereby adversely impacting environmental perception, path planning, and task execution.

[0004] In existing technologies, stability control methods based on attitude state feedback are commonly used to improve vehicle stability. For example, inertial measurement units (IMUs) acquire vehicle attitude angle and angular velocity information, and control inputs are calculated based on the error between the target attitude and the actual attitude, thereby adjusting the vehicle's drive system or steering system. This type of control method can achieve certain results under relatively smooth or minimally disturbed road conditions, but it still has limitations in complex road environments.

[0005] With the development of visual sensors and LiDAR technology, some autonomous vehicles have acquired the ability to perceive the road environment ahead, such as identifying road boundaries through cameras or acquiring 3D point cloud information through LiDAR. However, in current applications, this perceived information is mainly used for obstacle detection or path planning, and is rarely used in the vehicle stability control process. In other words, current technology has not fully utilized the spatial geometry of the road surface to characterize the external disturbances that the vehicle will be subjected to.

[0006] From a vehicle dynamics perspective, road surface unevenness is essentially an external excitation with spatial distribution characteristics. Its impact on vehicle attitude is not only related to changes in road surface height, but also closely related to factors such as vehicle speed, acceleration, and vehicle structural parameters. If control adjustments are made solely based on the vehicle's current attitude state, it is difficult to accurately reflect the transmission process of disturbances between the environment, tires, chassis, and vehicle body. Therefore, control systems often lack foresight.

[0007] Therefore, in the field of unmanned vehicle stability control, how to effectively combine road environment perception information with vehicle attitude control process, establish a road disturbance expression method oriented towards vehicle dynamic response, and realize disturbance advance compensation in the control system has become an important technical direction for improving the stable operation capability of unmanned vehicles under complex terrain conditions. Summary of the Invention

[0008] The purpose of this section is to outline some aspects of the embodiments of the present invention and to briefly describe some preferred embodiments. Simplifications or omissions may be made in this section, as well as in the abstract and title of this application, to avoid obscuring the purpose of these documents; however, such simplifications or omissions should not be construed as limiting the scope of the invention.

[0009] To address the aforementioned technical problems, according to one aspect of the present invention, the present invention provides the following technical solution:

[0010] A method for predicting road disturbances and actively maintaining stability in unmanned vehicles includes the following steps:

[0011] S1: Road environment information and vehicle motion status acquisition: Real-time acquisition of road surface spatial information in front of the vehicle and vehicle motion status information to establish a three-dimensional road surface set;

[0012] S2: Multi-source information time synchronization and spatial unification: Unified processing of multi-source data collected;

[0013] S3: Local 3D road surface geometry modeling: Based on the synchronized road surface point cloud data, 3D geometry modeling is performed on the area in front of the vehicle. After obtaining the road surface model, the road surface geometric feature parameters are further calculated, including: road surface slope, road surface curvature, and road surface roughness.

[0014] S4: Road disturbance prediction modeling: Based on the current motion state of the vehicle and the geometric characteristics of the road surface, establish a disturbance prediction model for the vehicle's dynamic response to estimate the disturbances that the vehicle will experience in the future time window.

[0015] S5: Disturbance feedforward compensation control quantity calculation: Based on the predicted future disturbance, calculate the disturbance feedforward compensation control quantity so that the control system can pre-adjust before the disturbance takes effect;

[0016] S6: Attitude Feedback Stability Control Calculation: Based on feedforward compensation, a vehicle attitude feedback control mechanism is established, specifically using a cascade control structure: Attitude Angle Control Loop: Generates the desired angular velocity based on the error between the target attitude and the actual attitude; Angular Velocity Control Loop: Calculates the control input based on the error between the desired angular velocity and the actual angular velocity;

[0017] S7: Fusion of feedforward and feedback control: The disturbance feedforward compensation control quantity and the attitude feedback control quantity are fused to obtain the final vehicle stability control input;

[0018] S8: Actuator attitude adjustment: Control and adjust the vehicle actuator according to the balance control input.

[0019] As a preferred embodiment of the unmanned vehicle road disturbance prediction and active stability control method described in this invention, in step S1, the road environment information includes: road surface image information and road surface three-dimensional point cloud information; the vehicle motion state information includes: vehicle attitude angle, vehicle angular velocity, vehicle linear acceleration, and vehicle speed.

[0020] The method for establishing the three-dimensional road surface set is as follows: Let the vehicle coordinate system be denoted as... ,in The axis is the direction of vehicle movement. The axis is the lateral direction of the vehicle. The axis is the vertical direction of the vehicle, and the set of points on the road surface ahead collected by the lidar is as follows:

[0021]

[0022] in, Indicates the number of sampling points. Indicates the first The three-dimensional coordinates of a road surface point in the vehicle coordinate system;

[0023] The vehicle state vector is defined as:

[0024]

[0025] in, This is the roll angle. The pitch angle, Yaw angle The roll rate is angular velocity. The pitch angular velocity, Yaw angular velocity, For the longitudinal speed of the vehicle, , , These represent the linear accelerations of the vehicle in the longitudinal, lateral, and vertical directions, respectively.

[0026] As a preferred embodiment of the unmanned vehicle road disturbance prediction and active stability control method described in this invention, the unified processing in S2 includes: aligning the data of various sensors in time using a time synchronization algorithm; uniformly converting the road point cloud information and vehicle motion state information to the vehicle coordinate system or global coordinate system through coordinate calibration and coordinate transformation; and filtering and fusing multi-source information through a state fusion algorithm to obtain a unified environmental state vector.

[0027] The specific method is as follows: Let the lidar timestamp be... IMU timestamp is Wheel speed timestamp is Select a unified control time Then, various types of data are mapped to the same moment through interpolation:

[0028]

[0029] in, and These are the sampling times adjacent to the control time, respectively. These are the estimated vehicle states after synchronization.

[0030] For spatial unification, let the extrinsic transformation matrix from the lidar coordinate system to the vehicle coordinate system be:

[0031]

[0032] in, For rotation matrix, If the translation vector is used, then the lidar point After conversion to the vehicle coordinate system, it becomes:

[0033]

[0034] Through the above processing, road surface point cloud and vehicle status information under a unified time reference and a unified coordinate system are obtained.

[0035] As a preferred embodiment of the unmanned vehicle road disturbance prediction and active stability control method described in this invention, the three-dimensional geometric modeling in S3 specifically includes: filtering and removing outliers from point cloud data; identifying road surface areas using a ground point extraction algorithm; and establishing a local road surface height function model using a surface fitting method.

[0036] The specific method is as follows: Select a length in front of the vehicle... Width is Using the road surface area as the modeling window, a quadratic surface fitting is performed on the ground points within the window to obtain a local road surface height function model:

[0037]

[0038] in, Indicates the forward distance of the vehicle. Indicates the lateral distance of the vehicles. Indicates the road surface height at the corresponding point; The surface coefficients to be estimated;

[0039] The surface coefficients are solved using the least squares method, i.e., minimizing the objective function:

[0040]

[0041] Based on the fitting results, calculate the road surface geometric characteristic parameters;

[0042] The method for calculating the road surface slope characteristic parameters is as follows: the road surface at... direction and The first-order partial derivatives in the directions are as follows:

[0043]

[0044]

[0045] The local slope is defined as follows:

[0046]

[0047] in, Indicates the road surface at point The slope amplitude at the location;

[0048] The method for calculating the road surface curvature characteristic parameters is as follows: To characterize the rate of change of local undulations, a second-order change quantity is defined:

[0049]

[0050] in, Represents the longitudinal curvature component. Represents the lateral curvature component. Indicates the coupled curvature components;

[0051] The method for calculating the road surface roughness parameter is as follows: Let the actual point cloud height be... The height of the fitted surface is The local roughness is then defined as:

[0052]

[0053] in, It represents the road surface roughness, reflecting the degree of high-frequency undulations near the fitted plane.

[0054] As a preferred embodiment of the unmanned vehicle road disturbance prediction and active stability control method described in this invention, the specific method of step S4 is as follows: Let the prediction time domain length be... The vehicle in the future The forward distance reached is approximately:

[0055]

[0056] in, For future preview location, The current longitudinal velocity, This represents the current longitudinal acceleration;

[0057] Let the positions of the centers of the left and right wheels of the vehicle in the lateral direction be respectively. and ,in Given the wheelbase, the preview heights for the left and right wheels are as follows:

[0058]

[0059] Let the equivalent preview position difference in the front and rear wheelbase directions of the vehicle be... The height difference in the front and rear directions is expressed as:

[0060]

[0061] The height difference in the left and right directions is expressed as:

[0062]

[0063] Therefore, estimates of pitch and roll disturbances are established:

[0064]

[0065]

[0066] in, To predict pitch disturbance angle, To predict the roll disturbance angle;

[0067] Further considering the coupling effect of roughness and velocity, a comprehensive disturbance prediction quantity is constructed:

[0068]

[0069] in, This is the predicted amount of roll disturbance. This is the pitch disturbance prediction quantity; , , , These are the perturbation mapping coefficients, obtained through experimental calibration or system identification;

[0070] In actual control, the weighted average value in the prediction time domain is selected as the equivalent disturbance input for the current control cycle:

[0071] .

[0072] As a preferred embodiment of the unmanned vehicle road disturbance prediction and active stability control method described in this invention, the specific method of step S5 is as follows: based on the predicted disturbance amount... Calculate the disturbance feedforward compensation control quantity, and assume the feedforward gain matrix is:

[0073]

[0074] The disturbance feedforward control quantity is:

[0075]

[0076] in, These represent the feedforward compensation amounts for the roll and pitch directions, respectively. , For feedforward compensation gain;

[0077] If we further consider the rate of change of the disturbance, we add a predictive differential compensation term:

[0078]

[0079] in, The perturbation rate of change gain matrix, This represents the rate of change of the predicted disturbance.

[0080] As a preferred embodiment of the unmanned vehicle road disturbance prediction and active stability control method described in this invention, wherein the attitude angle control loop in S6 is:

[0081] Let the target roll angle and target pitch angle be respectively , The actual roll angle and pitch angle are respectively , Then the attitude error is:

[0082]

[0083] The outer ring generates the desired angular velocity command:

[0084]

[0085]

[0086] in, For the desired roll rate, The desired pitch rate; , , These are the outer ring ratio, integral, and differential coefficients of the roll attitude angle, respectively. , , These are the outer ring proportional, integral, and differential coefficients of the pitch attitude angle, respectively.

[0087] Angular velocity control loop:

[0088] Let the actual roll rate and pitch rate be respectively , Then the angular velocity error is:

[0089]

[0090] The inner loop control output is:

[0091]

[0092]

[0093] in, , For feedback control; , , These are the inner loop parameters for roll rate; , , These are the inner loop parameters for pitch angular velocity.

[0094] As a preferred embodiment of the unmanned vehicle road disturbance prediction and active stability control method described in this invention, the specific method of S7 is as follows: feedforward compensation and feedback control are integrated to obtain the final control input:

[0095]

[0096] in,

[0097]

[0098] To avoid control saturation, an amplitude limit constraint is added:

[0099]

[0100] in, Represents the amplitude limiting function. and These represent the minimum and maximum allowed outputs for the corresponding control channels, respectively.

[0101] As a preferred embodiment of the unmanned vehicle road disturbance prediction and active stability control method described in this invention, the specific method of S8 is as follows: when the vehicle adopts a left-right wheel differential drive mode, the attitude control quantity is allocated as the additional torque of the left and right drive wheels:

[0102]

[0103] in, and Additional torque for the left and right wheels respectively. This is the torque distribution coefficient;

[0104] When a vehicle is equipped with active suspension, pitch and roll control values ​​are also mapped to suspension damping adjustment values:

[0105]

[0106]

[0107] in, , These represent the damping adjustment amounts of the front and rear suspensions, respectively. , These represent the damping adjustment amounts for the left and right suspensions, respectively. , , , For the mapping coefficient of the actuator.

[0108] Compared with the prior art, the beneficial effects of the present invention are: 1. By collecting road environment information in front of the vehicle and establishing a local three-dimensional road surface geometric model, the present invention enables the control system to obtain the road surface spatial change characteristics of the area that the vehicle is about to pass through, thereby providing prior environment input for vehicle attitude stability control, which is conducive to improving the vehicle's adaptability to complex terrain conditions.

[0109] 2. This invention establishes a disturbance prediction model that couples road surface geometry with vehicle motion state. It can estimate in advance the road surface excitation that the vehicle may receive within the future prediction time window, enabling the control system to generate control presets before the disturbance acts on the vehicle, thereby reducing control lag and improving the timeliness of vehicle attitude adjustment.

[0110] 3. This invention integrates disturbance feedforward compensation control and attitude feedback control, enabling the vehicle control system to simultaneously possess disturbance advance suppression capability and closed-loop stability adjustment capability. This helps to reduce the amplitude of vehicle pitch and roll attitude fluctuations, reduce control oscillations, and shorten attitude recovery time, thereby improving vehicle ride comfort.

[0111] 4. In one embodiment, by constructing a cascade control structure consisting of an attitude angle control loop and an angular velocity control loop, the vehicle attitude control system can have a fast dynamic response capability while ensuring control accuracy, thereby maintaining good operational stability under high-frequency road disturbances or speed change conditions.

[0112] 5. Improved vehicle attitude stability reduces attitude disturbances of onboard cameras and LiDAR sensors, thereby enhancing the stability and reliability of environmental perception results and further improving the quality of unmanned vehicle path tracking and task execution.

[0113] This invention can improve the attitude stability and anti-disturbance ability of unmanned vehicles under complex road conditions, and enhance the operational safety and adaptability of vehicles in practical application scenarios. Attached Figure Description

[0114] To more clearly illustrate the technical solutions of the embodiments of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and detailed embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0115] Figure 1 This is a flowchart of a method for predicting road disturbances and actively stabilizing control of unmanned vehicles according to the present invention;

[0116] Figure 2 This is a schematic diagram of a local three-dimensional road surface set model in front of the vehicle, which is a method for predicting road disturbances and actively stabilizing control of unmanned vehicles according to the present invention.

[0117] Figure 3 This is a block diagram illustrating the disturbance feedforward compensation control principle of the unmanned vehicle road disturbance prediction and active stabilization control method of the present invention. Detailed Implementation

[0118] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0119] Secondly, the present invention is described in detail with reference to the schematic diagrams. When detailing the embodiments of the present invention, for ease of explanation, the cross-sectional views illustrating the device structure may be partially enlarged, not according to the usual scale. Furthermore, the schematic diagrams are merely examples and should not limit the scope of protection of the present invention. In addition, actual fabrication should include three-dimensional spatial dimensions of length, width, and depth.

[0120] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0121] The purpose of this invention is to provide a method for predicting road disturbances and actively stabilizing control of unmanned vehicles. By collecting and modeling information about the road environment ahead of the vehicle and combining this information with the vehicle's motion state to establish a road disturbance prediction mechanism, the control system can generate feedforward compensation control quantities before disturbances occur. Simultaneously, by integrating disturbance feedforward compensation with attitude feedback control, active stabilization adjustment of the vehicle's attitude is achieved, thereby improving the stability, disturbance resistance, and operational safety of unmanned vehicles in complex road environments.

[0122] This invention first establishes a calculable local road surface height function model based on the spatial information of the road surface ahead. Then, it derives the predicted roll and pitch disturbances of the vehicle in the future time domain from this model, and further maps these predicted disturbances directly to control feedforward compensation quantities. Finally, it integrates this with cascade feedback control. In other words, the core of this invention is not "correcting attitude deviations after detection," but rather "intervening in advance based on the geometric trends of the road surface ahead before the disturbances act on the vehicle body." Therefore, this invention exhibits significant forward-looking control characteristics.

[0123] For details, please refer to Figure 1 A method for predicting road disturbances and actively maintaining stability in unmanned vehicles includes the following steps:

[0124] S1: Road environment information and vehicle motion status acquisition: During vehicle operation, real-time road environment information and vehicle motion status information within a certain space in front of the vehicle are acquired.

[0125] The road environment information includes: road surface image information and road surface 3D point cloud information.

[0126] Vehicle motion status information includes: vehicle attitude angle, vehicle angular velocity, vehicle linear acceleration, and vehicle speed.

[0127] The above information is used to characterize the vehicle's current motion state and the road surface geometry of the area it will pass through in the future.

[0128] like Figure 2 As shown, the specific method is as follows: During vehicle operation, real-time spatial information of the road surface in front of the vehicle and the vehicle's own motion status information are collected. Let the vehicle coordinate system be denoted as... ,in The axis is the direction of vehicle movement. The axis is the lateral direction of the vehicle. The axis is the vertical direction of the vehicle. The set of points on the road ahead collected by the lidar is as follows:

[0129]

[0130] in, Indicates the number of sampling points. Indicates the first The three-dimensional coordinates of a road surface point in the vehicle coordinate system. The vehicle state vector is defined as:

[0131]

[0132] in, This is the roll angle. The pitch angle, Yaw angle The roll rate is angular velocity. The pitch angular velocity, Yaw angular velocity, For the longitudinal speed of the vehicle, , , These represent the linear accelerations of the vehicle in the longitudinal, lateral, and vertical directions, respectively.

[0133] S2: Multi-source information time synchronization and spatial unification: Unified processing of collected multi-source data, including:

[0134] 1) Time synchronization algorithms are used to align data from various sensors, ensuring that data from different sources have a unified time reference;

[0135] 2) Through coordinate calibration and coordinate transformation, the road surface point cloud information and vehicle motion state information are uniformly converted to the vehicle coordinate system or the global coordinate system;

[0136] 3) The multi-source information is filtered and fused through a state fusion algorithm to obtain a unified environmental state vector.

[0137] After processing, we can obtain: highly consistent road space data and smooth and stable vehicle state estimation results.

[0138] The specific method is as follows: Since the sampling frequencies of road surface point cloud, image, IMU, and wheel speed data are different, time synchronization is required. Let the lidar timestamp be... IMU timestamp is Wheel speed timestamp is Select a unified control time Then, various types of data are mapped to the same moment through interpolation:

[0139]

[0140] in, and These are the sampling times adjacent to the control time, respectively. This is the estimated vehicle state after synchronization.

[0141] For spatial unification, let the extrinsic transformation matrix from the lidar coordinate system to the vehicle coordinate system be:

[0142]

[0143] in, For rotation matrix, Let be the translation vector. Then the lidar point After conversion to the vehicle coordinate system, it becomes:

[0144]

[0145] Through the above processing, road point cloud and vehicle state information under a unified time reference and a unified coordinate system are obtained, providing consistent input for subsequent modeling.

[0146] S3: Local 3D Road Surface Geometry Modeling: Based on synchronized road surface point cloud data, perform 3D geometry modeling of the area in front of the vehicle. Specifically includes:

[0147] 1) Filter and remove outliers from the point cloud data;

[0148] 2) Identify road surface areas using ground point extraction algorithms;

[0149] 3) Establish a local road surface height function model using surface fitting method.

[0150] After obtaining the road surface model, the geometric characteristic parameters of the road surface are further calculated, including: road surface slope, road surface curvature, and road surface roughness. These characteristics are used to characterize the degree of road surface unevenness and its spatial variation trend.

[0151] The specific method is as follows: Select a length in front of the vehicle... Width is The road surface area is used as the modeling window. A quadratic surface fitting is performed on the ground points within the window to obtain a local road surface height function model:

[0152]

[0153] in, Indicates the forward distance of the vehicle. Indicates the lateral distance of the vehicles. Indicates the road surface height at the corresponding point; These are the surface coefficients to be estimated.

[0154] The surface coefficients are solved using the least squares method, i.e., minimizing the objective function:

[0155]

[0156] Based on the fitting results, calculate the road surface geometric characteristic parameters.

[0157] 1. Slope characteristics: The road surface is in direction and The first-order partial derivatives in the directions are as follows:

[0158]

[0159]

[0160] The magnitude of the local slope can then be defined as:

[0161]

[0162] in, Indicates the road surface at point The slope amplitude at that location.

[0163] 2. Curvature characteristics: To characterize the rate of change of local undulations, a second-order change quantity is defined:

[0164]

[0165] in, Represents the longitudinal curvature component. Represents the lateral curvature component. This represents the coupled curvature components.

[0166] 3. Roughness characteristics: Assume the actual point cloud height is... The height of the fitted surface is The local roughness is then defined as:

[0167]

[0168] in, It represents the road surface roughness, reflecting the degree of high-frequency undulations near the fitted plane.

[0169] S4: Road Disturbance Prediction Modeling: Based on the current motion state of the vehicle and the geometric characteristics of the road surface, establish a disturbance prediction model for the vehicle's dynamic response to estimate the disturbances that the vehicle will experience in the future time window.

[0170] The prediction model comprehensively considers: changes in road height, road slope, road curvature distribution, road roughness, vehicle speed, vehicle acceleration, and vehicle angular velocity. This establishes a spatiotemporal coupling relationship for the disturbance. The predicted disturbance can be further decomposed into: a roll direction disturbance component, a pitch direction disturbance component, and a component representing the excitation effect of the external road surface on the vehicle's attitude.

[0171] The specific method involves coupling road surface geometry with vehicle motion state to construct a posture perturbation prediction model within a future time window. Let the prediction time domain length be... The vehicle in the future The forward distance reached is approximately:

[0172]

[0173] in, For future preview location, The current longitudinal velocity, This represents the current longitudinal acceleration.

[0174] Let the positions of the centers of the left and right wheels of the vehicle in the lateral direction be respectively. and ,in This is the wheelbase. Therefore, the preview heights for the left and right wheels are as follows:

[0175]

[0176] Let the equivalent preview position difference in the front and rear wheelbase directions of the vehicle be... The height difference in the front and rear directions can be expressed as:

[0177]

[0178] The height difference in the left and right directions can be expressed as:

[0179]

[0180] From this, estimates of pitch and roll disturbances can be established:

[0181]

[0182]

[0183] in, To predict pitch disturbance angle, To predict the roll disturbance angle.

[0184] Further considering the coupling effect of roughness and velocity, a comprehensive disturbance prediction quantity can be constructed:

[0185]

[0186] in, This is the predicted amount of roll disturbance. This is the pitch disturbance prediction quantity; , , , These are the perturbation mapping coefficients, which can be obtained through experimental calibration or system identification.

[0187] In actual control, the weighted average value in the prediction time domain can be selected as the equivalent disturbance input for the current control cycle:

[0188] .

[0189] S5: Calculation of Disturbance Feedforward Compensation Control Quantity: Based on the predicted future disturbance, calculate the disturbance feedforward compensation control quantity, enabling the control system to pre-adjust before the disturbance occurs. The compensation control quantity is obtained through the mapping relationship between the disturbance compensation gain and the predicted disturbance. This step achieves: earlier intervention of control action and reduced control lag.

[0190] like Figure 3 As shown, the specific method is as follows: based on the predicted disturbance amount... Calculate the disturbance feedforward compensation control quantity. Let the feedforward gain matrix be:

[0191]

[0192] The disturbance feedforward control quantity is:

[0193]

[0194] in, These represent the feedforward compensation amounts for the roll and pitch directions, respectively. , This is the gain for feedforward compensation.

[0195] If the rate of change of the disturbance is further considered, a predictive differential compensation term can be added:

[0196]

[0197] in, The perturbation rate of change gain matrix, This represents the rate of change of the predicted disturbance.

[0198] S6: Attitude Feedback Stability Control Calculation: Based on feedforward compensation, a vehicle attitude feedback control mechanism is established. Specifically, a cascade control structure is adopted.

[0199] 1) Attitude angle control loop: Generates the desired angular velocity based on the error between the target attitude and the actual attitude;

[0200] 2) Angular velocity control loop: The control input is calculated based on the error between the desired angular velocity and the actual angular velocity.

[0201] This control structure can improve control response speed and stability.

[0202] Attitude angle control loop: Let the target roll angle and target pitch angle be respectively , The actual roll angle and pitch angle are respectively , Then the attitude error is:

[0203]

[0204] The outer ring generates the desired angular velocity command:

[0205]

[0206]

[0207] in, For the desired roll rate, The desired pitch rate; , , These are the outer ring ratio, integral, and differential coefficients of the roll attitude angle, respectively. , , These are the outer ring proportional, integral, and differential coefficients of the pitch attitude angle, respectively.

[0208] Angular velocity control loop: Let the actual roll angular velocity and pitch angular velocity be respectively... , Then the angular velocity error is:

[0209]

[0210] The inner loop control output is:

[0211]

[0212]

[0213] in, , For feedback control; , , These are the inner loop parameters for roll rate; , , Pitch angular velocity inner loop parameters

[0214] S7: Fusion of Feedforward and Feedback Control: The disturbance feedforward compensation control quantity and the attitude feedback control quantity are fused to obtain the final vehicle stability control input. Through this fusion mechanism, the following are achieved: early disturbance suppression, rapid attitude recovery, and reduced control oscillation.

[0215] The specific method is as follows: feedforward compensation and feedback control are integrated to obtain the final control input.

[0216]

[0217] in,

[0218]

[0219] To avoid control saturation, a limiting constraint can be added:

[0220]

[0221] in, Represents the amplitude limiting function. and These represent the minimum and maximum allowed outputs for the corresponding control channels, respectively.

[0222] S8: Actuator Attitude Adjustment: Based on the stability control input, the vehicle actuators are controlled and adjusted, including: drive differential adjustment, steering angle adjustment, and suspension stiffness or damping adjustment. This achieves active attitude stability control of the vehicle under complex road conditions.

[0223] The specific method is as follows: When the vehicle adopts a left-right wheel differential drive mode, the attitude control quantity can be distributed as additional torque to the left and right drive wheels:

[0224]

[0225] in, and Additional torque for the left and right wheels respectively. This is the torque distribution coefficient.

[0226] When a vehicle is equipped with an active suspension, the pitch and roll control values ​​can also be mapped to the suspension damping adjustment value:

[0227]

[0228]

[0229] in, , These represent the damping adjustment amounts of the front and rear suspensions, respectively. , These represent the damping adjustment amounts for the left and right suspensions, respectively. , , , For the mapping coefficient of the actuator.

[0230] Example

[0231] This invention applies to a four-wheel independently driven unmanned inspection vehicle. The vehicle has a body length of 1.2m and a wheelbase of... The equivalent preview wheelbase is 0.8m. The vehicle has a height of 0.9m. A 3D LiDAR is mounted on the front of the vehicle, an inertial measurement unit is mounted on the top, and wheel speed encoders are installed on the rear wheels. The onboard controller is an embedded computing platform. When the vehicle travels at 1.5m / s on a mixed gravel and undulating road surface, the LiDAR collects real-time point cloud data of the road surface within an 8m range ahead. In each control cycle, the controller extracts a 3m wide and 1.5m wide section of the road surface ahead, performs ground point extraction and quadratic surface fitting, and obtains the local road surface height function. Subsequently, the system calculates the slope, curvature, and roughness based on the fitted surface, and combines the current vehicle speed and acceleration to obtain the preview position within the predicted time domain in the next 0.5 seconds. Then, it calculates the preview height difference between the left and right wheels and the preview height difference in the front and rear directions, obtaining the predicted roll disturbance and pitch disturbance quantities. Based on the predicted disturbances, the feedforward controller generates compensation control quantities in advance; simultaneously, the attitude outer loop generates the desired angular velocity based on the error between the target attitude and the current attitude, and the angular velocity inner loop generates the feedback control quantity based on the error between the desired angular velocity and the actual angular velocity. Finally, the controller fuses the feedforward compensation quantity and the feedback control quantity and outputs them to the left and right drive wheel torque distribution unit and the active suspension damping adjustment unit to achieve active stabilization control of the vehicle's roll and pitch attitude. Experimental results show that, compared to using only attitude feedback control, this embodiment can significantly reduce the fluctuation amplitude of the vehicle's roll and pitch angles under the same road conditions, shorten the vehicle's attitude recovery time, and improve the vehicle's driving stability, sensor measurement stability, and operational safety on complex unstructured road surfaces.

[0232] In another specific embodiment, the present invention is applied to unmanned agricultural vehicles. A binocular camera and LiDAR are installed at the front of the vehicle, utilizing a visual and point cloud fusion method to improve the recognition accuracy of furrows, field ridges, and undulating terrain. The system employs the same road surface modeling, disturbance prediction, and feedforward-feedback fusion control process as described above. When the vehicle passes through the furrow boundary area, it can identify the road surface elevation change trend in advance and generate pitch compensation control, thereby reducing the vehicle's pitch fluctuations and improving the stable alignment capability of the operating equipment with the target area. This embodiment illustrates that the present invention is not only applicable to inspection scenarios but also to various unmanned platforms such as agricultural machinery and mining transport vehicles.

[0233] Although the present invention has been described above with reference to embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of the invention. In particular, as long as there is no structural conflict, the features in the disclosed embodiments can be combined with each other in any manner. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, the present invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for predicting road disturbances and actively maintaining stability in unmanned vehicles, characterized in that, Includes the following steps: S1: Road environment information and vehicle motion status acquisition: Real-time acquisition of road surface spatial information in front of the vehicle and vehicle motion status information to establish a three-dimensional road surface set; S2: Multi-source information time synchronization and spatial unification: Unified processing of multi-source data collected; S3: Local 3D road surface geometry modeling: Based on the synchronized road surface point cloud data, 3D geometry modeling is performed on the area in front of the vehicle. After obtaining the road surface model, the road surface geometric feature parameters are further calculated, including: road surface slope, road surface curvature, and road surface roughness. S4: Road disturbance prediction modeling: Based on the current motion state of the vehicle and the geometric characteristics of the road surface, establish a disturbance prediction model for the vehicle's dynamic response to estimate the disturbances that the vehicle will experience in the future time window. S5: Disturbance feedforward compensation control quantity calculation: Based on the predicted future disturbance, calculate the disturbance feedforward compensation control quantity so that the control system can pre-adjust before the disturbance takes effect; S6: Attitude Feedback Stability Control Calculation: Based on feedforward compensation, a vehicle attitude feedback control mechanism is established, specifically using a cascade control structure: Attitude Angle Control Loop: Generates the desired angular velocity based on the error between the target attitude and the actual attitude; Angular Velocity Control Loop: Calculates the control input based on the error between the desired angular velocity and the actual angular velocity; S7: Fusion of feedforward and feedback control: The disturbance feedforward compensation control quantity and the attitude feedback control quantity are fused to obtain the final vehicle stability control input; S8: Actuator attitude adjustment: Control and adjust the vehicle actuator according to the balance control input.

2. The method for predicting road disturbances and actively stabilizing control of unmanned vehicles according to claim 1, characterized in that, In S1, the road environment information includes: road surface image information and road surface three-dimensional point cloud information; the vehicle motion state information includes: vehicle attitude angle, vehicle angular velocity, vehicle linear acceleration, and vehicle speed. The method for establishing the three-dimensional road surface set is as follows: Let the vehicle coordinate system be denoted as... ,in The axis is the direction of vehicle movement. The axis is the lateral direction of the vehicle. The axis is the vertical direction of the vehicle, and the set of points on the road surface ahead collected by the lidar is as follows: in, Indicates the number of sampling points. Indicates the first The three-dimensional coordinates of a road surface point in the vehicle coordinate system; The vehicle state vector is defined as: in, This is the roll angle. Yaw angle The roll rate is angular velocity. The pitch angular velocity, Yaw angular velocity, For the longitudinal speed of the vehicle, , , These represent the linear accelerations of the vehicle in the longitudinal, lateral, and vertical directions, respectively.

3. The method for predicting road disturbances and actively maintaining stability for unmanned vehicles according to claim 1, characterized in that, The unified processing in S2 includes: aligning the data from various sensors in time using a time synchronization algorithm; converting the road point cloud information and vehicle motion state information to the vehicle coordinate system or global coordinate system through coordinate calibration and transformation; and filtering and fusing multi-source information using a state fusion algorithm to obtain a unified environmental state vector. The specific method is as follows: Let the lidar timestamp be... IMU timestamp is Wheel speed timestamp is Select a unified control time Then, various types of data are mapped to the same moment through interpolation: in, and These are the sampling times adjacent to the control time, respectively. These are the estimated vehicle states after synchronization. For spatial unification, let the extrinsic transformation matrix from the lidar coordinate system to the vehicle coordinate system be: in, For rotation matrix, If the translation vector is used, then the lidar point After conversion to the vehicle coordinate system, it becomes: Through the above processing, road surface point cloud and vehicle status information under a unified time reference and a unified coordinate system are obtained.

4. The method for predicting road disturbances and actively maintaining stability for unmanned vehicles according to claim 1, characterized in that, The three-dimensional geometric modeling in S3 specifically includes: filtering and removing outliers from point cloud data; identifying road surface areas using a ground point extraction algorithm; and establishing a local road surface height function model using a surface fitting method. The specific method is as follows: Select a length in front of the vehicle... Width is Using the road surface area as the modeling window, a quadratic surface fitting is performed on the ground points within the window to obtain a local road surface height function model: in, Indicates the forward distance of the vehicle. Indicates the lateral distance of the vehicles. Indicates the road surface height at the corresponding point; The surface coefficients to be estimated; The surface coefficients are solved using the least squares method, i.e., minimizing the objective function: Based on the fitting results, calculate the road surface geometric characteristic parameters; The method for calculating the road surface slope characteristic parameters is as follows: the road surface at... direction and The first-order partial derivatives in the directions are as follows: The local slope is defined as follows: in, Indicates the road surface at point The slope amplitude at the location; The method for calculating the road surface curvature characteristic parameters is as follows: To characterize the rate of change of local undulations, a second-order change quantity is defined: in, Represents the longitudinal curvature component. Represents the lateral curvature component. Indicates the coupled curvature components; The method for calculating the road surface roughness parameter is as follows: Let the actual point cloud height be... The height of the fitted surface is The local roughness is then defined as: in, It represents the road surface roughness, reflecting the degree of high-frequency undulations near the fitted plane.

5. The method for predicting road disturbances and actively maintaining stability for unmanned vehicles according to claim 1, characterized in that, The specific method of S4 is as follows: Let the prediction time domain length be... The vehicle in the future The forward distance reached is: in, For future preview location, The current longitudinal velocity, This represents the current longitudinal acceleration; Let the positions of the centers of the left and right wheels of the vehicle in the lateral direction be respectively. and ,in Given the wheelbase, the preview heights for the left and right wheels are as follows: Let the equivalent preview position difference in the front and rear wheelbase directions of the vehicle be... The height difference in the front and rear directions is expressed as: The height difference in the left and right directions is expressed as: Therefore, estimates of pitch and roll disturbances are established: in, To predict pitch disturbance angle, To predict the roll disturbance angle; Further considering the coupling effect of roughness and velocity, a comprehensive disturbance prediction quantity is constructed: in, This is the predicted amount of roll disturbance. This is the pitch disturbance prediction quantity; , , , These are the perturbation mapping coefficients, obtained through experimental calibration or system identification; In actual control, the weighted average value in the prediction time domain is selected as the equivalent disturbance input for the current control cycle: 。 6. The method for predicting road disturbances and actively maintaining stability for unmanned vehicles according to claim 1, characterized in that, The specific method of S5 is as follows: based on the predicted disturbance amount Calculate the disturbance feedforward compensation control quantity, and assume the feedforward gain matrix is: The disturbance feedforward control quantity is: in, These represent the feedforward compensation amounts for the roll and pitch directions, respectively. , For feedforward compensation gain; If we further consider the rate of change of the disturbance, we add a predictive differential compensation term: in, The perturbation rate of change gain matrix, This represents the rate of change of the predicted disturbance.

7. The method for predicting road disturbances and actively maintaining stability for unmanned vehicles according to claim 1, characterized in that, The attitude angle control loop in S6: Let the target roll angle and target pitch angle be respectively , The actual roll angle and pitch angle are respectively , Then the attitude error is: The outer ring generates the desired angular velocity command: in, For the desired roll rate, The desired pitch rate; , , These are the outer ring ratio, integral, and differential coefficients of the roll attitude angle, respectively. , , These are the outer ring proportional, integral, and differential coefficients of the pitch attitude angle, respectively. Angular velocity control loop: Let the actual roll rate and pitch rate be respectively , Then the angular velocity error is: The inner loop control output is: in, , For feedback control; , , These are the inner loop parameters for roll rate; , , These are the inner loop parameters for pitch angular velocity.

8. The method for predicting road disturbances and actively maintaining stability for unmanned vehicles according to claim 1, characterized in that, The specific method of S7 is as follows: feedforward compensation and feedback control are integrated to obtain the final control input. in, To avoid control saturation, an amplitude limit constraint is added: in, Represents the amplitude limiting function. and These represent the minimum and maximum allowed outputs for the corresponding control channels, respectively.

9. The method for predicting road disturbances and actively maintaining stability for unmanned vehicles according to claim 1, characterized in that, The specific method of S8 is to distribute the attitude control quantity as additional torque to the left and right drive wheels when the vehicle adopts a left-right wheel differential drive mode: in, and Additional torque for the left and right wheels respectively. This is the torque distribution coefficient; When a vehicle is equipped with active suspension, pitch and roll control values ​​are also mapped to suspension damping adjustment values: in, , These represent the damping adjustment amounts of the front and rear suspensions, respectively. , These represent the damping adjustment amounts for the left and right suspensions, respectively. , , , For the mapping coefficient of the actuator.

Images