Automatic guided vehicle trajectory tracking control method applied to attachment coefficient change

Abstract

The invention discloses an automatic guided vehicle trajectory tracking control method applied to attachment coefficient change. The method comprises the following steps: acquiring road surface information of a driving road surface of an automatic guided vehicle based on an environment sensor; acquiring vehicle body information of the automatic guided vehicle based on a vehicle body sensing sensor; establishing a whole vehicle dynamic model of the automatic guided vehicle; estimating a road adhesion coefficient according to the road information, the vehicle body information and a vehicle dynamics model to obtain an estimation result; and tracking and controlling the track of the automated guided vehicle based on an improved model predictive control algorithm according to an estimation result. The characteristics of low vehicle tracking control efficiency and low robustness under different road adhesion coefficients are solved, an improved model prediction control method is adopted to enable the automatic guided vehicle to track a target trajectory more flexibly and efficiently, and the influence of complex and changeable environmental conditions on vehicle trajectory tracking is avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of the automobile industry, in particular to a track tracking control method for an unmanned carrier applied to a changing adhesion coefficient. BACKGROUND

[0002] At present, there are many studies on the track tracking control of unmanned carriers. The existing technology discloses a road surface adaptive MPC track tracking control for unmanned vehicles. A discrete linear time-varying prediction model is obtained from a 3-degree-of-freedom nonlinear dynamics model of a vehicle. The road adhesion coefficient detected by a sensor is used to realize the road surface adaptive vehicle speed range matching. A vehicle speed increment constraint condition is formulated. The driving safety is evaluated in combination with the tire side slip mechanism. The track tracking stability zone and the instability zone are divided in all working conditions. The tracking accuracy of the vehicle is improved. However, the real-time performance of the tracking cannot be guaranteed. When facing complex and variable environmental conditions, the vehicle is easily disturbed, and the high-precision control requirement cannot be met. After the track tracking of the vehicle is realized, the obtained track is not verified, the track tracking effect cannot be verified, and the track tracking accuracy is low. SUMMARY

[0003] The present application aims to at least partly solve one of the above technical problems. To this end, the present application aims to propose a track tracking control method for an unmanned carrier applied to a changing adhesion coefficient, solve the low vehicle tracking control efficiency and low robustness under different road adhesion coefficients, and use an improved model predictive control method to make the unmanned carrier more flexible and efficient in tracking the target track, and avoid the influence of complex and variable environmental conditions on the vehicle track tracking. Through the improved model predictive control algorithm, the track tracking accuracy and real-time performance of the unmanned carrier are improved, and efficient control under the variable adhesion coefficient working condition is realized. At the same time, the state observer designed by the adaptive singular value decomposition unscented Kalman filter algorithm with noise estimator can also be dynamically adjusted according to the actual situation, adapt to the needs of different working scenes, and improve the information acquisition ability of the unmanned carrier.

[0004] To achieve the above-mentioned purpose, the embodiment of the present application proposes a track tracking control method for an unmanned carrier applied to a changing adhesion coefficient, comprising:

[0005] Obtaining road surface information of a driving road surface of the unmanned carrier based on an environmental sensor;

[0006] Obtaining vehicle body information of the unmanned carrier based on a vehicle body perception sensor;

[0007] Establishing a whole vehicle dynamics model of the unmanned carrier;

[0008] Estimating the road adhesion coefficient according to the road surface information, the vehicle body information and the whole vehicle dynamics model, and obtaining an estimation result;

[0009] According to the estimation result, track control of the unmanned carrier is performed based on an improved model predictive control algorithm.

[0010] According to some embodiments of the present application, road surface information of a road surface on which the unmanned carrier travels is acquired based on an environment sensor, including:

[0011] Laser radar included in the environment sensor scans the surrounding environment through laser beams to determine laser point cloud data;

[0012] A camera included in the environment sensor acquires road surface images;

[0013] A temperature and humidity sensor included in the environment sensor acquires temperature and humidity information of the road surface on which the unmanned carrier is located;

[0014] The road surface information is determined according to the laser point cloud data, the road surface images and the temperature and humidity information.

[0015] According to some embodiments of the present application, vehicle body information of the unmanned carrier is acquired based on a vehicle body perception sensor, including:

[0016] An accelerometer included in the vehicle body perception sensor acquires acceleration of the unmanned carrier;

[0017] A gyroscope included in the vehicle body perception sensor acquires angular velocity of the unmanned carrier;

[0018] A wheel speed sensor included in the vehicle body perception sensor acquires wheel speed information of wheels of the unmanned carrier;

[0019] A vibration sensor included in the vehicle body sensor acquires vibration information of the unmanned carrier;

[0020] A pressure sensor included in the vehicle body sensor acquires pressure information of the unmanned carrier on the road surface;

[0021] A height sensor included in the vehicle body sensor measures a distance between the vehicle body and the road surface to evaluate a vehicle body height;

[0022] The vehicle body information of the unmanned carrier is determined according to the acceleration, the angular velocity, the wheel speed information, the vibration information, the pressure information and the vehicle body height.

[0023] According to some embodiments of the present application, a whole vehicle dynamics model of the unmanned carrier is established, including:

[0024] Vehicle information of the unmanned carrier is acquired; the vehicle information includes total mass, center of mass position, wheelbase, track, tire size and type, and suspension system type of the unmanned carrier;

[0025] Define a global coordinate system and a vehicle coordinate system; the global coordinate system is used to describe the position and motion state of the vehicle in the motion range; the vehicle coordinate system is used to describe the dynamic behavior of the vehicle in the motion process;

[0026] Determine the kinematics of the vehicle in the plane, including the relationship between the linear velocity, angular velocity and acceleration of the vehicle, and the influence of the steering and suspension system of the vehicle on the kinematics;

[0027] According to the kinematics of the vehicle in the plane, the whole vehicle dynamics model of the vehicle is established.

[0028] According to some embodiments of the present application, the road adhesion coefficient is estimated according to the road surface information, the vehicle body information and the whole vehicle dynamics model, and the estimation result is obtained, including:

[0029] According to the road surface information and the vehicle body information, the vehicle motion state is determined;

[0030] Based on the whole vehicle dynamics model, the road adhesion coefficient is regarded as a state variable, and the correlation between the vehicle motion state and the state variable is established;

[0031] According to the correlation, the state observer is designed by using the adaptive singular value decomposition unscented Kalman filter algorithm with noise estimator to estimate the road adhesion coefficient, and the estimation result is obtained.

[0032] According to some embodiments of the present application, according to the correlation, the state observer is designed by using the adaptive singular value decomposition unscented Kalman filter algorithm with noise estimator to estimate the road adhesion coefficient, and the estimation result is obtained, including:

[0033] Initialization: set the initial parameters of the adaptive singular value decomposition unscented Kalman filter algorithm, including the state vector, the covariance matrix, and the noise statistical characteristics;

[0034] Prediction step: predict the vehicle motion state and the road adhesion coefficient at the next time based on the correlation by using the whole vehicle dynamics model;

[0035] Update step: update the predicted state by using the vehicle motion state information measured by the sensor and the ASVD-UKF algorithm;

[0036] Adaptive singular value decomposition: in the update step, the covariance matrix is processed by using the singular value decomposition technology;

[0037] Noise estimator: estimate the statistical characteristics of the process noise and the measurement noise in real time according to the observation data;

[0038] Iteration and convergence: repeatedly perform the prediction and update steps until the algorithm converges or reaches a preset number of iterations, and during the convergence process, the estimated value of the road adhesion coefficient will gradually approach the true value.

[0039] According to some embodiments of the application, the improved model predictive control algorithm is used for tracking control of the trajectory of the unmanned carrier vehicle based on the estimation result, which includes:

[0040] Determining the control quantity and the constraint condition of the vehicle according to the estimation result;

[0041] The gradient projection algorithm is introduced to improve the model predictive control algorithm, the projection direction is taken as the steepest descent direction of the objective function, and the constraint condition and the relaxation factor are added to increase the tracking precision and prevent falling into a local optimal solution, so as to obtain the improved model predictive control algorithm;

[0042] The improved model predictive control algorithm is used for tracking control of the trajectory of the unmanned carrier vehicle.

[0043] According to some embodiments of the application, the trajectory information obtained by the tracking control is verified, and an alarm prompt is issued when it is determined that the verification fails.

[0044] According to some embodiments of the application, the trajectory information obtained by the tracking control is verified, which includes:

[0045] In the process of tracking control of the trajectory of the unmanned carrier vehicle, two frames of scene images including the unmanned carrier vehicle are obtained;

[0046] Feature extraction is performed on the two frames of scene images, a first color histogram corresponding to the unmanned carrier vehicle in the first frame of scene image is extracted, and the proportion coefficient of the pixel point with a pixel value of λ in the first color histogram a ε is calculated; a second color histogram corresponding to the unmanned carrier vehicle in the second frame of scene image is extracted, and the proportion coefficient of the pixel point with a pixel value of λ in the second color histogram b ε is calculated. ε λ ;

[0047]

[0048] wherein a ε λ is the proportion coefficient of the pixel point with a pixel value of λ in the first color histogram a ε ; d(w j , p) is the distance between the jth pixel point in the first color histogram and the center pixel point p in the first color histogram; d0 is the maximum value of d(w j , p); w j is the jth pixel point in the first color histogram; k(w jis a pixel value of a jth pixel point in the first color histogram; is a weight coefficient of the jth pixel point in the first color histogram; is a weight coefficient of the jth pixel point in the first color histogram; a is a number of pixel points included in the first color histogram; T is a number of pixel points with a pixel value of λ in the first color histogram;

[0049] According to a proportion coefficient of pixel points with a pixel value of λ in the first color histogram a ε and a proportion coefficient of pixel points with a pixel value of λ in the second color histogram b ε , the similarity sim(a, b) between the first frame scene image a and the second frame scene image b is calculated. ε λ

[0050]

[0051] wherein, d a is a moving length of the AGV in the first frame scene image; d b is a moving length of the AGV in the second frame scene image; d(a0, b0) is a distance between a midpoint a0 of the AGV in the first frame scene image and a midpoint b0 of the AGV in the second frame scene image; A1 is a first weight related to the moving length of the AGV; M a is a number of pixel points included in the first frame scene image; M b is a number of pixel points included in the second frame scene image; A2 is a second weight related to the number of pixel points included in the AGV; A3 is a third weight related to the color histogram corresponding to the AGV; A1+A2+A3=1; C is a number of pixel values included in the first color histogram, and is also a number of pixel values included in the second color histogram, and the number of pixel values included in the first color histogram is equal to the number of pixel values included in the second color histogram;

[0052] The similarity between the first frame scene image a and the second frame scene image b determines a moving track of the AGV, and the track information obtained by the tracking control is verified according to the moving track.

[0053] ​The application provides a track tracking control method for an unmanned carrier applied to a changing adhesion coefficient, solves the characteristics of low tracking control efficiency and low robustness of the vehicle under different road adhesion coefficients, and makes the unmanned carrier more flexible and efficient in tracking a target track by using an improved model prediction control method, so that the influence of complex and changeable environmental conditions on vehicle track tracking is avoided.

[0054] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent to those skilled in the art from the description, or can be learned by practice of the present application. The objects and other advantages of the present application can be realized and attained by the structure particularly pointed out in the written description and the accompanying drawings.

[0055] The technical solutions of the present application will be further described in detail below with the help of the accompanying drawings and examples. BRIEF DESCRIPTION OF DRAWINGS

[0056] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the specification, and are used to explain the present application together with the embodiments of the present application, and do not constitute a limitation on the present application. In the drawings:

[0057] Figure 1 is a flowchart of a track tracking control method for an unmanned carrier applied to a changing adhesion coefficient according to an embodiment of the present application. DETAILED DESCRIPTION

[0058] The preferred embodiments of the present application will be described below with reference to the accompanying drawings, and it should be understood that the preferred embodiments described herein are only used to illustrate and explain the present application, and do not limit the present application.

[0059] As shown in Figure 1 , the present application provides a track tracking control method for an unmanned carrier applied to a changing adhesion coefficient, comprising steps S1-S5:

[0060] S1, obtaining road surface information of a driving road surface of the unmanned carrier based on an environmental sensor;

[0061] S2, obtaining vehicle body information of the unmanned carrier based on a vehicle body perception sensor;

[0062] S3, establishing a whole vehicle dynamics model of the unmanned carrier;

[0063] S4, estimate the road adhesion coefficient based on the road information, vehicle body information and vehicle dynamics model, and obtain an estimation result;

[0064] S5, track control the trajectory of the unmanned carrier vehicle based on the improved model predictive control algorithm according to the estimation result.

[0065] The working principle of the above technical solution is that the environmental sensors such as cameras and laser radars installed on the unmanned carrier vehicle are used to collect road information. These information includes the material, humidity, roughness and slope of the road, which will directly affect the driving stability and adhesion coefficient of the unmanned carrier vehicle. The unmanned carrier vehicle obtains its state information such as acceleration, speed and attitude angle through vehicle perception sensors such as acceleration sensors, gyroscopes and speed sensors. Based on the collected road information and vehicle body information, a vehicle dynamics model describing the dynamic behavior of the unmanned carrier vehicle can be established. This model will include the mass, inertia, tire characteristics and road adhesion coefficient of the vehicle, which will be used for subsequent road adhesion coefficient estimation and trajectory tracking control. Model predictive control is an advanced control method that can calculate the optimal control strategy by considering the dynamic behavior of the system in the future. By introducing the estimation result of the road adhesion coefficient, the accuracy and stability of trajectory tracking can be further improved.

[0066] The beneficial effects of the above technical solution are that the low efficiency and low robustness of vehicle tracking control under different road adhesion coefficients are solved, and the improved model predictive control method makes the unmanned carrier vehicle more flexible and efficient in tracking the target trajectory, avoiding the influence of complex and variable environmental conditions on vehicle trajectory tracking. Through the improved model predictive control algorithm, the trajectory tracking accuracy and real-time performance of the unmanned carrier vehicle are improved, and efficient control under variable adhesion coefficient conditions is realized. At the same time, the state observer designed by the adaptive singular value decomposition unscented Kalman filter algorithm with noise estimator can also be dynamically adjusted according to the actual situation to adapt to the needs of different working scenarios, and the information acquisition ability of the unmanned carrier vehicle is improved.

[0067] According to some embodiments of the present application, the road information of the road on which the unmanned carrier vehicle travels is obtained based on the environmental sensors, including:

[0068] The laser radar included in the environmental sensor scans the surrounding environment through laser beams to determine the laser point cloud data;

[0069] The camera included in the environmental sensor obtains the road image;

[0070] The temperature and humidity sensor included in the environmental sensor obtains the temperature and humidity information of the road on which the unmanned carrier vehicle is located;

[0071] Determine the road surface information according to the laser point cloud data, the road surface image and the temperature and humidity information.

[0072] The working principle of the above technical solution: based on the laser point cloud data, the three-dimensional shape of the road surface, the obstacle distribution, the slope and the possible potholes or protrusions and other key information can be obtained. Through the camera, the unmanned vehicle can obtain the texture, color, brightness and traffic signs of the road surface. These information is crucial for determining the material of the road surface (such as asphalt, concrete or sandstone, etc.), the degree of dryness and wetness, and whether there is oil or ice and snow and other potential risks. The temperature and humidity sensor can directly measure the temperature and humidity information of the road surface where the unmanned vehicle is located. These information is convenient for judging the physical state of the road surface (such as whether it is icy, whether it is too slippery, etc.) and predicting future changes in the road surface (such as whether it is likely to have water accumulation or frost, etc.).

[0073] The beneficial effects of the above technical solution: the unmanned vehicle integrates multiple environmental sensors and uses advanced algorithms for information fusion and processing, which can realize comprehensive perception and accurate judgment of the road surface information, so as to ensure its safe driving under various complex road conditions.

[0074] According to some embodiments of the application, the vehicle body information of the unmanned vehicle is obtained based on the vehicle body perception sensor, including:

[0075] The acceleration of the unmanned vehicle is obtained based on the accelerometer included in the vehicle body perception sensor;

[0076] The angular velocity of the unmanned vehicle is obtained based on the gyroscope included in the vehicle body perception sensor;

[0077] The wheel speed information of the wheels of the unmanned vehicle is obtained based on the wheel speed sensor included in the vehicle body perception sensor;

[0078] The vibration information of the unmanned vehicle is obtained based on the vibration sensor included in the vehicle body sensor;

[0079] The pressure information of the unmanned vehicle to the road surface is obtained based on the pressure sensor included in the vehicle body sensor;

[0080] The height of the vehicle body is evaluated based on the height sensor included in the vehicle body sensor to measure the distance between the vehicle body and the road surface;

[0081] The vehicle body information of the unmanned vehicle is determined based on the acceleration, angular velocity, wheel speed information, vibration information, pressure information and vehicle body height.

[0082] The working principle of the above technical solution is as follows: The accelerometer is usually used to monitor the longitudinal, lateral and vertical acceleration of the vehicle, and these acceleration information is crucial for judging the dynamic behavior of the vehicle such as acceleration, deceleration, turning and jolt. The gyroscope is mainly used to monitor the yaw angular velocity, pitch angular velocity and roll angular velocity of the vehicle, and these information is very important for judging the attitude stability, steering performance and possible rollover of the vehicle. The data of the gyroscope can also be used for vehicle attitude estimation and navigation positioning. The wheel speed sensor is usually installed on each wheel to monitor the wheel speed and vehicle speed in real time. These information is very important for the brake control, anti-skid control and traction control of the vehicle. In addition, by analyzing the wheel speed difference, the steering angle and driving trajectory of the vehicle can be judged. The vibration sensor is used to monitor the vibration of the unmanned carrier during operation. These vibrations may come from uneven road surface, loose or damaged vehicle parts, etc. The pressure sensor is used to measure the pressure distribution of the unmanned carrier on the road surface. These pressure information is convenient for judging the load condition of the vehicle, the degree of tire wear and the road adhesion condition. The height sensor is used to measure the distance between the vehicle body and the road surface, so as to evaluate the body height of the vehicle.

[0083] The beneficial effects of the above technical solution are as follows: The unmanned carrier can comprehensively process the information from the accelerometer, gyroscope, wheel speed sensor, vibration sensor, pressure sensor and height sensor, and fuse and analyze them through algorithm, so as to obtain comprehensive and accurate vehicle body information. It can realize comprehensive perception and accurate judgment of the vehicle body state, so as to ensure its safe, stable and efficient operation under various complex conditions.

[0084] According to some embodiments of the present application, a whole vehicle dynamics model of the unmanned carrier is established, comprising:

[0085] Obtaining vehicle information of the unmanned carrier; the vehicle information includes total mass, mass center position, wheelbase, track, tire size and type, and suspension system type of the unmanned carrier;

[0086] Defining a global coordinate system and a vehicle coordinate system; the global coordinate system is used to describe the position and motion state of the unmanned carrier in the motion range; the vehicle coordinate system is used to describe the dynamic behavior of the unmanned carrier in the motion process;

[0087] Determining the kinematics relationship of the unmanned carrier in the plane, including the relationship between the linear velocity, angular velocity and acceleration of the vehicle, and the influence of the steering and suspension system of the vehicle on the kinematics relationship;

[0088] According to the kinematics relationship of the vehicle in the plane, the whole vehicle dynamics model of the unmanned carrier is established.

[0089] The working principle and beneficial effects of the above technical solution are as follows: the parameters of the unmanned forklift are collected, including but not limited to the total mass of the vehicle, the center of mass position, the wheelbase, the track, the size and type of the tire, and the type of the suspension system. These information constitutes the basis for building the dynamic model, which will directly affect the dynamic behavior of the vehicle. Global coordinate system: a fixed reference frame used to describe the absolute position and velocity of the unmanned forklift in the entire range of motion. It provides a unified reference to accurately track the motion trajectory of the vehicle. Vehicle coordinate system: a frame of reference with the vehicle itself as the reference, which moves with the vehicle. In this coordinate system, the dynamic behavior of the vehicle such as steering, acceleration, etc. can be more conveniently described. In the plane, the kinematic relationship of the unmanned forklift mainly involves the interaction between its linear velocity, angular velocity, acceleration, and the influence of the steering and suspension system on these relationships. Start building the whole vehicle dynamics model of the unmanned forklift. This model will be based on Newton's second law (F=ma) and the principle of conservation of momentum, integrating the parameters and kinematic relationships of the vehicle together. Consider the longitudinal, lateral and vertical motion of the vehicle, as well as their interactions. This includes vehicle acceleration, deceleration, steering, suspension system dynamic response, etc. At the same time, the friction and adhesion between the tire and the road surface, and their influence on the dynamic behavior of the vehicle, also need to be considered. Get a model containing a system of multiple differential equations, which describes the dynamic behavior of the unmanned forklift under various working conditions. This model can be used to predict the vehicle's driving trajectory, stability, handling, and other key performance indicators, and provide a basis for subsequent trajectory tracking control.

[0090] According to some embodiments of the present application, the road adhesion coefficient is estimated according to the road surface information, the vehicle body information and the whole vehicle dynamics model, and an estimation result is obtained, including:

[0091] Determine the vehicle motion state according to the road surface information and the vehicle body information;

[0092] Based on the whole vehicle dynamics model, the road adhesion coefficient is regarded as a state variable, and a correlation between the vehicle motion state and the state variable is established;

[0093] According to the correlation, a state observer is designed using an adaptive singular value decomposition unscented Kalman filter algorithm with noise estimator to estimate the road adhesion coefficient, and an estimation result is obtained.

[0094] The working principle and beneficial effects of the above technical solution are: the road surface information and the vehicle body information, such as the material, humidity and temperature of the road surface, and the acceleration, angular velocity and wheel speed of the vehicle, are integrated to determine the current accurate motion state of the unmanned carrier. With the help of a carefully constructed vehicle dynamics model, the road adhesion coefficient is regarded as a state to be solved. This model will describe the complex relationship between the vehicle motion state and the road adhesion coefficient based on the physical characteristics and motion laws of the vehicle. An advanced state observer is designed by using the adaptive singular value decomposition unscented Kalman filter (SVD-UKF) algorithm with noise estimator. This algorithm combines the efficiency of unscented Kalman filter (UKF) and the stability of singular value decomposition (SVD), so that it can maintain high estimation accuracy in a noisy environment. By using the road surface information, vehicle body information and vehicle dynamics model, combined with the advanced SVD-UKF algorithm, the road adhesion coefficient can be accurately estimated, and the control performance and driving safety of the unmanned carrier can be improved.

[0095] According to some embodiments of the application, the adaptive singular value decomposition unscented Kalman filter algorithm with noise estimator is used to design a state observer to estimate the road adhesion coefficient according to the correlation relationship, and an estimation result is obtained, including:

[0096] Initialization: set the initial parameters of the adaptive singular value decomposition unscented Kalman filter algorithm, including the state vector, covariance matrix and noise statistical characteristics;

[0097] Prediction step: predict the vehicle motion state and road adhesion coefficient at the next time based on the correlation relationship by using the vehicle dynamics model;

[0098] Update step: update the predicted state by using the vehicle motion state information measured by the sensor and the ASVD-UKF algorithm;

[0099] Adaptive singular value decomposition: in the update step, the covariance matrix is processed by using the singular value decomposition technology;

[0100] Noise estimator: estimate the statistical characteristics of the process noise and measurement noise in real time according to the observation data;

[0101] Iteration and convergence: repeat the prediction and update steps until the algorithm converges or reaches the preset number of iterations. During the convergence process, the estimated value of the road adhesion coefficient will gradually approach the true value.

[0102] The working principle and beneficial effects of the above technical solution are as follows: At the beginning of the algorithm, a series of initial parameters are set for the ASVD-UKF algorithm. These parameters include: state vector: it covers all the states we need to estimate, such as the speed, acceleration, attitude angle of the vehicle, and road adhesion coefficient, etc. Covariance matrix: it describes the uncertainty or correlation between the elements in the state vector. Noise statistical characteristics: this includes the mean, variance and other statistical information of the process noise and measurement noise, which will be used in the subsequent filtering and updating steps. Using the known vehicle dynamics model and the relationship between the vehicle and the road, the vehicle motion state and road adhesion coefficient at the next time are predicted. This prediction will be based on the current state vector and covariance matrix, as well as the physical laws in the dynamics model. When new sensor data arrives, use these data to update the predicted state. The ASVD-UKF algorithm will play a key role here, combining the efficiency of Unscented Kalman Filter (UKF) and the stability of Singular Value Decomposition (SVD) to more accurately handle nonlinear systems and noise. Use sensor data: obtain real-time motion state information of the vehicle from sensors, such as acceleration, angular velocity, etc. ASVD-UKF update: use these data and ASVD-UKF algorithm to update the predicted state, to get more accurate state estimation value. In the updating step, in order to maintain the numerical stability and computational efficiency of the covariance matrix, the Singular Value Decomposition (SVD) technique is used to process it. SVD can decompose the covariance matrix into a linear combination of a series of orthogonal vectors, making it easier to handle its inverse matrix and square root operations. Since the statistical characteristics of noise in the actual environment are often unknown or time-varying, a noise estimator is designed to estimate the statistical characteristics of the process noise and measurement noise in real time. This estimation will be based on the observation data and the filtering results of the ASVD-UKF algorithm, so as to ensure that the algorithm can adapt to different noise environments. Repeat the prediction and updating steps until the algorithm converges or reaches the preset number of iterations. During the convergence process, the estimated value of the road adhesion coefficient will gradually approach its true value. Observe the stability and accuracy of the estimated value to determine whether the algorithm has converged. The observer combines advanced filtering technology and noise processing technology, and can provide accurate road information for the autonomous vehicle in complex driving environments.

[0103] According to some embodiments of the present application, the trajectory tracking control of the autonomous vehicle based on the improved model predictive control algorithm according to the estimation result comprises:

[0104] determining the control quantity and the constraint condition of the vehicle according to the estimation result;

[0105] The gradient projection algorithm is introduced to improve the model predictive control algorithm, the projection direction is taken as the steepest descent direction of the objective function, and a constraint condition and a relaxation factor are added to increase tracking accuracy and prevent falling into a local optimal solution, so that the improved model predictive control algorithm is obtained.

[0106] The improved model predictive control algorithm is used for trajectory tracking control of the unmanned forklift.

[0107] The working principle and beneficial effects of the above technical solution are as follows: the control amount and the constraint condition of the vehicle are determined according to the estimation result; the control amount includes the acceleration and the steering angle of the vehicle, which will directly affect the motion trajectory of the vehicle. The constraint condition considers the physical limitations (such as maximum acceleration and maximum steering angle) of the vehicle and the road conditions (such as curvature and slope), so as to ensure that the control instruction is within a feasible range. The gradient projection algorithm is introduced. This algorithm takes the projection direction as the steepest descent direction of the objective function, thereby accelerating the optimization process. At the same time, a constraint condition and a relaxation factor are added to enhance the tracking accuracy and prevent the algorithm from falling into a local optimal solution. By calculating the gradient of the objective function with respect to the control amount and projecting along the gradient direction, we can find the control amount update direction that makes the objective function value descend the fastest. In the optimization process, it will be ensured that the control amount always meets the preset constraint condition, so as to ensure the feasibility of the control instruction. In order to balance the tracking accuracy and the stability of the algorithm, the relaxation factor is introduced to allow a certain control error. This helps to prevent the algorithm from falling into a local optimal solution under strict constraints. Using the current state of the vehicle and the dynamic model, the motion trajectory of the vehicle in the future period of time can be predicted. Rolling optimization: in each control period, the improved MPC algorithm is used to solve the optimal control sequence according to the prediction result and the constraint condition. Only the first control instruction of the sequence is executed, and this process is repeated in the next control period. Feedback correction: by comparing the difference between the actual motion trajectory and the predicted trajectory, the prediction model is corrected to improve the accuracy of subsequent control. The trajectory tracking control method based on the estimation result and the improved MPC algorithm can realize accurate control of the motion trajectory of the unmanned forklift. This method combines the advantages of the gradient projection algorithm, improves the performance of the MPC algorithm, and ensures the feasibility and stability of the control instruction.

[0108] According to some embodiments of the present application, the trajectory information obtained by the tracking control is verified, and an alarm prompt is issued when it is determined that the verification fails.

[0109] The working principle and beneficial effects of the above technical solution are as follows: the trajectory information obtained by the tracking control is verified, and the effect of trajectory tracking is judged, and an alarm prompt is issued when it is determined that the verification fails, so as to timely adjust the control strategy of trajectory tracking and improve the accuracy of trajectory tracking control.

[0110] According to some embodiments of the present application, the trajectory information obtained by tracking control is verified, including:

[0111] In the process of tracking control on the trajectory of the unmanned carrier, two frames of scene images including the unmanned carrier are obtained;

[0112] Feature extraction is performed on the two frames of scene images, a first color histogram corresponding to the unmanned carrier in the first frame of scene image is extracted, and a proportion coefficient b of pixel points with a pixel value of λ in the first color histogram a is calculated. ε A second color histogram corresponding to the unmanned carrier in the second frame of scene image is extracted, and a proportion coefficient b of pixel points with a pixel value of λ in the second color histogram b is calculated. ε ε λ ;

[0113]

[0114] wherein, a ε λ is the proportion coefficient of pixel points with a pixel value of λ in the first color histogram a ε ; d(w j , p) is the distance between the jth pixel point in the first color histogram and the center pixel point p in the first color histogram; d0 is the maximum value of d(w j , p); w j is the jth pixel point in the first color histogram; k(w j ) is the pixel value of the jth pixel point in the first color histogram; is the weight coefficient of ; M a is the number of pixel points included in the unmanned carrier in the first color histogram; T is the number of pixel points with a pixel value of λ in the first color histogram;

[0115] According to the proportion coefficient of pixel points with a pixel value of λ in the first color histogram a ε and the proportion coefficient b of pixel points with a pixel value of λ in the second color histogram b ε , the similarity sim(a, b) between the first frame of scene image a and the second frame of scene image b is calculated. ε λ

[0116]

[0117] wherein, d a is the moving length of the unmanned carrier in the first frame of scene image; d b ​​is the moving length of the second frame scene image; d(a0, b0) is the distance between the midpoint a0 of the first frame scene image and the midpoint b0 of the second frame scene image; A1 is the first weight related to the moving length of the second frame scene image; M a is the number of pixel points included in the first frame scene image; M b is the number of pixel points included in the second frame scene image; A2 is the second weight related to the number of pixel points included in the second frame scene image; A3 is the third weight related to the color histogram corresponding to the second frame scene image; A1+A2+A3=1; C is the number of pixel values included in the first color histogram, and is also the number of pixel values included in the second color histogram, and the number of pixel values included in the first color histogram is equal to the number of pixel values included in the second color histogram;

[0118] The similarity between the first frame scene image a and the second frame scene image b determines the moving trajectory of the second frame scene image, and the trajectory information obtained by the tracking control is verified according to the moving trajectory.

[0119] The working principle of the above technical solution is that the features of the second frame scene image are extracted, mainly focusing on color features. For the first frame scene image, the first color histogram corresponding to the second frame scene image is extracted, and the proportion coefficient of the pixel point with the pixel value λ is calculated. The proportion coefficient is calculated by the formula, which considers the distance between the pixel point and the center pixel point, the weight of the pixel value and other factors. For the second frame scene image, the second color histogram corresponding to the second frame scene image is also extracted, and the proportion coefficient of the pixel point with the pixel value λ is calculated. The information of the first color histogram and the second color histogram is used to calculate the similarity between the two frames of scene images, and the similarity calculation formula considers the moving length of the second frame scene image, the number of pixel points contained and the similarity of the color histogram. According to the calculated similarity between the two frames of scene images, we can determine the moving trajectory of the second frame scene image. The moving trajectory is compared with the trajectory information, and when the difference is less than the preset difference threshold, it means that the verification is passed; otherwise, it means that the verification is not passed.

[0120] The beneficial effects of the above technical solution are that the verification method can effectively verify the trajectory information obtained by the tracking control of the second frame scene image through steps such as feature extraction, color histogram calculation and similarity calculation, and ensure the accuracy and reliability of the trajectory.

[0121] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application also intends to include these modifications and variations.

Claims

1. A trajectory tracking control method for a driverless vehicle applied to a change in a coefficient of adhesion, characterized by, The application relates to a method for estimating road adhesion coefficient of an unmanned vehicle, and belongs to the field of intelligent transportation. The method comprises the following steps: acquiring road surface information of a driving road surface of the unmanned vehicle based on an environment sensor; acquiring vehicle body information of the unmanned vehicle based on a vehicle body perception sensor; establishing a whole-vehicle dynamics model of the unmanned vehicle; estimating the road adhesion coefficient based on the road surface information, the vehicle body information and the whole-vehicle dynamics model to obtain an estimation result; 2. The track following control method for a palletless vehicle applied to a change in a coefficient of adhesion according to claim 1, wherein tracking and controlling a trajectory of the unmanned vehicle based on an improved model predictive control algorithm according to the estimation result. The method for acquiring the road surface information of the driving road surface of the unmanned vehicle based on the environment sensor comprises the following steps: determining laser point cloud data by scanning the surrounding environment through a laser beam based on a laser radar included in the environment sensor; acquiring road surface images based on a camera included in the environment sensor; acquiring temperature and humidity information of the road surface where the unmanned vehicle is located based on a temperature and humidity sensor included in the environment sensor; 3. The track following control method for a palletless vehicle applied to a change in a coefficient of adhesion according to claim 1, wherein determining the road surface information based on the laser point cloud data, the road surface images and the temperature and humidity information. The method for acquiring the vehicle body information of the unmanned vehicle based on the vehicle body perception sensor comprises the following steps: acquiring acceleration of the unmanned vehicle based on an accelerometer included in the vehicle body perception sensor; acquiring angular velocity of the unmanned vehicle based on a gyroscope included in the vehicle body perception sensor; acquiring wheel speed information of a wheel of the unmanned vehicle based on a wheel speed sensor included in the vehicle body perception sensor; acquiring vibration information of the unmanned vehicle based on a vibration sensor included in the vehicle body sensor; acquiring pressure information of the unmanned vehicle on the road surface based on a pressure sensor included in the vehicle body sensor; evaluating the height of the vehicle body based on a height sensor included in the vehicle body sensor by measuring the distance between the vehicle body and the road surface; 4. The track following control method for a palletless vehicle applied to a change in a coefficient of adhesion according to claim 1, wherein determining the vehicle body information of the unmanned vehicle based on the acceleration, the angular velocity, the wheel speed information, the vibration information, the pressure information and the height of the vehicle body. The method for establishing the whole-vehicle dynamics model of the unmanned vehicle comprises the following steps: acquiring vehicle information of the unmanned vehicle; the vehicle information comprises the total mass, the center of mass position, the wheelbase, the track, the tire size and type and the suspension system type of the unmanned vehicle; defining a global coordinate system and a vehicle coordinate system; the global coordinate system is used for describing the position and motion state of the unmanned vehicle in a motion range; the vehicle coordinate system is used for describing the dynamic behavior of the unmanned vehicle in the motion process of the unmanned vehicle itself; determining the kinematic relationship of the unmanned vehicle in a plane, including the relationship among the linear velocity, the angular velocity and the acceleration of the vehicle, and the influence of the steering and the suspension system of the vehicle on the kinematic relationship; 5. The track following control method for a palletless vehicle applied to a change in a coefficient of adhesion according to claim 1, wherein establishing the whole-vehicle dynamics model of the unmanned vehicle according to the kinematic relationship of the vehicle in the plane. The method for estimating the road adhesion coefficient based on the road surface information, the vehicle body information and the whole-vehicle dynamics model to obtain the estimation result comprises the following steps: determining the motion state of the vehicle according to the road surface information and the vehicle body information; regarding the road adhesion coefficient as a state variable based on the whole-vehicle dynamics model, establishing the correlation between the motion state of the vehicle and the state variable; 6. The track following control method for a palletless vehicle applied to a change in a coefficient of adhesion according to claim 5, wherein designing a state observer based on an adaptive singular value decomposition unscented Kalman filter algorithm with a noise estimator according to the correlation to estimate the road adhesion coefficient to obtain the estimation result. The method for estimating the road adhesion coefficient based on the correlation according to the adaptive singular value decomposition unscented Kalman filter algorithm with the noise estimator to design the state observer to obtain the estimation result comprises the following steps: Initialization: Set the initial parameters of the adaptive singular value decomposition unscented Kalman filter algorithm, including the state vector, covariance matrix, and noise statistics; Prediction step: Use the vehicle dynamics model to predict the vehicle motion state and road adhesion coefficient at the next time step based on the correlation; Update step: Update the predicted state using the sensor-measured vehicle motion state information and the ASVD-UKF algorithm; Adaptive singular value decomposition: In the update step, use the singular value decomposition technique to process the covariance matrix; Noise estimator: Estimate the statistical properties of the process noise and measurement noise in real time based on the observation data; Iteration and convergence: Repeat the prediction and update steps until the algorithm converges or reaches the preset number of iterations. During the convergence process, the estimated value of the road adhesion coefficient will gradually approach the true value.

7. The track following control method for a palletless vehicle applied to a change in a coefficient of adhesion according to claim 1, wherein Based on the estimation results, use the improved model predictive control algorithm to track the trajectory of the unmanned vehicle, including: Determine the control amount and constraint conditions of the vehicle based on the estimation results; Introduce the gradient projection algorithm to improve the model predictive control algorithm, taking the projection direction as the steepest descent direction of the objective function, and adding constraint conditions and relaxation factors to increase tracking accuracy and prevent falling into local optimal solutions, obtaining the improved model predictive control algorithm; Track the trajectory of the unmanned vehicle based on the improved model predictive control algorithm.

8. The track following control method for a palletless vehicle applied to a change in a coefficient of adhesion according to claim 1, wherein Verify the trajectory information obtained by tracking control. When it is determined that it does not pass the verification, an alarm prompt is issued.

9. The track following control method for a palletless vehicle applied to a change in a coefficient of adhesion according to claim 8, wherein Verify the trajectory information obtained by tracking control, including: During the tracking control of the unmanned vehicle trajectory, two frames of scene images including the unmanned vehicle are obtained; extracting a first color histogram corresponding to the automated guided vehicle in the first frame of the scene image, and calculating the first color histogram a ε the proportion coefficient of the pixel point with the pixel value of λ; extracting a second color histogram corresponding to the automated guided vehicle in the second frame of the scene image, and calculating the second color histogram b ε the proportion coefficient of the pixel point with the pixel value of λ ε λ ; wherein a ε λ is the first color histogram a ε is the proportion coefficient of the pixel point with the pixel value λ in the first color histogram; d(w j , p) is the distance between the jth pixel point in the first color histogram and the center pixel point p in the first color histogram; d0 is the maximum value of d(w j , p); w j is the jth pixel point in the first color histogram; k(w j ) is the pixel value of the jth pixel point in the first color histogram; is the weight coefficient of ; M a is the number of pixel points included in the automated guided vehicle in the first color histogram; T is the number of pixel points with the pixel value λ in the first color histogram; According to the first color histogram a ε The proportion coefficient of the pixel points with the pixel value λ in the second color histogram b ε The proportion coefficient of the pixel points with the pixel value λ in the second color histogram b ε λ , and calculating the similarity sim(a, b) of the first frame scene image a and the second frame scene image b. wherein d a is the moving length of the AGV in the first frame of scene image; d b is the moving length of the AGV in the second frame of scene image; d(a0, b0) is the distance between the midpoint a0 of the AGV in the first frame of scene image and the midpoint b0 of the AGV in the second frame of scene image; A1 is the first weight related to the moving length of the AGV; M a is the number of pixel points included in the AGV in the first frame of scene image; M b is the number of pixel points included in the AGV in the second frame of scene image; A2 is the second weight related to the number of pixel points included in the AGV; A3 is the third weight related to the color histogram corresponding to the AGV; A1+A2+A3=1; C is the number of pixel values included in the first color histogram, and is also the number of pixel values included in the second color histogram, and the number of pixel values included in the first color histogram is equal to the number of pixel values included in the second color histogram; The similarity between the first frame of scene image a and the second frame of scene image b determines the moving trajectory of the unmanned vehicle, and the trajectory information obtained by tracking control is verified according to the moving trajectory.

Images