Kalman filtering-based automatic driving vehicle accumulative error reduction method

Abstract

The invention discloses an automatic driving vehicle accumulative error reduction method based on Kalman filtering, and aims to improve the precision of vehicle trajectory prediction and reduce accumulative errors. The system obtains a first attitude information vector phi k1 and a second attitude information vector phi k2 of a vehicle through a wheel encoder and a vehicle-mounted GPS. And performing prediction calculation on the two attitude vectors according to the constructed Kalman filter to obtain a corrected optimal attitude information vector phi opt. And substituting the optimal attitude information vector phi opt into the starting and stopping conditions of the automatic driving vehicle, and comparing the distance obtained by calculation with a set threshold value, so as to perform judgment and calibration according to the size of an error. A simulation test shows that the actual position data of the vehicle is calibrated by the steps when the vehicle is parked each time, so that the accumulative error of the vehicle in a multi-time starting and stopping mode is reduced, the system effectively reduces the error superposition of each section of track, and the accuracy of predicting the long-distance multi-section driving track is improved.

Description

Technical Field

[0001] The present invention belongs to the field of autonomous driving and intelligent transportation system (ITS), and specifically relates to a multi-sensor data fusion method, a method for reducing the cumulative error of an autonomous driving vehicle based on Kalman filtering. Background Art

[0002] With the rapid development of autonomous driving technology, the accuracy requirements for positioning and path identification are also constantly increasing. The traditional global positioning system (GPS) has good coverage and ease of use, but in complex urban environments, due to signal interference and environmental occlusion, the positioning accuracy of GPS is insufficient and cannot meet the needs of autonomous driving vehicles in high-precision, real-time scenarios. To this end, technologies that integrate multiple sensor data have been widely used, such as inertial navigation systems, lidar, and vehicle-mounted cameras, to improve the robustness and accuracy of the system.

[0003] Existing multi-sensor fusion methods are usually based on the extended Kalman filter (EKF) algorithm, which predicts the vehicle state through a dynamic system model and updates the state estimate in combination with actual measurement data. However, these methods still have high cumulative errors when dealing with complex traffic scenarios such as stop-and-go.

[0004] To address the above problems, the present invention predicts and fuses data from GPS and wheel encoders under an extended Kalman filter to obtain the optimal posture information vector of the autonomous driving vehicle, and sets an error threshold in the stop-and-start situation for calibration, so as to improve the trajectory recognition accuracy of the autonomous driving vehicle in such repeated start-and-stop scenarios. Summary of the invention

[0005] The purpose of the present invention is to solve the problem that in complex urban environments, due to signal interference and environmental occlusion, the positioning accuracy of GPS is insufficient, and the trajectory identification of autonomous driving vehicles still has a high cumulative error. To achieve the above purpose, the present invention provides the following technical solutions:

[0006] 1. A method and system for reducing cumulative error of an autonomous driving vehicle based on Kalman filtering, characterized in that the method comprises:

[0007] Step 1: Generate the first attitude information vector Φ of the autonomous driving vehicle in the navigation coordinate system from the wheel encoder and the inertial measurement unit of the vehicle. k1 ;

[0008] Step 2: Obtain the second posture information vector Φ from the vehicle positioning system GPS of the autonomous driving vehicle k2 ;

[0009] Step 3: According to the constructed Kalman filter, the above two posture information vectors are predicted and fused to obtain the corrected optimal posture information vector Φ opt As a Kalman filter result;

[0010] Step 4: The Kalman filter result obtained in step 3 above, that is, the corrected optimal posture information vector Φ opt The parameter is brought into the start-stop environment of the autonomous driving vehicle to update and calibrate the trajectory error.

[0011] In step 1, the first posture information vector Φ of the autonomous driving vehicle in the navigation coordinate system is generated from the wheel encoder of the vehicle. k1 , characterized in that:

[0012] Step 11: Obtain the left and right wheel speed information v of the autonomous driving vehicle l and v r ;

[0013] Step 12: Generate the vehicle lateral velocity v in the vehicle body coordinate system x and the vehicle longitudinal velocity v y .

[0014] Step 13: convert the above vehicle lateral velocity v x and the vehicle longitudinal velocity v y Converted to the velocity vector V in the vehicle body coordinate system sdv , that is, V sdv =[v x ,v y ,0] T

[0015] Step 14: transform the velocity vector V in the vehicle body coordinate system sdv Transformed into the first attitude information vector Φ based on the GPS navigation coordinate system k1 .

[0016] In step 2, the second posture information vector Φ is obtained from the vehicle positioning system GPS of the autonomous driving vehicle. k2 Features:

[0017] Step 21, setting a data transmission rate of 10 Hz to receive vehicle GPS positioning data, including is the latitude positioning information of the autonomous driving vehicle, λ GPS Autonomous driving vehicle longitude positioning information;

[0018] Step 22: convert the above data into the slip angle α of the autonomous driving vehicle sa , relative azimuth β ra and the relative travel distance d GPS ;

[0019] Step 23, the above phase information is respectively used in the due east and due north directions in the navigation coordinate system to obtain the velocity v ge and v gn , get the second posture information vector Φ k2 .

[0020] In step 3, the above two sensor data are predicted and fused through the extended Kalman filter equation group including the prediction equation, the state estimation equation, the estimated mean square error equation and the Kalman gain equation to obtain the optimal attitude information vector Φ opt , characterized in that:

[0021] (1) The prediction step predicts the system through the system's dynamic model and process covariance error.

[0022] The prediction equation is

[0023] (2) The state estimation update equation is:

[0024] (3) The Kalman filter gain equation is:

[0025] (4) The equation for estimating the mean square error is:

[0026] in, and is the prior state estimate of the autonomous driving vehicle's posture at time steps k-1 and k, and are the pose state estimates of the autonomous driving vehicle at time steps k-1 and k, and u k-1 is the control input, is a nonlinear prediction function of the system state. k is the observed value at time step k, and h is a nonlinear function of the known observation model of the system. k is the adaptive factor. B is the control matrix preset by the system. H k is the Jacobian matrix of the observation model, is its transpose, R k is the custom measurement noise mean square error matrix at time step k. is the prior error covariance matrix at time step k, P k-1 is the error covariance matrix at time step k-1.

[0027] Step 31, before using the extended Kalman filter listed above, the position state estimate of the autonomous driving vehicle is initialized to and the predicted mean square error matrix is ​​initialized as in is the initial velocity mean square error.

[0028] Step 32: After initialization, the first posture information vector Φ in the above step 1 is k1 The prior state in the state estimation update equation of the Kalman filter Then according to the second posture information vector Φ k2 The observation value μ at time step k is brought into the Kalman filter k Yes, we can use the pre-built Kalman filter to perform attitude estimation and obtain As the corrected optimal posture information vector Φ opt .

[0029] In step 4, the Kalman filter result obtained in step 3 above, i.e., the corrected optimal posture information vector Φ opt It is used as a parameter in the stop-start situation of the autonomous driving vehicle to update and calibrate the trajectory error.

[0030] The specific idea is: let the distance of the autonomous driving vehicle relative to the pedestrian traffic sign when it stops be D0, let the distance measurement error of the autonomous driving vehicle be D e , let D T is the distance change threshold of the autonomous driving vehicle, which is related to the state estimation result of the Kalman filter constructed above. T =K t ∫ΔΦ opt (t)dt,K t is the preset correlation coefficient, ΔΦ opt (t) is the difference of the optimal attitude information vector obtained in the Kalman filter in step 3 above, ΔΦ opt (t) = Φ opt (t1)-Φ opt (t0), t1-t0=20s. After the autonomous vehicle stops and restarts in front of the pedestrian traffic sign, the pedestrian traffic sign position d recorded by GPS is recorded every 20s. pcl Relative position information with the autonomous driving vehicle d GPS According to the similar coordinate transformation in step 1 and step 2 above and the difference, the relative distance change D of the start-stop measurement of the autonomous driving vehicle is obtained. i (i=1,2,3......10). If D i >D T , (i=1,2,......,10), it is considered that the error caused by the autonomous driving vehicle after starting and stopping affects the estimation of the vehicle's subsequent driving position trajectory. Then the relative distance change D of the autonomous driving vehicle when starting and stopping is obtained. iThe mean D of (i=1,2,......10) is used to subtract K from the mean D at this time. t D0 gets the difference ΔD. If the difference is greater than the distance measurement error D of the autonomous driving vehicle, e , it is also believed that the error caused by the autonomous driving vehicle after starting and stopping affects the estimation of the vehicle's subsequent driving position trajectory.

[0031] In view of the influence of the above error on the position trajectory, for the current driving section, that is, when the error from stopping to calculating the position is too large, the wheel encoder is used to estimate the position trajectory. It is also believed that the measurement error accumulated in the previous driving process of the autonomous driving vehicle is too large, so the on-board GPS is reset and re-calibrated to eliminate the accumulated error of the autonomous driving vehicle in the previous section. This ensures that the autonomous driving vehicle reduces the positioning error or other factors caused by the large error between the driving position distance and the actual position when it starts from the parking state during long-distance and multi-segment driving.

[0032] The effective effect of the present invention is to limit the cumulative error to the current trajectory segment and prevent it from propagating to subsequent trajectories, thereby effectively improving the accuracy of the autonomous driving vehicle trajectory and reducing the cumulative error of the complete trajectory of the autonomous driving vehicle. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 These are the main process steps of the present invention.

[0034] Figure 2 This is the data fusion flow chart.

[0035] Figure 3 A simulated road map for vehicles driving in a complex urban environment.

[0036] Figure 4 Schematic diagram of the simulation of the vehicle starting and stopping at the pedestrian crossing traffic sign.

[0037] Figure 5 KF trajectory identification curve diagram when there are 2 start and stop points.

[0038] Figure 6 KF trajectory identification curve diagram when there are 4 start and stop points. DETAILED DESCRIPTION

[0039] After the autonomous driving system of the autonomous driving vehicle in the simulation platform is started, the autonomous driving vehicle starts to drive with large buildings as obstacles around it. It is set that there will be 4 pedestrian traffic signs during the driving process to make the autonomous driving vehicle start and stop in the middle. Before each start and stop, the position of the autonomous driving vehicle is measured by predicting the vehicle wheel encoder model and combining the on-board GPS of the autonomous driving vehicle. The present invention provides a multi-sensor data fusion method based on the extended Kalman filter algorithm, and its specific steps are as follows: Figure 1 As shown in the figure, the accumulated error of each start and stop is eliminated, so as to obtain a more accurate trajectory distance estimation curve for a complete path.

[0040] Step 1: Generate the first attitude information vector Φ of the autonomous driving vehicle in the navigation coordinate system from the wheel encoder and the inertial measurement unit of the vehicle. k1 ;

[0041] Step 11: Obtain the left and right wheel speed information v of the autonomous driving vehicle l and v r .

[0042] Here the number of wheel revolutions of the autonomous driving vehicle is ω l =n l / θ res and ω r =n r / θ res , where θ res Represents the resolution of the wheel encoder. Finally, the current left wheel speed v of the autonomous vehicle can be obtained by multiplying the rotation speed by the wheel arc length l. l and right wheel speed v r .

[0043] Step 12: Generate the vehicle lateral velocity v in the vehicle body coordinate system x and the vehicle longitudinal velocity v y .

[0044] Here, according to the above-mentioned left and right wheel speed information v l and v r Through vehicle kinematic analysis, the movement and direction of the vehicle can be obtained. The equation of motion of a moving vehicle can be expressed as Among them, d is the distance between the left and right wheels, that is, the width of the vehicle chassis.

[0045] Step 13: convert the above vehicle lateral velocity v x and the vehicle longitudinal velocity v y Converted to the velocity vector V in the vehicle body coordinate system sdv , that is, V sdv =[v x ,v y ,0] T

[0046] Step 14: transform the velocity vector V in the vehicle body coordinate system sdv Transformed into the first attitude information vector Φ based on the GPS navigation coordinate system k1 .

[0047] Specifically, the autonomous driving vehicle is obtained in the inertial measurement unit IMU. Including the first heading angle α, the first pitch angle β and the first roll angle γ, so as to construct the rotation matrix R from the vehicle body coordinate system to the navigation coordinate system k , the expression is

[0048]

[0049] Then the first posture information vector Φ k1 =R k V sdv =[v ke ,v kn ,v ku ], here v ke ,v kn ,v ku The velocity components of the autonomous driving vehicle in the east, north, and celestial directions in the navigation coordinate system respectively.

[0050] Step 2: Obtain the second posture information vector Φ from the vehicle positioning system GPS of the autonomous driving vehicle k2 , specifically including:

[0051] Step 21: When the autonomous driving vehicle is driving, the vehicle GPS positioning data is received at a data transmission rate of 10 Hz. The positioning data includes is the latitude positioning information of the autonomous driving vehicle, λ GPS Autonomous driving vehicle longitude positioning information.

[0052] Step 22: Autonomous driving vehicle latitude positioning information, λ GPS The longitude positioning information of the autonomous driving vehicle and the earth radius parameter under GPS are converted into the slip angle α of the autonomous driving vehicle sa , relative azimuth β ra and the relative travel distance d GPS , and its combined expression is as follows:

[0053] d GPS =R GPS β ra

[0054] Step 23: convert the relative azimuth angle β ra and the relative travel distance d GPS Get the speed in the due east and due north directions in the navigation coordinate system respectively and ΔT GPS The second attitude information vector Φ is obtained by measuring the time of the positioning system. k2 =[v ge ,vgn ,0].

[0055] Step 3: The first posture information vector Φ of the autonomous driving vehicle at the same time is calculated based on the constructed Kalman filter. k1 and the second posture information vector Φ of the autonomous driving vehicle k2 Perform prediction fusion to obtain the corrected optimal posture information vector Φ opt As a result of Kalman filtering.

[0056] For the knowledge about the principle of extended Kalman filter, please refer to the relevant technology, which will not be further elaborated here; the prediction ability of extended Kalman filter can be divided into two key stages: prediction and state update. Figure 2 The extended Kalman filter of the embodiment of the present invention comprises a prediction equation, a state estimation equation, an estimated mean square error equation and a Kalman gain equation:

[0057] (1) The prediction step predicts the system through the system's dynamic model and process covariance error.

[0058] The prediction equation is

[0059] (2) The state estimation update equation is:

[0060] (3) The Kalman filter gain equation is:

[0061] (4) The equation for estimating the mean square error is:

[0062] The following is an explanation of the parameters of each equation:

[0063] In the prediction equation, is the prior state estimation result of the position of the autonomous driving vehicle at time step k, is the state estimate of the previous time step k-1, u k-1 is the control input at the previous time step k-1. is a nonlinear prediction function of the system state, which is linearized here as I is the identity matrix, A is the preset 3*3 dimensional state transfer matrix, and dt is the update period of the extended Kalman filter.

[0064] In the state estimation update equation, is the posterior state estimate of the autonomous vehicle position at time step k. k is the observed value at time step k, and h is a nonlinear function of the known observation model of the system. The designed adaptive factor λ is added to the Kalman gain kUsed to adjust the filter parameters to adapt to changes in system characteristics. Used to adjust the measurement noise covariance R and system noise covariance Q.

[0065] In the Kalman filter gain equation, the adaptive factor formula is: Where A is the above preset 3*3 dimensional state transfer matrix, and B is the system preset control matrix. k is the Jacobian matrix of the observation model, is its transpose, R k is the custom measurement noise mean square error matrix at time step k, It is a custom noise parameter. is the a priori error covariance matrix at time step k, The calculation formula is Q k-1 is the custom system noise mean square error matrix Q, is the custom system noise parameter. k-1 is the error covariance matrix of the previous time step k-1, Q k-1 is the process noise covariance matrix of the previous time step k-1, F k is the Jacobian matrix of the system state;

[0066] Step 31: Before using the extended Kalman filter equations listed above, the estimated value of the position state of the autonomous driving vehicle needs to be initialized at time k = 1 as and the predicted mean square error matrix is ​​initialized as in It is the mean square error of the initial velocity on the positive half axis of X, Y, and Z respectively in the navigation coordinate system of the autonomous driving vehicle system.

[0067] Step 32, after initialization, the above Substitute the P0 value into the prediction equation and estimate the mean square error equation P k-1 , and then the first posture information vector Φ in step 1 above k1 The prior state in the state estimation update equation of the Kalman filter Then according to the second posture information vector Φ k2 The observation value μ at time step k is brought into the Kalman filter k Yes, we can estimate the attitude based on the pre-built Kalman filter and get As the corrected optimal posture information vector Φ opt .

[0068] Step 4: The Kalman filter result obtained in step 3 above, that is, the corrected optimal posture information vector Φ optIt is brought into the autonomous driving stop-start environment as a parameter to update and calibrate the trajectory error.

[0069] Figure 3 The route used in the simulation environment is a traffic road in the city, which contains some environmental interference, such as simulated large buildings beside the road, which will reduce the accuracy of GPS positioning. The total length between the start and end points is about 200 meters, and 4 pedestrian crossings are set as vehicle start and stop points. The simulation aims to identify the positioning error from GPS by processing the dynamic movement data collected by GPS and vehicle wheel encoders.

[0070] During the repeated start-stop process of the autonomous driving vehicle, software filtering is used to correct the wheel encoder estimation error caused by wheel slippage in the vehicle start-stop state.

[0071] The specific approach is: Assume that the autonomous vehicle is 1 meter ahead of the pedestrian traffic sign each time it stops. There are a total of four pedestrian traffic signs that require the autonomous vehicle to start and stop. The simulation of the start-stop environment during the driving process of the autonomous vehicle is as follows: Figure 4 .

[0072] Assume that the initial distance of the autonomous vehicle relative to the pedestrian traffic sign is D0, and the distance measurement error of the autonomous vehicle is D e , let D T is the distance change threshold of the autonomous driving vehicle, which is related to the state estimation result of the Kalman filter constructed above. T =K t ∫ΔΦ opt (t)dt,K t is the preset correlation coefficient, ΔΦ opt (t) is the difference of the optimal attitude information vector obtained in the Kalman filter in step 3 above, ΔΦ opt (t) = Φ opt (t1)-Φ opt (t0), t1-t0=20s. After the autonomous vehicle stops and restarts in front of the pedestrian traffic sign, the pedestrian traffic sign position d recorded by GPS is recorded every 20s. pcl Relative position information with the autonomous driving vehicle d GPS According to the similar coordinate transformation in steps 1 and 2 above, the two parameters are differentiated to obtain the relative distance change D when the autonomous driving vehicle starts and stops. i (i=1,2,3......10), if the position relative distance measured by GPS changes relatively steadily, that is, from D1 to D 10 The difference between each adjacent pair is less than D T , that is, Di ≤D T ,(i=1,2,......,10), at this time we believe that the error caused by the start and stop of the autonomous driving vehicle can be ignored. i >D T , (i=1,2,......,10), we believe that the error caused by the autonomous driving vehicle after starting and stopping affects the estimation of the vehicle's subsequent driving position trajectory. Then the relative distance change D of the autonomous driving vehicle start-stop measurement obtained by the current calculation is obtained i The mean of (i=1,2,......10) Use this time mean Subtract K t D0 gets the difference ΔD. If the difference at this time is greater than the distance measurement error D of the autonomous driving vehicle e , we believe that the error caused by the autonomous driving vehicle after starting and stopping affects the estimation of the vehicle's subsequent driving position trajectory.

[0073] In view of the impact of the start-stop error on the position trajectory, for the current driving section, that is, when the calculated position error is too large from stopping, the wheel encoder is used to estimate the position trajectory. It is also believed that the automatic driving vehicle has accumulated errors at the start-stop moment due to the occlusion of complex obstacles in the city during the previous driving process. The on-board GPS is reset and re-positioned and calibrated to eliminate the accumulated errors of the automatic driving vehicle in the previous section. This ensures that the automatic driving vehicle reduces the positioning error after each start-stop state when driving a long distance, or the problem of large errors between the driving position distance and the actual position caused by other factors.

[0074] The results of the motion tests by selecting 2 and 4 stopping and starting points are as follows: Figure 5 and Figure 6 As expected, for the test with 4 stop and start points, compared with the test with 2 stop and start points, the trajectory estimate is dynamically updated by using the extended Kalman filter (EKF) and the above-mentioned specific method at each stop point as the reference point to ensure that the subsequent trajectory points accurately reflect the actual position of the vehicle. After each stop, the accumulated error is limited to the current trajectory segment by recalibrating the position information to prevent it from propagating to the subsequent trajectory, and a more accurate trajectory estimate is obtained.

Claims

1. A method for reducing the cumulative error of an autonomous driving vehicle based on Kalman filtering, characterized in that: The method comprises: Step 1: Generate the first attitude information vector Φ of the autonomous driving vehicle in the navigation coordinate system from the wheel encoder and the inertial measurement unit of the vehicle. k1 ; Step 2: Obtain the second posture information vector Φ from the vehicle positioning system GPS of the autonomous driving vehicle k2 ; Step 3: According to the constructed Kalman filter, the two posture information vectors are predicted and fused to obtain the corrected optimal posture information vector Φ opt As a result of Kalman filtering; Step 4: The Kalman filter result obtained in step 3 above, that is, the corrected optimal posture information vector Φ opt The parameter is brought into the start-stop environment of the autonomous driving vehicle to update and calibrate the trajectory error.

2. The method for reducing the cumulative error of an autonomous driving vehicle based on Kalman filtering according to claim 1, characterized in that: In step 1, the first attitude information vector Φ of the autonomous driving vehicle in the navigation coordinate system is generated from the wheel encoder of the vehicle. k1 ; Specifically including: Step 11, obtaining the left and right wheel speed information v of the autonomous driving vehicle l and v r ; Step 12, generate the vehicle lateral velocity v in the vehicle body coordinate system x and the vehicle longitudinal velocity v y ; Step 13, the above vehicle lateral speed v x and the vehicle longitudinal velocity v y Converted to the velocity vector V in the vehicle body coordinate system sdv , that is, V sdv =[v x ,v y ,0] T ; Step 14, the velocity vector V in the above vehicle body coordinate system sdv Transformed into the first attitude information vector Φ based on the GPS navigation coordinate system k1 .

3. The method for reducing the cumulative error of an automatic driving vehicle based on Kalman filtering according to claim 1, characterized in that: In step 2, the second posture information vector Φ is obtained from the vehicle positioning system GPS of the autonomous driving vehicle. k2 ; Specifically, step 21 is to receive vehicle GPS positioning data at a data transmission rate of 10 Hz, including is the latitude positioning information of the autonomous driving vehicle, λ GPS The longitude positioning information of the autonomous driving vehicle; Step 22, converting the above data into the slip angle α of the autonomous driving vehicle sa , relative azimuth β ra and the relative travel distance d GPS ; Step 23, obtain the speed v in the due east and due north directions of the navigation coordinate system respectively ge and v gn , get the second posture information vector Φ k2 .

4. The method for reducing cumulative error of an autonomous driving vehicle based on Kalman filtering according to claim 1, characterized in that: In step 3, the above two sensor data are predicted and fused through the extended Kalman filter equation group including the prediction equation, state estimation equation, estimated mean square error equation and Kalman gain equation to obtain the optimal attitude information vector Φ opt ; Specifically include: (1) The prediction step predicts the system through the system's dynamic model and process covariance error; the prediction equation is (2) The state estimation update equation is: The Kalman filter gain equation is: The estimated mean square error equation is in, and is the prior state estimate of the autonomous driving vehicle's posture at time steps k-1 and k, and are the pose state estimates of the autonomous driving vehicle at time steps k-1 and k, and u k-1 is the control input, is the nonlinear prediction function of the system state; μ k is the observed value at time step k, h is a nonlinear function of the known observation model of the system; λ k is the adaptive factor; B is the control matrix preset by the system; H k is the Jacobian matrix of the observation model, is its transpose, R k is the custom measurement noise mean square error matrix at time step k; is the prior error covariance matrix at time step k, P k-1 is the error covariance matrix at time step k-1.

5. The method for reducing cumulative error of an automatic driving vehicle based on Kalman filtering according to claim 4, characterized in that: Before using the extended Kalman filter listed above, the estimated value of the position state of the autonomous driving vehicle is initialized as and the predicted mean square error matrix is ​​initialized as in is the initial velocity mean square error.

6. The method for reducing cumulative error of an automatic driving vehicle based on Kalman filtering according to claim 4, characterized in that: After initialization, the first posture information vector Φ in step 1 above is k1 The prior state in the state estimation update equation of the Kalman filter Then according to the second posture information vector Φ k2 The observation value μ at time step k is brought into the Kalman filter k Yes, we can use the pre-built Kalman filter to perform attitude estimation and obtain As the corrected optimal posture information vector Φ opt .

7. The method for reducing cumulative error of an automatic driving vehicle based on Kalman filtering according to claim 1, characterized in that: In step 4, the Kalman filter result obtained in step 3, that is, the corrected optimal posture information vector Φ opt As a parameter, it is introduced into the parking and starting situation of the autonomous driving vehicle, so as to update and calibrate the trajectory error; it is characterized in that: the distance of the autonomous driving vehicle relative to the pedestrian traffic sign when it stops is set as D0, and the distance measurement error of the autonomous driving vehicle is set as D e , let D T is the distance change threshold of the autonomous driving vehicle, which is related to the state estimation result of the Kalman filter constructed above. T =K t ∫ΔΦ opt (t)dt,K t is the preset correlation coefficient, ΔΦ opt (t) is the difference of the optimal attitude information vector obtained in the Kalman filter in step 3 above, ΔΦ opt (t) = Φ opt (t1)-Φ opt (t0), t1-t0=20s; after the autonomous vehicle stops and restarts in front of the pedestrian traffic sign, the pedestrian traffic sign position d recorded by GPS is recorded every 20s. pcl Relative position information with the autonomous vehicle d GPS According to the similar coordinate transformation in step 1 and step 2 above and the difference, the relative distance change D of the start-stop measurement of the autonomous driving vehicle is obtained. i If D i >D T , it is considered that the error caused by the autonomous driving vehicle after starting and stopping affects the estimation of the vehicle's subsequent driving position trajectory; then the relative distance change D of the autonomous driving vehicle when starting and stopping is obtained. i The mean Use this time mean Subtract K t D0 gets the difference ΔD. If the difference is greater than the distance measurement error D of the autonomous driving vehicle, e , it is also believed that the error caused by the autonomous driving vehicle after starting and stopping affects the estimation of the vehicle's subsequent driving position trajectory.

8. The method for reducing cumulative error of an automatic driving vehicle based on Kalman filtering according to claim 7, characterized in that: In view of the influence of the above error on the position trajectory, for the current driving section, that is, when the error from stopping to calculating the position is too large, the focus is on using the wheel encoder to estimate the position trajectory; and it is considered that the autonomous driving vehicle has accumulated too much measurement error in the previous driving process, so the on-board GPS is reset and re-positioned and calibrated to eliminate the accumulated error of the autonomous driving vehicle in the previous section. In this way, the autonomous driving vehicle can reduce the positioning error or other factors that cause a large error between the driving position distance and the actual position when driving from a stop to a multi-segment driving distance over a long distance.

Images