CN107340298B - Balance car system measurement and control method based on camera road surface detection - Google Patents

Balance car system measurement and control method based on camera road surface detection Download PDF

Info

Publication number
CN107340298B
CN107340298B CN201710515182.XA CN201710515182A CN107340298B CN 107340298 B CN107340298 B CN 107340298B CN 201710515182 A CN201710515182 A CN 201710515182A CN 107340298 B CN107340298 B CN 107340298B
Authority
CN
China
Prior art keywords
value
angle
road surface
follows
noise
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201710515182.XA
Other languages
Chinese (zh)
Other versions
CN107340298A (en
Inventor
胡维平
胡帆
蒙佳
欧纯洋
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guangxi Normal University
Original Assignee
Guangxi Normal University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Guangxi Normal University filed Critical Guangxi Normal University
Priority to CN201710515182.XA priority Critical patent/CN107340298B/en
Publication of CN107340298A publication Critical patent/CN107340298A/en
Application granted granted Critical
Publication of CN107340298B publication Critical patent/CN107340298B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • G01N21/88—Investigating the presence of flaws or contamination
    • G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • G01N21/88—Investigating the presence of flaws or contamination
    • G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
    • G01N2021/8887—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges based on image processing techniques

Abstract

The invention discloses a balance car system measurement and control method based on camera road surface detection, which comprises the steps of utilizing a camera to carry out gradient difference on LBP texture characteristic statistical information in a collected road surface image to obtain road surface texture distribution, matching with comparison difference before and after filtering of an inertial sensor, carrying out weighted summation on the two to obtain -classified estimation of road surface disease degree, finally improving an inertial sensor measurement and calculation system of a balance car, adaptively changing a Kalman filter in real time to realize more accurate prediction angle, and selecting a motor control strategy according to a camera feedback value.

Description

Balance car system measurement and control method based on camera road surface detection
Technical Field
The invention relates to the technical field of balance car measurement and control, in particular to balance car system measurement and control methods based on camera road surface detection.
Background
With the rapid development of the balance car market, a large number of inferior products in the market are also emitted, safety accidents are frequent, and the improvement of the robustness and the practical range of a balance car control system is of great importance.
At present, the manned balance vehicles sold in the market or the manned balance vehicles developed by research units are modified and adjusted in fine branches and minor branches on the basis of the traditional inertial sensors, the limitation of the single sensor on measuring and calculating control signals cannot be broken through all the time, the technical research of the manned balance vehicles is not stopped, and large blank areas are left in safety problems.
Disclosure of Invention
Aiming at the defects in the prior art, the invention aims to improve the control precision of the balance car and ensure the overall safety.
In order to solve the technical problems, the technical scheme adopted by the invention is balance car system measurement and control methods based on camera road surface detection, a method of gradient difference is carried out on LBP texture feature statistical information in a collected road surface image by using a camera to obtain road surface texture distribution, difference comparison before and after filtering is matched with an inertial sensor, the road surface texture distribution and the difference comparison are weighted and summed to obtain estimation of road surface disease degrees, and the estimation and calculation system is finally used for improving an inertial sensor measurement and calculation system of a balance car, a real-time self-adaptive Kalman filter is changed to realize more accurate prediction angle, and a motor control strategy is selected according to a camera feedback value, and the method comprises the following specific processes:
the process 1 uses an accelerometer and a gyroscope to collect data, utilizes a Kalman filter to realize the fusion of collected data signals, denoises to obtain the real-time inclination angle Pitch information of a vehicle body in the Y-axis direction, namely the front and back directions, obtains the predicted convergent angle information, and finally performs difference with the integral angle of the gyroscope, and comprises the following specific steps:
(1) the system incorporating discrete control processes is described by Linear machine differential equations (Linear stochastics difference equation):
X(k|k-1)＝AX(k-1|k-1)+BU(k) (1)
according to the equation, substituting the current time measured value of the gyroscope and adding the drift amount of the gyroscope to obtain the predicted value Angle of the current Angle, and writing the predicted value Angle into a matrix form is as follows:
Figure GDA0002297981090000021
(2) in the prediction of the covariance matrix, two input values need to be defined in the calculation process, which are respectively: drift noise of the gyro sensor and angle noise of the accelerometer sensor,
P(k|k-1)＝AP(k-1|k-1)A’+Q (3)
wherein Q in the formula is a vectorThe covariance matrix of (a), namely:
Figure GDA0002297981090000023
since the drift noise and also the angle noise are independent from each other;
cov(Q_bias,Angel)＝0 (5)
cov(Angel,Q_bias)＝0 (6)
the defined process angle noise covariance parameter and the measured noise covariance parameter in the program are respectively:
Q_angle＝0.001；Q_gyro＝0.003；R_angle＝0.5；
(3) calculating a gain factor of the Kalman filter, which is a two-dimensional vector of gain factors set to
Figure GDA0002297981090000024
For the correction of the angle and the angular velocity, the expression is:
Kg(k)＝P(k|k-1)H’/(HP(k|k-1)H’+R) (7)
the constants R appearing in the equation refer to the noise figure of the acceleration measurement, and the figure in the program is initially defined as R _ angle ═ 0.5;
(4) correcting a predicted value by using a Kalman gain coefficient, taking a difference value of angles obtained by an accelerometer and a gyroscope as an error value Angle _ err, and correcting the predicted value by using a product of the Kalman gain coefficient and the error, wherein the formula is as follows:
X(k|k)＝X(k|k-1)+Kg(k)(Z(k)-HX(k|k-1)) (8)
updating the prediction error value and the angle value simultaneously:
Angel+＝K_0*Angel_err (9)
Q_Bias+＝K_1*Angel_err (10)
(5) the matrix covariance P matrix is updated this time, and is mainly used for predicting covariance in the following iterations, and the equation is as follows:
P(k|k-1)＝(I-Kg(k)H)P(k|k-1) (11)
(6) measuring the angle X at the moment k(k|k-1) differences are made with the predicted angle value at the k moment, average angle average error measurements are counted in every 20 angle measurements,
Figure GDA0002297981090000031
in the process 2, real-time angular velocity of the mass center of the vehicle body on the Z axis and inclination angle data of the X axis are obtained through a gyroscope sensor, rolls (namely real-time inclination angles in the left and right directions) of the mass center of the vehicle body in the horizontal direction are obtained through Kalman fusion filtering, and the data are subjected to the operation of conversion to obtain mathematical statistical characteristics of pavement disease degree representation, wherein the method comprises the following specific steps of:
(1) performing Kalman prediction fusion by using gyroscope to measure and calculate Z-axis acceleration information and accelerometer to measure and calculate X-axis angle information to obtain posterior estimation values theta converging to accurate valuesr,
(2) The angle Roll measured on the X axis is used for solving the sine value sin theta of the angle RollrAnd the sine value is multiplied by the wheel base length D between the left wheel and the right wheel of the vehicle body to obtain a value h which can be expressed as the height of the center of mass of the vehicle body supported by the uneven road surface,
h＝D*sinθr；
(3) dividing the height value h of the vehicle body raised by the uneven road surface by the height 2d of the chassis of the vehicle body to obtain the tilt rate Q of the whole vehicle, finally representing the Q value as the tilt degree of the road surface to the vehicle body of the balance vehicle, and when the value is 1, achieving the ultimate safety state that the vehicle body can bear the maximum tilt, and achieving the effect of returning to , wherein the formula is as follows:
Q＝D*Sinθr/2d (13)；
the process 3 continuously acquires image information of a road in front of a balance car body through an image sensor, and performs graying, image segmentation, LBP feature extraction statistics and binning conversion processing on the image information, and specifically comprises the following steps:
(1) collecting road surface picture digital signal a by utilizing COMS camera(t)Extracting LBP characteristics and carrying out statistics, the steps are as follows:
1) converting the RGB color image into a gray-scale image;
aGray(t)＝R*0.299+G*0.587+B*0.144
2) optimizing the conversion of a floating-point operation structure;
aGray(t)＝(R*299+G*587+B*114+500)/1000
3) dividing the image digital signal into 1 × 10-10 sub-regions by dividing the image into equal vertical segments; then the pixel resolution for each sub-region is 640 x 48;
Figure GDA0002297981090000041
(2) local binarization; a local dynamic threshold value method is selected for binaryzation, so that the accurate extraction of the target texture is realized; the specific extraction steps are as follows:
1) in the image traversal process, 3 x 3 gray windows are extracted each time as objects, and the central gray value of the window is selected as the threshold value a of the current binarization(t)(i,j)Respectively connecting 8 adjacent pixel points with the threshold value a(t)(i,j)Comparing, if the value is larger than the threshold value, assigning 1, otherwise, assigning 0;
Figure GDA0002297981090000051
2) numbering from the top left corner results in binary digits of 8 bits, followed by bi-1，j-1bi-1，jbi-1，j+1bi，j+1bi+1，j+1bi+1，jbi+1，j-1bi，j-1](ii) a The binary number of 8 bits is the LBP value of the pixel point of the center window;
LBP(t)(i,j)＝sum[bi-1,j-1,bi-1,j,bi-1,j+1, bi,j-1,bi,j,bi-1,j+1, bi+1,j-1,bi+1,j,bi+1,j+1]
3) listing all eight 8-bit binary rotation arrangement modes, and selecting the minimum LBP as the LBP value of a central window pixel point to represent the texture information of the points;
Figure GDA0002297981090000052
Figure GDA0002297981090000053
(3) counting LBP histogram gradient distribution statistics, and making a pavement damage degree estimation value into , wherein the specific steps are as follows;
1) performing LBP histogram statistics on each subregion map, dividing the distribution histograms from 0 to 255 into 16 intervals for statistics, and counting the number of pixel points in each interval to account for the global proportion to obtain set arrays R containing 16 elementsnm(1 ≦ n ≦ 10; 1 ≦ m ≦ 16), and new sequences R generated by operating the absolute difference of these 16 elementsnk(1 ≦ n ≦ 10; 1 ≦ k ≦ 8), thereby characterizing the LBP histogram probability density distribution gradient;
Rn,2*k＝Rn,2*k-Rn,2*k-1
2) count its array RnmSum of Cn(1≦n≦10)；
Figure 100002_1
CnConcentrating in a fixed interval, the minimum value is 0 (at the moment, the texture distribution of the road surface is minimum), and the meaning of the representation is that the road surface is in the most flat state, otherwise, the C isnThe value is closer to 1 (the texture contrast is maximum), which indicates that the road surface has serious diseases;
process 4 quantifying value C of degree of road surface damage acquired from image sensor and inertial sensornAnd PiWeighted summation is carried out, and the two can be combinedThe final reliability decision obtains measurement variances for measuring the road surface damage degree, namely sensor noise values, the current vehicle body X-axis inclination angle is theta, a noise variance estimation value I is shown as follows:
process 5 performs a debugging method based on the accuracy and reliability of the respective sensors.
method for determining weight factor K of features by empirical debugging1，K2，K3The values of the three statistical characteristics are weighted and summed, and three parameters are adjusted until the balance of the vehicle body is optimal, wherein the parameter adjustment rule is as follows:
K1+K2＝1；
K3＜1
the second method comprises the following steps: according to the measured C under different road surfacesnQ and PiPerforming least square fitting under matlab to perform parameter setting according to the noise variance estimation value I corresponding to the value, fitting an appropriate -order function of a four-dimensional space, and obtaining a parameter K1，K2And K3The function is as follows:
Figure GDA0002297981090000063
the process 6 uses the measurement noise variance estimation value I introduced for the road surface diseases as an input value of an initial value of the Kalman prediction at the next moments to predict the angle value of the balance car body at the next moments, and comprises the following specific steps:
(1) referring to the implementation process of the kalman filter in process 1, the defined process noise covariance parameters in the program are:
Q_angle＝0.001；Q_gyro＝0.003；
(2) the noise variance estimation value I introduced by the road surface is used as a result and is output to a measuring and calculating system of the balance car to update R _ angle in real time, the Kalman filter interface is adaptively changed, Q _ angle is 0.001, and Q _ gyro value is unchanged, and the formula is as follows:
R_angel＝I；
(3) and (3) obtaining a roll angle (pitch) as a final inclination angle of the balance car according to the complete calculation process in the process 1, sending the roll angle (pitch) into a motor controller for calculation, and obtaining and executing motor control quantity.
Process 7 is according to the estimated value of calculating of camera to road surface disease degree, contrast threshold value I, select following motor control strategy, this scheme sets up I0.6 (the value is different between the different balance cars), when the detected value is less than threshold value I, consider the road surface flat, motor control strategy selection fuzzy PID control this moment, if the detected value is greater than threshold value I, choose motor control strategy for cascade PID control, angle after will optimizing simultaneously, during angular velocity measured value substitutes PID controller, obtain and carry out motor control volume:
Figure GDA0002297981090000071
the method comprises the following specific steps:
(1) and after calculating the noise variance estimation value I of the road surface bumping degree, storing and finally using the noise variance estimation value I as a decision basis of a motor control strategy.
(2) When I is(t)When the signal is smaller than I, the system considers that the road surface is relatively flat, and the measuring signal of the sensor is stable and has less noise. The fuzzy PID control is selected, the operation quantity provided by a user is adapted according to the model with better signal following capability and faster response capability of the system, and the fuzzy PID also has the advantages of accurate control quantity execution and simple operation, and the specific formula is as follows:
Figure GDA0002297981090000072
the parameter adjustment equation is as follows:
Kp＝Kp0+Δkp
Ki＝Ki0+Δki
Kd＝Kd0+Δkd
(3) when I is(t)When the amplitude is larger than I, the system considers that the road surface is bumpy, and the measuring signal of the sensor contains non-artificial jitter and has high noise content. The cascade PID control is selected, compared with a single-ring PID, the cascade PID control is simple in structure and accurate in control, the anti-interference performance of a system is enhanced, however, due to the fact that the integral terms are too many due to the structural diversification of the cascade PID control, the control lags behind the system control, the cascade PID control is just suitable for a bumpy road surface, the system is required to be suitable for a dullness point of a peak burr caused by the bumpy road surface, and therefore the cascade PID control is more beneficial to safe operation of.
(4) After control cycles are completed, wait for the system clock to arrive, and cyclically repeat process 1 to process 7.
The technical scheme of the invention can obtain the following beneficial effects:
(1) under different pavement conditions, visual cooperation with an inertial sensor is introduced for data fusion, and the pavement disease flatness is automatically detected and quantified.
(2) Under the condition of complex road surface, the road surface information obtained by vision is accessed into the Kalman prediction model, dynamic self-adaptation of the model is realized, the convergence speed in the iterative process of the filter can be accelerated, and the following performance of the prediction signal can be accelerated.
(3) And (3) making a decision on the visually obtained road surface information to select a motor control method (cascade PID and fuzzy PID), and inputting a control signal predicted by the Kalman into a PID controller to obtain a motor control quantity more suitable for the road surface.
(4) Visual information collected by the visual sensor can be stored, and can be taken for manual pavement analysis or early warning treatment when needed, so that man-machine interaction is enhanced, and overall safety is improved.
Drawings
FIG. 1 is a flow chart embodying the present invention;
FIG. 2 is a comparison graph of convergence rates for adaptive input Kalman prediction and fixed input Kalman prediction;
FIG. 3 is an enlarged detail view of the adaptive input Kalman prediction and fixed input Kalman prediction schemes for the measurement and calculation of 10-degree noisy signals;
FIG. 4 is a graph comparing the effect of the adaptive input Kalman prediction scheme and the fixed input Kalman prediction scheme.
FIG. 5 is a comparison graph of step response curves for bit-cascaded PID and fuzzy PID.
Detailed Description
The following is a description of an embodiment of the present invention, but not intended to be limiting.
Fig. 1 shows a specific flow chart of the present invention, road surface evaluation schemes based on the combination of LBP features and inertial sensors and an improved balance car angle measurement and calculation method, a method of gradient difference is performed on LBP texture feature statistical information in a collected road surface image by using a camera to obtain road surface texture distribution, comparison difference before and after filtering of an inertial sensor is matched, the two are weighted and summed to obtain road surface disease degree estimates, which are used for improving an inertial sensor measurement and calculation system of a balance car, and finally the road surface disease degree estimates are used as a control strategy selection basis to further optimize motor control, and the method comprises the following specific processes:
the method comprises the following steps that 1, an accelerometer and a gyroscope are used for respectively acquiring real-time angular velocity of a mass center of a vehicle body on an X axis and inclination angle data of a Y axis, a Kalman filter is used for realizing data fusion of acquired data signals, real-time inclination angle Pitch information of the vehicle body in the Y axis direction, namely the front side direction and the rear side direction, is obtained through denoising, predicted convergent angle information is obtained, and finally difference is carried out on the predicted convergent angle information and an integral dive angle of the gyroscope; the method comprises the following specific steps:
(1) the system incorporating discrete control processes is described by Linear machine differential equations (Linear stochastics difference equation):
X(k|k-1)＝AX(k-1|k-1)+BU(k) (1)
according to the equation, substituting the current time measured value of the gyroscope and adding the drift amount of the gyroscope to obtain the predicted value Angle of the current Angle, and writing the predicted value Angle into a matrix form is as follows:
(2) in the prediction of the covariance matrix, two input values need to be defined in the calculation process, which are respectively: drift noise of the gyro sensor and angle noise of the accelerometer sensor,
P(k|k-1)＝AP(k-1|k-1)A’+Q (3)
wherein Q in the formula is a vector
Figure GDA0002297981090000102
The covariance matrix of (a), namely:
Figure GDA0002297981090000103
since the drift noise and also the angle noise are independent from each other;
cov(Q_bias,Angel)＝0 (5)
cov(Angel,Q_bias)＝0 (6)
the defined angular velocity noise and gyroscope noise scale factors in the program are respectively:
Q_angle＝0.001；Q_gyro＝0.003；
(3) calculating a gain factor of the Kalman filter, which is a two-dimensional vector of gain factors set to
Figure GDA0002297981090000104
For the correction of the angle and the angular velocity, the expression is:
Kg(k)＝P(k|k-1)H’/(HP(k|k-1)H’+R) (7)
the constants R appearing in the equation refer to the noise figure of the acceleration measurement, and the figure in the program is initially defined as R _ angle ═ 0.5;
(4) correcting a predicted value by using a Kalman gain coefficient, taking a difference value of angles obtained by an accelerometer and a gyroscope as an error value Angle _ err, and correcting the predicted value by using a product of the Kalman gain coefficient and the error, wherein the formula is as follows:
X(k|k)＝X(k|k-1)+Kg(k)(Z(k)-HX(k|k-1)) (8)
updating the prediction error value and the angle value simultaneously:
Angel+＝K_0*Angel_err (9)
Q_Bias+＝K_1*Angel_err (10)
(5) the matrix covariance P matrix is updated this time, and is mainly used for predicting covariance in the following iterations, and the equation is as follows:
P(k|k-1)＝(I-Kg(k)H)P(k|k-1) (11)
(6) measuring the angle X at the moment k(k|k-1) differences are made with the predicted angle value at the k moment, average angle average error measurements are counted in every 20 angle measurements,
in the process 2, real-time angular velocity of the mass center of the vehicle body on the Z axis and inclination data of the X axis are obtained through a gyroscope sensor, rolls (namely real-time inclinations in the left and right sides) of the mass center of the vehicle body in the horizontal direction of the mass center of the vehicle body on the X axis are obtained through Kalman fusion filtering, and the data are subjected to the operation of conversion to obtain mathematical statistical characteristics (inclination degree) of pavement disease degree representation, and the method specifically comprises the following steps:
(1) performing Kalman prediction fusion by using gyroscope to measure and calculate Z-axis acceleration information and accelerometer to measure and calculate X-axis angle information to obtain posterior estimation values theta converging to accurate valuesr,
(2) The angle Roll measured on the X axis is used for solving the sine value sin theta of the angle RollrAnd the sine value is multiplied by the wheel base length D between the left wheel and the right wheel of the vehicle body to obtain a value h which can be expressed as the height of the center of mass of the vehicle body supported by the uneven road surface,
h＝D*sinθr；
(3) dividing the height value h of the vehicle body raised by the uneven road surface by the height 2d of the chassis of the vehicle body to obtain the inclination rate Q of the whole vehicle, finally representing the Q value as the bumping degree of the road surface to the vehicle body of the balance vehicle, and when the value is 1, achieving the limit state of the maximum inclination which can be borne by the vehicle body, achieving the effect of , wherein the formula is as follows:
Q＝D*Sinθr/2d (13)；
the process 3 continuously acquires image information of a road in front of a balance car body through an image sensor, and performs graying, image segmentation, LBP feature extraction statistics and binning conversion processing on the image information, and specifically comprises the following steps:
(1) collecting road surface picture digital signal a by utilizing COMS camera(t)Extracting LBP characteristics and carrying out statistics, the steps are as follows:
1) converting the RGB color image into a gray-scale image;
2) optimizing the conversion of a floating-point operation structure;
Figure GDA0002297981090000122
3) dividing the image digital signal into 1 × 10-10 sub-regions; then the pixel resolution for each sub-region is 640 x 48;
Figure GDA0002297981090000123
(2) local binarization; a local adaptive threshold method is selected for binaryzation, so that the accurate extraction of the target texture is realized; the specific extraction steps are as follows:
1) in the image traversal process, 3 x 3 gray windows are extracted each time as objects, and the central gray value of the window is selected as the threshold value a of the current binarization(t)(i,j)(ii) a Respectively connecting 8 adjacent pixel points with the threshold value a(t)(i,j)(ii) a Comparing, if the value is larger than the threshold value, assigning 1, otherwise, assigning 0;
Figure GDA0002297981090000124
2) numbering from the top left corner results in binary digits of 8 bits, followed by bi-1，j-1bi-1，jbi-1，j+1bi，j+1bi+1，j+1bi+1，jbi+1，j-1bi，j-1](ii) a The binary number of 8 bits is the LBP value of the pixel point of the center window;
LBP(t)(i,j)＝sum[bi-1,j-1,bi-1,j,bi-1,j+1, bi,j-1,bi,j,bi-1,j+1, bi+1,j-1,bi+1,j,bi+1,j+1]
3) listing all eight 8-bit binary rotation arrangement modes, and selecting the minimum LBP as the LBP value of a central window pixel point to represent the texture information of the points;
Figure GDA0002297981090000132
(3) counting LBP histogram gradient distribution statistics, and making a pavement damage degree estimation value into , wherein the specific steps are as follows;
1) performing LBP histogram statistics on each subregion map, dividing the distribution histograms from 0 to 255 into 16 intervals for statistics, and counting the number of pixel points in each interval to account for the global proportion to obtain set arrays R containing 16 elementsnm(1 ≦ n ≦ 10; 1 ≦ m ≦ 16), and new sequences R generated by operating the absolute difference of these 16 elementsnk(1≦n≦10；1≦k≦8)；
rN,2*K＝RN,2*K-RN,2*K-1
2) Count its array rnmSum of Cn(1≦n≦10)；
Figure GDA0002297981090000133
CnCentered in a fixed interval, the minimum value is 0, which characterizes the most flat state of the road surface, whereas S is the minimum valuenThe value is closer to 1, indicating the damage course of the road surfaceThe degree is severe;
process 4 quantifying value C of degree of road surface damage acquired from image sensor and inertial sensornDegree of inclination Q and PiAnd performing linear programming combined weighted summation, and finally determining to obtain introduced measurement variances for measuring the road surface damage degree by integrating the credibility of the three, namely the noise value of the sensor, wherein the current Y-axis inclination angle of the vehicle body is theta, and the noise variance estimation value I is as follows:
Figure GDA0002297981090000141
process 5 is an empirical tuning method based on the accuracy and reliability of each sensor.
method for determining weight factor K of features by empirical debugging1，K2，K3The values of the three statistical characteristics are weighted and summed, and three parameters are adjusted until the balance of the vehicle body is optimal, wherein the parameter adjustment rule is as follows:
K1+K2＝1；
K3＜1
the second method comprises the following steps: according to the measured C under different road surfacesnQ and PiPerforming least square fitting under matlab to perform parameter setting according to the noise variance estimation value I corresponding to the value, fitting an appropriate -order function of a four-dimensional space, and obtaining a parameter K1，K2And K3The function is as follows:
Figure GDA0002297981090000142
the process 6 uses the measurement noise variance estimation value I introduced for the road surface diseases as an input value of an initial value of the Kalman prediction at the next moments to predict the angle value of the balance car body at the next moments, and comprises the following specific steps:
(1) referring to the implementation process of the kalman filter in process 1, the defined process noise covariance parameters in the program are:
Q_angle＝0.001；Q_gyro＝0.003；
(2) the noise variance estimation value I introduced by the road surface is used as a result and is output to a measuring and calculating system of the balance car to update R _ angle in real time, the Kalman filter interface is adaptively changed, Q _ angle is 0.001, and Q _ gyro value is unchanged, and the formula is as follows:
R_angel＝I；
(3) and (3) obtaining a roll angle (pitch) as a final inclination angle of the balance car according to the complete calculation process in the process 1, sending the roll angle (pitch) into a motor controller for calculation, and obtaining and executing motor control quantity.
Process 7 is according to the estimated value of calculating of camera to road surface disease degree, contrast threshold value I, select following motor control strategy, this scheme sets up I0.6, when the detected value is less than threshold value I, think the road surface flat, motor control strategy selection fuzzy PID control this moment, if the detected value is greater than threshold value I, choose motor control strategy for cascade PID control, the angle after will optimizing simultaneously, during angular velocity measured value substitutes the PID controller, obtain and carry out the motor control volume:
Figure GDA0002297981090000151
the method comprises the following specific steps:
(1) and after calculating the noise variance estimation value I of the road surface bumping degree, storing and finally using the noise variance estimation value I as a decision basis of a motor control strategy.
(2) When I is(t)When the signal is smaller than I, the system considers that the road surface is relatively flat, and the measuring signal of the sensor is stable and has less noise. The fuzzy PID control is selected, the operation quantity provided by a user is adapted according to the model with better signal following capability and faster response capability of the system, and the fuzzy PID also has the advantages of accurate control quantity execution and simple operation, and the specific formula is as follows:
the parameter adjustment equation is as follows:
Kp＝Kp0+Δkp
Ki＝Ki0+Δki
Kd＝Kd0+Δkd
(3) when I is(t)When the amplitude is larger than I, the system considers that the road surface is bumpy, and the measuring signal of the sensor contains non-artificial jitter and has high noise content. The cascade PID control is selected, compared with a single-ring PID, the cascade PID control is simple in structure and accurate in control, the anti-interference performance of a system is enhanced, however, due to the fact that the integral terms are too many due to the structural diversification of the cascade PID control, the control lags behind the system control, the cascade PID control is just suitable for a bumpy road surface, the system is required to be suitable for a dullness point of a peak burr caused by the bumpy road surface, and therefore the cascade PID control is more beneficial to safe operation of.
(4) After control cycles are completed, the process of 1-7 is repeated circularly after the arrival of the system clock.
FIG. 2 shows that in Process 5, when K is1，K2Under the condition that the data are both 0, the data feedback of a single camera sensor is used for comparing the convergence speed of variable Kalman prediction and fixed Kalman prediction of adaptive adjustment input, comparisons are carried out on two angle measurement and calculation schemes of the balance car in the scheme, wherein the standard input angle is a signal of 10 degrees, the two angle measurement and calculation schemes can well keep the following performance on the measurement angle in the whole view, and the convergence performance on the accurate angle is realized, but the self-adaptive Kalman prediction scheme is improved by 6 percentage points on the traditional basis by comparing the variance of the results in the two angle measurement and calculation schemes.
Fig. 3 shows detail amplification of the measurement of a 10-degree noisy signal for two angle measurement schemes of adaptive input prediction and conventional prediction, both of which show -degree prediction performance at the initial moment of iteration and do not show better recognition.
Fig. 4 shows the comparison between the adaptive input kalman prediction scheme and the fixed input kalman prediction scheme, which shows definite differences after the iteration times are increased as time is accumulated, and especially when the measurement signal has large jumps, the adaptive prediction scheme shows better convergence and less accumulated error.
FIG. 5 is a graph comparing step response curves for a cascade PID and a fuzzy PID. It can be seen that the response time and static error of the fuzzy PID control are superior to those of the cascade PID control, the quick response may not be suitable for the control of the bumpy road surface, and the time lag existing in the cascade PID control is used for the bumpy road surface and the trend of positive and negative offset of noise may exist, so as to obtain a more real control quantity.
The technical scheme of the invention can obtain the following beneficial effects:
(1) under different pavement conditions, visual cooperation with an inertial sensor is introduced for data fusion, and the pavement disease flatness is automatically detected and quantified.
(2) Under the condition of complex road surface, the road surface information obtained by vision is accessed into the Kalman prediction model, dynamic self-adaptation of the model is realized, the convergence speed in the iterative process of the filter can be accelerated, and the following performance of the prediction signal can be accelerated.
(3) And (3) making a decision on the visually obtained road surface information to select a motor control method (cascade PID and fuzzy PID), and inputting a control signal predicted by the Kalman into a PID controller to obtain a motor control quantity more suitable for the road surface.
(4) Visual information collected by the visual sensor can be stored, and can be taken for manual pavement analysis or early warning treatment when needed, so that man-machine interaction is enhanced, and overall safety is improved.
The embodiments of the present invention have been described in detail with reference to the accompanying drawings, but the present invention is not limited to the described embodiments. It will be apparent to those skilled in the art that various changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention.

Claims (8)

  1. The method is characterized in that a camera is used for carrying out gradient difference on LBP texture feature statistical information in a collected pavement image to obtain pavement texture distribution, the difference is compared before and after filtering of an inertial sensor, the two are weighted and summed to obtain estimation of pavement damage degrees, and finally the estimation method is used for improving an inertial sensor measuring and calculating system of the balance car, a Kalman filter is changed in a real-time self-adaptive mode to achieve more accurate prediction angles, and a motor control strategy is selected according to a camera feedback value, and the method comprises the following specific processes:
    in the process 1, an accelerometer and a gyroscope are used for acquiring data, a Kalman filter is used for realizing signal fusion of the acquired data, real-time inclination angle Pitch information of a vehicle body in the Y-axis direction, namely the front side and the rear side directions, is obtained by denoising, predicted convergent angle information is obtained, and difference is finally carried out on the predicted convergent angle information and an integral angle of the gyroscope;
    in the process 2, real-time angular velocity data of a vehicle mass center on a Z axis and inclination angle data of an X axis are obtained through a gyroscope sensor, rolls of the X axis in the horizontal direction of the vehicle mass center, namely real-time inclination angles in the left and right side directions are obtained through Kalman fusion filtering, and the data are subjected to the operation of conversion to obtain mathematical statistical characteristics of pavement disease degree representation;
    continuously acquiring image information of a road in front of a balance car body through an image sensor, and performing graying, image segmentation, LBP feature extraction statistics and binning transformation processing on the image information;
    process 4 quantifying value C of degree of road surface damage acquired from image sensor and inertial sensornAnd PiCarrying out weighted summation, and finally determining to obtain measurement variances for measuring the degree of the road surface diseases by integrating the reliability of the weighted summation and the reliability of the weighted summation, namely the noise value of the sensor, wherein the current Y-axis inclination angle of the vehicle body is theta, and the noise variance estimation value I is represented by the following formula:
    step 5, carrying out a debugging method according to the accuracy and the reliability of each sensor;
    the process 6 takes the measurement noise variance estimation value I introduced for the road surface diseases as an input value of an initial value of the Kalman prediction at the next moments to predict the angle value of the balance car body at the next moments;
    and 7, selecting a next motor control strategy according to the estimated value of the road surface disease degree measured by the camera and the noise variance estimated value I, considering that the road surface is flat when the detection value is smaller than the noise variance estimated value I, selecting fuzzy PID control as the motor control strategy at the moment, selecting the motor control strategy as cascade PID control if the detection value is larger than the noise variance estimated value I, and substituting the optimized angle and angular velocity measured values into a PID controller to obtain and execute motor control quantity:
    Figure FDA0002297981080000021
  2. 2. the balance car system measurement and control method based on camera pavement detection according to claim 1, characterized in that: in the process 1, the specific steps are as follows:
    (1) the system incorporating discrete control processes is described by linear machine differential equations:
    X(k|k-1)＝AX(k-1|k-1)+BU(k) (1)
    according to the equation, substituting the current time measured value of the gyroscope and adding the drift amount of the gyroscope to obtain the predicted value Angle of the current Angle, and writing the predicted value Angle into a matrix form is as follows:
    Figure FDA0002297981080000022
    (2) in the prediction of the covariance matrix, two input values need to be defined in the calculation process, which are respectively: drift noise of the gyro sensor and angle noise of the accelerometer sensor,
    P(k|k-1)＝AP(k-1|k-1)A’+Q (3)
    wherein Q in the formula is a vector
    Figure FDA0002297981080000023
    The covariance matrix of (a), namely:
    Figure FDA0002297981080000024
    since the drift noise and also the angle noise are independent from each other;
    cov(Q_bias,Angel)＝0 (5)
    cov(Angel,Q_bias)＝0 (6)
    the defined process angle noise covariance parameter and the measured noise covariance parameter in the program are respectively:
    Q_angle＝0.001；Q_gyro＝0.003；R_angle＝0.5；
    (3) calculating a gain factor of the Kalman filter, which is a two-dimensional vector of gain factors set to
    Figure FDA0002297981080000031
    For the correction of the angle and the angular velocity, the expression is:
    Kg(k)＝P(k|k-1)H’/(HP(k|k-1)H’+R) (7)
    the constants R appearing in the equation refer to the noise figure of the acceleration measurement, and the figure in the program is initially defined as R _ angle ═ 0.5;
    (4) correcting a predicted value by using a Kalman gain coefficient, taking a difference value of angles obtained by an accelerometer and a gyroscope as an error value Angle _ err, and correcting the predicted value by using a product of the Kalman gain coefficient and the error, wherein the formula is as follows:
    X(k|k)＝X(k|k-1)+Kg(k)(Z(k)-HX(k|k-1)) (8)
    updating the prediction error value and the angle value simultaneously:
    Angel+＝K_0*Angel_err (9)
    Q_Bias+＝K_1*Angel_err (10)
    (5) the matrix covariance P matrix is updated this time, and is mainly used for predicting covariance in the following iterations, and the equation is as follows:
    P(k|k-1)＝(I-Kg(k)H)P(k|k-1) (11)
    (6) measuring the angle X at the moment k(k|k-1) differences are made with the angle predicted value at the k moment, and angles are counted in every 20 angle measurementsThe degree average error is measured on average,
    Figure FDA0002297981080000041
  3. 3. the balance car system measurement and control method based on camera road surface detection according to claim 1 or 2, characterized in that: in the process 2, the specific steps are as follows:
    (1) performing Kalman prediction fusion by using gyroscope to measure and calculate Z-axis acceleration information and accelerometer to measure and calculate X-axis angle information to obtain posterior estimation values theta converging to accurate valuesr；
    (2) The angle Roll measured on the X axis is used for solving the sine value sin theta of the angle RollrAnd the sine value is multiplied by the wheel base length D between the left wheel and the right wheel of the vehicle body to obtain a value h which can be expressed as the height of the center of mass of the vehicle body supported by the uneven road surface,
    h＝D*sinθr；
    (3) dividing the height value h of the vehicle body raised by the uneven road surface by the height 2d of the chassis of the vehicle body to obtain the tilt rate Q of the whole vehicle, finally representing the Q value as the tilt degree of the road surface to the vehicle body of the balance vehicle, and when the value is 1, achieving the ultimate safety state that the vehicle body can bear the maximum tilt, and achieving the effect of returning to , wherein the formula is as follows:
    Q＝D*Sinθr/2d (13)。
  4. 4. the balance car system measurement and control method based on camera road surface detection according to claim 1 or 2, characterized in that: in the process 3, the specific steps are as follows: (1) collecting road surface picture digital signal a by utilizing COMS camera(t)Extracting LBP characteristics and carrying out statistics, the steps are as follows:
    1) converting the RGB color image into a gray-scale image;
    Figure FDA0002297981080000042
    2) optimizing the conversion of a floating-point operation structure;
    Figure FDA0002297981080000043
    3) dividing the image digital signal into 1 × 10-10 sub-regions by dividing the image into equal vertical segments; then the pixel resolution for each sub-region is 640 x 48;
    Figure FDA0002297981080000051
    (2) local binarization; a local dynamic threshold value method is selected for binaryzation, so that the accurate extraction of the target texture is realized; the specific extraction steps are as follows:
    1) in the image traversal process, 3 x 3 gray windows are extracted each time as objects, and the central gray value of the window is selected as the threshold value a of the current binarization(t)(i,j)Respectively connecting 8 adjacent pixel points with the threshold value a(t)(i,j)Comparing, if the value is larger than the threshold value, assigning 1, otherwise, assigning 0;
    2) numbering from the top left corner results in binary digits of 8 bits, followed by bi-1，j-1bi-1，jbi-1，j+1bi，j+1bi+1，j+1bi+1，jbi+1，j-1bi，j-1](ii) a The binary number of 8 bits is the LBP value of the pixel point of the center window;
    LBP(t)(i,j)＝sum[bi-1,j-1,bi-1,j,bi-1,j+1,bi,j-1,bi,j,bi-1,j+1,bi+1,j-1,bi+1,j,bi+1,j+1]
    3) listing all eight 8-bit binary rotation arrangement modes, and selecting the minimum LBP as the LBP value of a central window pixel point to represent the texture information of the points;
    Figure FDA0002297981080000053
    Figure FDA0002297981080000054
    (3) counting LBP histogram gradient distribution statistics, and making a pavement damage degree estimation value into , wherein the specific steps are as follows;
    1) performing LBP histogram statistics on each subregion map, dividing the distribution histograms from 0 to 255 into 16 intervals for statistics, and counting the number of pixel points in each interval to account for the global proportion to obtain set arrays R containing 16 elementsnm(1 ≦ n ≦ 10; 1 ≦ m ≦ 16), and new sequences R generated by operating the absolute difference of these 16 elementsnk(1 ≦ n ≦ 10; 1 ≦ k ≦ 8), thereby characterizing the LBP histogram probability density distribution gradient;
    Rn,2*k＝Rn,2*k-Rn,2*k-1
    2) counting a summation Cn of an array Rnm thereof (1 ≦ n ≦ 10);
    Figure 1
    Cnthe minimum value is 0 when the interval is fixed, namely the distribution of the texture of the road surface is minimum, the meaning of the distribution is that the road surface is in the most flat state, otherwise, the C isnThe value is closer to 1, namely the texture contrast is the maximum, which indicates that the road surface is seriously damaged.
  5. 5. The balance car system measurement and control method based on camera road surface detection according to claim 1 or 2, characterized in that: in the process 5, the specific steps are as follows: method for determining weight factor K of features by using empirical debugging method1，K2，K3The values of the three statistical characteristics are weighted and summed, and three parameters are adjusted until the balance of the vehicle body is optimal, wherein the parameter adjustment rule is as follows:
    Figure FDA0002297981080000062
  6. 6. the balance car system measurement and control method based on camera road surface detection according to claim 1 or 2, characterized in that: in the process 5, the specific steps are as follows: according to the measured C under different road surfacesnQ and PiPerforming least square fitting under matlab to perform parameter setting according to the noise variance estimation value I corresponding to the value, fitting an appropriate -order function of a four-dimensional space, and obtaining a parameter K1，K2And K3The function is as follows:
    Figure FDA0002297981080000063
  7. 7. the balance car system measurement and control method based on camera road surface detection according to claim 1 or 2, characterized in that: in the process 6, the specific steps are as follows:
    (1) referring to the implementation process of the kalman filter in process 1, the defined process noise covariance parameters in the program are:
    Q_angle＝0.001；Q_gyro＝0.003；
    (2) the noise variance estimation value I introduced by the road surface is used as a result and is output to a measuring and calculating system of the balance car to update R _ angle in real time, the Kalman filter interface is adaptively changed, Q _ angle is 0.001, and Q _ gyro value is unchanged, and the formula is as follows:
    R_angel＝I；
    (3) and (3) obtaining a roll angle as a final inclination angle of the balance car according to the complete calculation process in the process 1, sending the roll angle into a motor controller for calculation, and obtaining and executing motor control quantity.
  8. 8. The balance car system measurement and control method based on camera road surface detection according to claim 1 or 2, characterized in that: in the process 7, the specific steps are as follows:
    (1) calculating a noise variance estimation value I of the road surface bumping degree, storing and finally using the noise variance estimation value I as a decision basis of a motor control strategy;
    (2) when I is(t)When the signal intensity is less than I, the system considers that the road surface is relatively flat, and the measuring signal of the sensor is stable and has less noise; the fuzzy PID control is selected, the operation quantity provided by a user is adapted according to the model with better signal following capability and faster response capability of the system, and the fuzzy PID also has the advantages of accurate control quantity execution and simple operation, and the specific formula is as follows:
    the parameter adjustment equation is as follows:
    Kp＝Kp0+Δkp
    Ki＝Ki0+Δki
    Kd＝Kd0+Δkd
    (3) when I is(t)When the signal is larger than I, the system considers that the road surface is bumpy, and the measuring signal of the sensor contains large non-artificial jitter and much noise; the cascade PID control is selected, compared with a single-ring PID, the cascade PID has the advantages that the structure is simple, the control is accurate, the anti-interference performance of the system is enhanced, the integral terms are too many due to the structural diversity, the control lag of the system is caused, the cascade PID control is just suitable for a bumpy road surface, and the system is required to have a proper slow point for a peak burr caused by the bumpy road surface, so that the safe operation of a user is facilitated;
    after control cycles are completed, wait for the system clock to arrive, and cyclically repeat process 1 to process 7.
CN201710515182.XA 2017-06-29 2017-06-29 Balance car system measurement and control method based on camera road surface detection Active CN107340298B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710515182.XA CN107340298B (en) 2017-06-29 2017-06-29 Balance car system measurement and control method based on camera road surface detection

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710515182.XA CN107340298B (en) 2017-06-29 2017-06-29 Balance car system measurement and control method based on camera road surface detection

Publications (2)

Publication Number Publication Date
CN107340298A CN107340298A (en) 2017-11-10
CN107340298B true CN107340298B (en) 2020-01-31

Family

ID=60218102

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710515182.XA Active CN107340298B (en) 2017-06-29 2017-06-29 Balance car system measurement and control method based on camera road surface detection

Country Status (1)

Country Link
CN (1) CN107340298B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108973576A (en) * 2018-08-12 2018-12-11 苏州青科艾莉电子科技有限公司 A kind of anti-rollover regulation method of road vehicle high-performance electronic control air suspension
CN109507871B (en) * 2018-12-11 2022-03-25 广东工业大学 PID parameter setting method and product for two-wheel balance vehicle body balance control
CN110554706A (en) * 2019-09-25 2019-12-10 江苏理工学院 visual navigation self-balancing vehicle and balancing method
CN114495068B (en) * 2022-04-18 2022-07-08 河北工业大学 Pavement health detection method based on human-computer interaction and deep learning

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001247296A (en) * 2000-03-08 2001-09-11 Komatsu Forklift Co Ltd Lpg cylinder bracket also used as truck
JP2003165431A (en) * 2001-11-30 2003-06-10 Honda Motor Co Ltd Body slip angle estimating method
CN101950156A (en) * 2010-09-06 2011-01-19 重庆大学 Adaptive cascade PID control method
CN102538781A (en) * 2011-12-14 2012-07-04 浙江大学 Machine vision and inertial navigation fusion-based mobile robot motion attitude estimation method
CN104657724A (en) * 2015-03-12 2015-05-27 福建依图网络科技有限公司 Method for detecting pedestrians in traffic videos
CN204423150U (en) * 2015-02-05 2015-06-24 重庆交通大学 Based on the two-wheeled control system of balance car of image
CN104932505A (en) * 2015-06-08 2015-09-23 华南理工大学 Automatic navigation system based on camera detection, control method of system and vertical balance car
CN106054931A (en) * 2016-07-29 2016-10-26 北方工业大学 Unmanned aerial vehicle fixed-point flight control system based on visual positioning
CN106353782A (en) * 2015-07-16 2017-01-25 深圳市华信天线技术有限公司 Ground point coordinates measuring method and device

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001247296A (en) * 2000-03-08 2001-09-11 Komatsu Forklift Co Ltd Lpg cylinder bracket also used as truck
JP2003165431A (en) * 2001-11-30 2003-06-10 Honda Motor Co Ltd Body slip angle estimating method
CN101950156A (en) * 2010-09-06 2011-01-19 重庆大学 Adaptive cascade PID control method
CN102538781A (en) * 2011-12-14 2012-07-04 浙江大学 Machine vision and inertial navigation fusion-based mobile robot motion attitude estimation method
CN204423150U (en) * 2015-02-05 2015-06-24 重庆交通大学 Based on the two-wheeled control system of balance car of image
CN104657724A (en) * 2015-03-12 2015-05-27 福建依图网络科技有限公司 Method for detecting pedestrians in traffic videos
CN104932505A (en) * 2015-06-08 2015-09-23 华南理工大学 Automatic navigation system based on camera detection, control method of system and vertical balance car
CN106353782A (en) * 2015-07-16 2017-01-25 深圳市华信天线技术有限公司 Ground point coordinates measuring method and device
CN106054931A (en) * 2016-07-29 2016-10-26 北方工业大学 Unmanned aerial vehicle fixed-point flight control system based on visual positioning

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
"Attitude measure system based on extended Kalman filter";Zhang Tiemin，et.al.;《Computers and Electronics in Agriculture》;20170119(第134期);第19-26页 *
"基于图像显著性的路面裂缝检测";徐威等;《中国图象图形学报》;20130131;第18卷(第1期);第69-77页 *

Also Published As

Publication number Publication date
CN107340298A (en) 2017-11-10

Similar Documents

Publication Publication Date Title
CN107340298B (en) Balance car system measurement and control method based on camera road surface detection
CN107230218B (en) Method and apparatus for generating confidence measures for estimates derived from images captured by vehicle-mounted cameras
US9797981B2 (en) Moving-object position/attitude estimation apparatus and moving-object position/attitude estimation method
CN104820996B (en) A kind of method for tracking target of the adaptive piecemeal based on video
EP2126843A2 (en) Method and system for video-based road lane curvature measurement
CN107850446A (en) Self-position estimating device and self-position presumption method
JPH0734162B2 (en) Analogical control method
CN110223317B (en) Moving target detection and track prediction method based on image processing
CN106128121B (en) Vehicle queue length fast algorithm of detecting based on Local Features Analysis
CN110765909B (en) Road surface estimation method based on vehicle-mounted camera auxiliary distributed driving electric vehicle
CN114612795A (en) Laser radar point cloud-based road surface scene target identification method
JP7173471B2 (en) 3D position estimation device and program
CN113848545B (en) Fusion target detection and tracking method based on vision and millimeter wave radar
CN113256679A (en) Electronic image stabilization algorithm based on vehicle-mounted rearview mirror system
CN113076988B (en) Mobile robot vision SLAM key frame self-adaptive screening method based on neural network
CN113221739B (en) Monocular vision-based vehicle distance measuring method
CN116935199B (en) Intelligent detection method and system for levelness of highway construction
EP3624064A1 (en) Vehicle-mounted environment recognition device
CN115035000B (en) Road dust image identification method and system
CN111207699A (en) Method and system for dynamically measuring tire corner in visual mode
CN113566828A (en) Impact-resistant scanning matching method and system based on multi-sensor decision fusion
CN115240170A (en) Road pedestrian detection and tracking method and system based on event camera
CN113643355A (en) Method and system for detecting position and orientation of target vehicle and storage medium
JP2003256854A (en) Apparatus for recognizing road white line
JP2551324B2 (en) Method and device for recognizing environment of moving vehicle

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant