Intelligent automobile seat adjusting method capable of sensing postures of driver and passengers

Abstract

The invention relates to an intelligent automobile seat adjusting method capable of sensing postures of a driver and passengers, which comprises the following steps of: acquiring a left view and a right view of the driver and passengers in an automobile on site through a binocular camera to calculate three-dimensional coordinates of key points of each part of the driver and passengers; and then the target joint angles of the driver and the passengers and the automobile seat adjusting parameters are calculated through the automobile seat adjusting model, so that the automobile seat is suitable for the drivers and the passengers of various body types, and the driving experience of the drivers and the passengers is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent adjustment of automobile seats, and in particular to an automatic adjustment method of an intelligent automobile seat based on driver and passenger posture perception. Background Art

[0002] With the continuous development of economy and the increase of personal income, people's demand for cars is rising. The production and ownership of cars are growing at a steady pace. As an indispensable part of cars, car seats are not only the key facilities to provide a comfortable riding experience, but also the requirements for comfort, environmental protection, lightweight, intelligence and safety are also increasing to meet consumers' pursuit of high-quality life.

[0003] As a key component of a car, the design level of car seats is directly related to the safety and comfort experience of drivers and passengers. At present, car seats are mainly divided into two types: independent and one-piece, and are changing from traditional manual seats to electric seats. Modern seats are generally equipped with height adjustment motors, slide motors, seat control units, ventilation and heating, and memory functions. However, the existing seat design does not fully meet the needs of future smart cars. With the simplification of smart car functions, seats will have more space. Therefore, improving the user's riding experience in smart cars has become a topic that needs to be solved urgently.

[0004] People of different body types have different needs for seat space. If the same space size is provided for people of all body types, the seat can only meet the comfort of some users, but cannot fully take into account the needs of other users. There are deficiencies in the personalization of the seat, and the adaptability of the person to the seat is poor. In addition, the sitting habits of drivers and passengers are different. At present, most seats require manual operation to adjust the angle to meet the needs, and cannot achieve autonomous adjustment, and the adjustment accuracy is low.

[0005] Therefore, a method for automatic adjustment of an intelligent automobile seat based on genetic driver and passenger posture perception is provided to solve the problem that existing automobile seats have poor adaptability to people of different body shapes.

[0006] At present, there are few solutions to the problem of intelligent car seat systems and methods for sensing the posture of drivers and passengers and automatically adjusting. It is necessary to solve the problem that existing car seats have poor adaptability to people of different body shapes. Summary of the invention

[0007] The technical problem to be solved by the present invention is how to improve the adaptability of automobile seats to people of different body shapes and to perform precise adjustments according to the body shapes of drivers and passengers.

[0008] The present invention provides an intelligent adjustment method for a car seat for sensing the posture of a driver and passenger, comprising: Step 1: The driver and passengers sit in the car seat and make preliminary adjustments; Step 2: Use binocular camera to collect A checkerboard image of a driver or passenger holding a checkerboard calibration plate; Step 3, perform checkerboard calibration, checkerboard corner point detection, binocular camera internal parameter matrix and external parameter matrix calculation, and set distortion coefficient on the checkerboard image; Step 4, calculating the reprojection error of the checkerboard corner points based on the checkerboard corner points and the distortion coefficients, and selecting the internal parameter matrix and the external parameter matrix corresponding to the checkerboard image with the smallest average reprojection error as the optimal internal parameter matrix and the optimal external parameter matrix; Step 5: adjust the relative position and posture of the binocular camera based on the optimal external parameter matrix; and collect the real scene inside the car through the adjusted binocular camera to obtain the left view and the right view; Step 6, performing distortion correction, stereo correction and remapping operations on the left view and the right view; Step 7, calculating the depth information of the vehicle occupants and the vehicle seats based on the remapped left view and right view; Step 8, constructing a human body posture model, and calculating the three-dimensional coordinates of key points of various parts of the driver and passenger in combination with the depth information of the driver and passenger; Step 9, constructing a car seat adjustment model, inputting the three-dimensional coordinates of key points of various parts of the driver and passenger, and outputting the target joint angles of the driver and passenger and the car seat adjustment parameters; Step 10, setting the priority of the car seat adjustment, and adjusting the car seat based on the priority and the car seat adjustment parameters.

[0009] Compared with the prior art, the present application has the following advantages: the left view and right view of the driver and passenger in the car are collected on-site by a binocular camera to calculate the three-dimensional coordinates of the key points of each part of the driver and passenger, and then the target joint angles and car seat adjustment parameters of the driver and passenger are calculated through the car seat adjustment model, so that the car seat can adapt to drivers and passengers of various body shapes and improve the driving experience of the driver and passenger.

[0010] In a possible implementation manner, the step 1 specifically includes: Step 101, a pressure sensor is arranged on the car seat, the binocular camera is arranged at the intersection of the car roof and the A-pillar, and a millimeter wave radar is arranged on the car steering wheel; Step 102: The driver and passenger sit on the car seat, and the weight parameter, height parameter, gender parameter, and distance parameter between the driver and passenger and the steering wheel are obtained according to the pressure sensor, binocular camera, and millimeter wave radar; Step 103, constructing an initial height adjustment model of the car seat: ; In the formula, Indicates the initial height of the car seat; Indicates the weight parameter of the driver and passengers; and are coefficients that vary in direct proportion to the weight parameter; the initial height of the car seat is preliminarily adjusted according to the weight parameter; Step 104, constructing a car seat back tilt angle adjustment model based on the sex parameter sex of the driver and passenger: ; In the formula, Represents the height parameter of male, Represents the height parameter of female; and represents the coefficient that varies in direct proportion to the height parameter of males; and represents the coefficient that changes in direct proportion to the height parameter of women; Indicates the tilt angle of the car seat back for male occupants; Indicates the tilt angle of the car seat back when the driver and passenger are female; preliminarily adjusts the tilt angle of the car seat back according to the gender parameter and the height parameter; Step 105, constructing a car seat position adjustment model based on the distance parameter between the driver and the steering wheel: ; In the formula, Indicates the amount of adjustment of the car seat position; Indicates the distance parameter between the driver and the steering wheel; Indicates the preset safety distance threshold; Represents the proportional coefficient; calculates the distance parameter between the driver and the steering wheel, calculates the car seat position adjustment amount and preliminarily adjusts the position of the car seat.

[0011] In a possible implementation manner, the step 2 specifically includes: Step 201, setting the parameters of the binocular camera to adjust the resolution and frame rate of the binocular camera; Step 202, selecting a checkerboard calibration plate according to the resolution and frame rate of the binocular camera; Step 203: The driver or passenger holds a checkerboard calibration plate in the field of view of the binocular camera and uses the binocular camera to capture the checkerboard calibration plate by changing the posture and distance of the checkerboard calibration plate. Checkerboard image.

[0012] In a possible implementation manner, the step 3 specifically includes: Step 301, import all chessboard images into MATLAB software; Step 302: setting the distortion coefficient according to the distortion characteristics of the binocular camera, wherein the distortion coefficient includes the radial distortion coefficient. and tangential distortion coefficient ; Step 303, using MATLAB software to detect all the checkerboard corner points in each checkerboard image, and calculate the internal parameter matrix and external parameter matrix of the binocular camera corresponding to each checkerboard image, wherein the internal parameter matrix includes the focal length of the binocular camera , principal point coordinates , radial distortion coefficient and tangential distortion coefficient ; The external parameter matrix includes a rotation matrix for describing the rotation direction of the binocular camera And the translation vector T used to describe the translation distance of the binocular camera.

[0013] In a possible implementation manner, step 4 specifically includes: Step 401, calculating the reprojection error of each checkerboard corner point based on the checkerboard corner points in each checkerboard image and the set distortion coefficient by using MATLAB software; Step 402, calculating the average reprojection error of the checkerboard corner points in each checkerboard image; Step 403 , selecting the checkerboard image with the smallest average reprojection error as the best calibration object, and taking the internal parameter matrix and the external parameter matrix of the binocular camera corresponding to the best calibration object as the best internal parameter matrix and the best external parameter matrix.

[0014] In a possible implementation manner, the distortion correction of the left view and the right view in step 6 specifically includes: Step 601A, combining the best internal parameter matrix and the distortion correction matrix to calculate and generate the distortion correction parameters, the distortion correction matrix expression is: ; In the formula, Represents the coordinates of the pixel points on the left view and the right view; Represents the coordinates of the pixel after distortion correction; Indicates the lens distortion of the binocular camera Distortion value in direction; Indicates the distortion value of the binocular camera's lens distortion in the Y direction; Step 602A, using the undistort Image function in MATLAB software to perform distortion correction on the pixels in the left view and the right view according to the distortion correction parameters and the optimal internal parameter matrix of the binocular camera to eliminate the influence of lens distortion. The distortion correction model is: ; In the formula, Indicates the distance from the pixel point on the left view and right view image plane to the center point of the image plane; ; represents the radial distortion coefficient; represents the tangential distortion coefficient; Indicates the coordinates of the pixel before distortion correction; Represents the coordinates of the pixel after distortion correction; The stereoscopic correction of the left view and the right view in step 6 specifically includes: Step 601B: The pixel coordinates after distortion correction are To perform coordinate transformation, the transformation formula is: ; In the formula, Indicates the translation amount of the binocular camera in three-dimensional space based on the translation vector. represents the scaling factor; Thus, we can get multiple sets of corresponding coordinates in the left view and the right view. and ; Step 602B, coordinate and Transform to the camera coordinate system through the optimal intrinsic parameter matrix: ; In the formula, and Represents the coordinates of corresponding points in the left view and the right view in the camera coordinate system; represents the optimal intrinsic parameter matrix of the left camera; represents the optimal intrinsic parameter matrix of the right camera; according to the epipolar relationship, there is an essential matrix , so that any pair of corresponding points in the left view and the right view and satisfy: , Represents the coordinate vector of the pixel point in the right view The transpose of Step 603B, based on , bring it into Get ; Solved by multiple groups of corresponding pixel points ; Used to adjust the essential matrix in epipolar relation ; Step 604B, using singular value decomposition method to solve the essential matrix ; The essential matrix Decompose into ; Represents the translation vector The opposition formed into a matrix, is the rotation matrix; The essential matrix Perform singular value decomposition into ; represents an orthogonal matrix, represents a diagonal matrix, , ; Based on the above decomposition, the rotation matrix of the left camera in the binocular camera is obtained and the rotation matrix of the right camera ; and the translation vector , , where Represents an orthogonal matrix The third column of Step 605B, based on the geometric principle of camera imaging and the rotation matrix and the translation vector Construct the left camera stereo rectification transformation matrix separately and the right camera stereo rectification transformation matrix , where the left camera stereo rectification transformation matrix is The formula is: ; In the formula, Indicates the coordinates of the principal point of the left view. represents the focal length of the left camera, Indicates that the left camera is Translation matrix in direction; Camera Stereo Rectification Transformation Matrix The formula is: ; In the formula, Indicates the coordinates of the principal point of the right view; represents the focal length of the right camera, Indicates that the right camera is Translation matrix in direction; Step 606B, using the left camera stereo correction transformation matrix and the right camera stereo rectification transformation matrix Perform stereo correction operations on the distortion-corrected left and right views respectively; The remapping operation of the left view and the right view after stereoscopic correction in step 6 specifically includes: Take the left camera stereo rectification transformation matrix and the right camera stereo rectification transformation matrix As the key parameter, the remapping matrix is ​​generated through the cv::initUndistortRectifyMap function, and the cv::remap function is used to remap each pixel of the left view and the right view.

[0015] In a possible implementation manner, the step 7 specifically includes: Step 701, using a deep learning-based image segmentation algorithm to segment the foreground and background in the remapped left view and right view; Step 702, using a stereo matching method, searching for the most similar pixel blocks of the driver and the car seat in the foreground of the left view on the corresponding epipolar lines of the right view through an SSD algorithm, thereby generating a disparity map; Step 703, using a median filter algorithm to process the disparity map, remove noise interference, and smooth the disparity data; Step 704: Calculate the depth estimation value of the driver and the car seat in the foreground based on the disparity map ; The calculation formula is: ; Indicates the focal length of the binocular camera; Indicates the baseline distance of the binocular camera; Indicates the disparity between corresponding points in the left and right views.

[0016] In a possible implementation manner, step 8 specifically includes: Step 801, using a Kinect V2 device to collect key points of various parts of the driver and passenger on the car seat and number them, the key points of various parts of the driver and passenger include: the focus of the head, the neck and the left and right shoulders, the right shoulder, the right elbow, the right wrist, the left shoulder, the left elbow, the left wrist, the right hip point, the right knee, the right ankle, the right foot, the left hip point, the left knee, the left ankle, and the left foot; Step 802: Calculate the depth estimate based on the lens distortion characteristics of the binocular camera, the thickness of the driver's clothing, and the lighting environment. The error rate with the real depth distance is corrected to obtain accurate depth information of the driver and the car seat; Step 803, constructing a human body posture model, and calculating the three-dimensional coordinates of key points of various parts of the human body based on the depth information. The human body posture model is: ; ; In the formula, Respectively represent the focal length of the binocular camera lens in the x-direction and the y-direction; Represents the coordinates of the principal point of the binocular camera in the pixel coordinate system; and Respectively represent the pixel coordinates of the left view and the right view; and ; Represents depth information.

[0017] In a possible implementation manner, step 9 specifically includes: Step 901, based on the three-dimensional coordinates of the key points of each part of the driver and passenger and the matrix formula of the size of each part, the size of each part of the driver and passenger is calculated; including the torso L1, the height from eyes to shoulders L2, the length of the left upper arm L3, the length of the right upper arm L4, the length of the left forearm L5, the length of the right forearm L6, the length of the left thigh L7, the length of the right thigh L8, the length of the left calf L9, the length of the right calf L10, the length of the left foot L11, the length of the right foot L12, and the height of the eyes L3; The matrix formula for each part size is: ; and Represents the three-dimensional coordinates of key points of adjacent parts; Step 902, constructing a car seat adjustment model, inputting the three-dimensional coordinates of key points of each part, and outputting the target joint angle and car seat adjustment parameters; the car seat adjustment model is: In the formula, Indicates the initial angle of the joint angles of each part of the driver and passenger; represents the target joint angle; represents the square of the length of one of the limb segments adjacent to the joint, represents the square of the length of the other limb segment adjacent to the joint, Indicates The comfortable angle value of each joint, represents the total number of joint angles; Indicates the adjustment parameters of the car seat, including the car seat back tilt angle, car seat height, and car sliding adjustment distance; Indicates the calibration scale factor, including the car seat back inclination factor , Car seat height coefficient and vehicle slip adjustment coefficient ; Indicates offset, including car seat offset , Car seat height offset and car slide adjustment offset .

[0018] In a possible implementation manner, the priority set in step 10 includes: Step 1001, preset the optimal height threshold of the driver's eyes, take the set optimal height threshold as the starting point, and calculate the coordinates of the preset hip point in combination with the driver's trunk tilt comfort angle value, the relative distance between the steering wheel and the body, and the trunk length ; The calculation formula is: ; ; ; In the formula, Indicates the length of the car seat, Indicates the width of the car seat, Indicates the height of the car seat, Indicates the leg length of the driver and passenger. Indicates the tilt angle of the car seat back; Step 1002, setting the position of the car seat as the first priority; adjusting the position of the car seat based on the preset hip point; Step 1003, setting the sliding adjustment distance and lifting height of the car seat as the second priority; based on the preset hip point and the fixed heel point, the hip point is moved twice by using the cosine theorem combined with the comfortable angle value of the knee joint and the length of the thigh and calf to obtain the coordinates of the ideal hip point, and the coordinate values ​​of the ideal hip point and the preset hip point are compared to obtain the sliding adjustment distance and height lifting value of the car seat; Step 1004, setting the tilt angle of the car seat back as the third priority; adjusting the angle of the car seat back based on the comfortable tilt angle of the driver's body and the initial value of the driver's body tilt. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a schematic diagram of the process of the present invention; Figure 2 The human body posture model diagram constructed by the present invention. DETAILED DESCRIPTION

[0020] First, those skilled in the art should understand that these implementations are only used to explain the technical principles of the embodiments of the present application, and are not intended to limit the protection scope of the embodiments of the present application. Those skilled in the art can make adjustments to them as needed to adapt to specific application scenarios.

[0021] In the description of the embodiments of the present application, it should be noted that, unless otherwise clearly specified and limited, the terms "connected" and "connection" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms in the embodiments of the present application can be understood according to specific circumstances.

[0022] In the embodiments of the present application, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may mean that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. Moreover, a first feature being "above", "above" or "above" a second feature may mean that the first feature is directly above or obliquely above the second feature, or simply means that the first feature is higher in level than the second feature. A first feature being "below", "below" or "below" a second feature may mean that the first feature is directly below or obliquely below the second feature, or simply means that the first feature is lower in level than the second feature.

[0023] The present application is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0024] See also Figure 1 , Figure 2 As shown, the embodiment of the present application discloses an intelligent adjustment method for a car seat for sensing the posture of a driver and passenger, which is applied to intelligently adjust the car seat in a smart car. The intelligent adjustment method for the car seat includes: Step 1: The driver and passengers sit in the car seat and make preliminary adjustments; specifically: Step 101, a pressure sensor is arranged on the car seat to collect the weight parameter of the driver and the passenger, the binocular camera is arranged at the intersection of the car roof and the A-pillar to collect the height parameter and gender parameter of the driver and the passenger, and a millimeter wave radar is arranged on the car steering wheel to collect the distance parameter between the driver and the steering wheel; Step 102: The driver and passenger sit on the car seat, and the weight parameter, height parameter, gender parameter, and distance parameter between the driver and passenger and the steering wheel are obtained according to the pressure sensor, binocular camera, and millimeter wave radar; Step 103, constructing an initial height adjustment model of the car seat: ; In the formula, Indicates the initial height of the car seat; Indicates the weight parameter of the driver and passengers; and are coefficients that vary in direct proportion to the weight parameter; the initial height of the car seat is preliminarily adjusted according to the weight parameter; in this specific embodiment, if the driver and passenger are light in weight, the initial height of the car seat is preliminarily set to be low, so as to facilitate the driver and passenger to get on or off the car; if the driver and passenger are heavy in weight, the load-bearing mode of the car seat is increased; Step 104, constructing a car seat back tilt angle adjustment model based on the sex parameter sex of the driver and passenger: ; In the formula, Represents the height parameter of male, Represents the height parameter of female; and represents the coefficient that varies in direct proportion to the height parameter of males; and represents the coefficient that changes in direct proportion to the height parameter of women; Indicates the tilt angle of the car seat back for male occupants; Indicates the tilt angle of the car seat back when the driver and passenger are female; preliminarily adjusts the tilt angle of the car seat back according to the gender parameter and the height parameter; for example, for a tall male, preliminarily adjusts the tilt angle of the car seat back to a 45° reclining position, and lengthens the leg support length of the car seat, that is, adjusts the car seat to slide backward; for a short female, adjusts the car seat back to the default vertical position, and adjusts the car seat to slide forward; Step 105, constructing a car seat position adjustment model based on the distance parameter between the driver and the steering wheel: ; In the formula, Indicates the amount of adjustment of the car seat position; Indicates the distance parameter between the driver and the steering wheel; Indicates the preset safety distance threshold; Indicates the proportional coefficient; calculates the distance parameter between the driver and the steering wheel, calculates the adjustment amount of the car seat position and preliminarily adjusts the position of the car seat; In addition, during driving, the system continuously monitors the parameter data of pressure sensors, binocular cameras and millimeter-wave radars, and dynamically adjusts the weight of each sensor data according to the vehicle's driving status (such as acceleration, braking, turning) and the real-time posture changes of the driver and passengers; The first is real-time sensor data collection and fusion: input pressure sensor data to monitor the force distribution of the seat and determine whether the center of gravity of the driver and passenger is offset; input binocular camera data to calculate the three-dimensional distance between the driver and passenger's torso, head and steering wheel in real time; input millimeter-wave radar data to accurately measure the straight-line distance between the driver and passenger and the steering wheel to supplement the visual data; in normal driving conditions, the S vision of the binocular camera is the main, and the S radar of the millimeter-wave radar is the auxiliary, and the final S is weightedly calculated: and Represent the weights of the binocular camera and millimeter-wave radar measurements respectively; Then the safety threshold S 0 Dynamic correction, such as increasing S when driving at high speed 0 Reserve more safety redundancy, automatically reduce S when driving fatigue 0 In order to shorten the reaction distance, the last is priority linkage. If the seat position adjustment (first priority) conflicts with the height adjustment (second priority) (such as the seat moving back resulting in insufficient height), the position adjustment is prioritized and the height compensation is triggered. The calculation formula is as follows: Therefore, the mathematical representation of dynamic adjustment is: D( t )= g ( t ) • [S( t )− S 0 ( t )] Where D(t) is the seat sliding adjustment amount, which means the forward and backward sliding distance that the car seat needs to be adjusted at time t, and g(t) is the dynamic proportional coefficient, which means the measured distance difference S(t)−S 0 (t) is the proportional factor converted into the actual adjustment amount, S(t) is the real-time fusion distance, which means the real-time distance between the driver and the steering wheel obtained by fusion of binocular camera and millimeter-wave radar data, S 0 (t) is the dynamic safety distance threshold, which represents the safety distance target value corrected in real time according to the driving status.

[0025] Step 2: Use binocular camera to collect A chessboard image of a driver or passenger holding a chessboard calibration plate; specifically including: Step 201, setting the parameters of the binocular camera to adjust the resolution and frame rate of the binocular camera; and adjusting the parameters of the binocular camera to keep the parameters consistent during the shooting process; the parameters of the binocular camera include exposure time, white balance and focus; this specific embodiment adopts a uniform light environment to collect chessboard images to avoid direct strong light or heavy shadows; and uses a tripod or other stable equipment to fix the binocular camera to ensure the quality of subsequent chessboard image collection; Step 202, select a checkerboard calibration plate according to the resolution and frame rate of the binocular camera; determine that the black and white squares on the checkerboard calibration plate have a high contrast; select a binocular camera with a high resolution Square; low-resolution stereo camera selection Grid; This specific embodiment selects Checkered checkerboard calibration board; Step 203: The driver or passenger holds a checkerboard calibration plate in the field of view of the binocular camera, and changes the posture (including translation, rotation, and tilt of the checkerboard calibration plate) and distance of the checkerboard calibration plate to obtain a picture with the binocular camera. Checkerboard images are collected to ensure that the collected checkerboard images are clear and visible, and the checkerboard corners are clear.

[0026] Step 3, perform checkerboard calibration on the checkerboard image, checkerboard corner point detection, calculate the stereo camera internal parameter matrix and external parameter matrix, and set the distortion coefficient; specifically including: Step 301, import all checkerboard images into the Stereo Camera Calibrator toolbox of MATLAB software; Step 302: setting the distortion coefficient according to the distortion characteristics of the binocular camera, wherein the distortion coefficient includes the radial distortion coefficient. and tangential distortion coefficient ; This specific embodiment sets two radial distortion coefficients and , when the binocular camera lens has tangential distortion, set two tangential distortion coefficients and ; Step 303, click the Detect Checkerboard Points button in the MATLAB software, and the MATLAB software automatically detects all checkerboard corner points in each checkerboard image; the user checks the detection results, and if there are checkerboard corner points that are not detected, manually adjust the checkerboard image or re-collect the checkerboard image; Next, click the Calibrate button in the MATLAB software, and the MATLAB software will start calculating the intrinsic and extrinsic matrix of the binocular camera corresponding to each checkerboard image; the intrinsic matrix includes the focal length of the binocular camera , principal point coordinates , radial distortion coefficient and tangential distortion coefficient ; The external parameter matrix includes a rotation matrix for describing the rotation direction of the binocular camera And the translation vector T used to describe the translation distance of the binocular camera.

[0027] Step 4, calculating the reprojection error of the checkerboard corner points based on the checkerboard corner points and the distortion coefficients, and selecting the internal parameter matrix and the external parameter matrix corresponding to the checkerboard image with the smallest average reprojection error as the optimal internal parameter matrix and the optimal external parameter matrix; specifically including: Step 401, using MATLAB software to calculate the reprojection error of each checkerboard corner point based on the checkerboard corner points in each checkerboard image and the set distortion coefficient; the Stereo Camera Calibrator toolbox calculates the reprojection error of each checkerboard corner point based on the detected checkerboard corner points and the set distortion coefficient; if the calculated reprojection error exceeds a preset threshold, the MATLAB software will automatically check whether the posture of the checkerboard calibration plate in the checkerboard image is horizontal and vertical, and if there is a deviation, the staff will be prompted to re-place the checkerboard calibration plate; in addition, the MATLAB software will also detect whether the size measurement of the calibration plate is accurate, and whether the lens of the binocular camera is damaged or blocked by foreign objects, to ensure the accuracy of the checkerboard image acquisition; Step 402, calculating the average reprojection error of the checkerboard corner points in each checkerboard image; Step 403 , selecting the checkerboard image with the smallest average reprojection error as the best calibration object, and taking the internal parameter matrix and the external parameter matrix of the binocular camera corresponding to the best calibration object as the best internal parameter matrix and the best external parameter matrix.

[0028] Step 5: adjust the relative position and posture of the binocular camera based on the optimal external parameter matrix; and collect the real scene inside the car through the adjusted binocular camera to obtain the left view and the right view.

[0029] Step 6: performing distortion correction, stereo correction and remapping operations on the left view and the right view; wherein: The distortion correction of the left view and the right view specifically includes: Step 601A, combining the best internal parameter matrix and the distortion correction matrix to calculate and generate the distortion correction parameters, the distortion correction matrix expression is: ; In the formula, Indicates the actual coordinates of the pixel points on the left view and the right view; Represents the coordinates of the pixel after distortion correction; Indicates the lens distortion of the binocular camera The distortion value in the direction is related to the pixel point Related, reflecting the lens distortion on the pixel in the x direction Influence; Indicates the distortion value of the binocular camera lens distortion in the Y direction, which reflects the lens distortion on the pixel point in the Y direction. The impact of Step 602A, using the undistort Image function in the MATLAB software to perform distortion correction on the pixels in the left view and the right view according to the distortion correction parameters and the best intrinsic parameter matrix of the binocular camera. The undistort Image function remaps the pixels in the left view and the right view according to the best intrinsic parameter matrix of the binocular camera to eliminate the influence of lens distortion. The distortion correction model is: ; In the formula, Indicates the distance from the pixel point on the left view and right view image plane to the center point of the image plane; ; represents the radial distortion coefficient; represents the tangential distortion coefficient; Indicates the actual coordinates of the pixel before distortion correction; Indicates the coordinates of the pixel after distortion correction.

[0030] The stereoscopic correction of the left view and the right view specifically includes: Step 601B: The pixel coordinates after distortion correction are To perform coordinate transformation, the transformation formula is: ; In the formula, Indicates the translation amount of the binocular camera in three-dimensional space based on the translation vector. In this specific embodiment, based on the optimal intrinsic parameter matrix and extrinsic parameter matrix of the binocular camera, the coordinate transformation criteria and mathematical derivation of the prior art are used to obtain and the optimal intrinsic parameter matrix; the optimal intrinsic parameter matrix accurately encodes the angle information of the stereo camera's rotation around each coordinate axis, and the translation vector accurately determines the translation amount of the stereo camera in three-dimensional space; Thus, we can get multiple sets of corresponding coordinates in the left view and the right view. and ; Step 602B, coordinate and Transform to the camera coordinate system through the optimal intrinsic parameter matrix: ; In the formula, and Represents the coordinates of corresponding points in the left view and the right view in the camera coordinate system; represents the optimal intrinsic parameter matrix of the left camera; represents the optimal intrinsic parameter matrix of the right camera; according to the epipolar relationship, there is an essential matrix , so that any pair of corresponding points in the left view and the right view and satisfy: , Represents the coordinate vector of the pixel point in the right view The transpose of Step 603B, based on , bring it into Get ; Solved by multiple groups of corresponding pixel points ; Used to adjust the essential matrix in epipolar relation ; Step 604B, using singular value decomposition method to solve the essential matrix ; The essential matrix Decompose into ; Represents the translation vector The opposition formed into a matrix, is the rotation matrix; the essential matrix Perform singular value decomposition into ; represents an orthogonal matrix, represents a diagonal matrix, , ; Based on the above decomposition, the rotation matrix of the left camera in the binocular camera is obtained and the rotation matrix of the right camera ; and the translation vector , , where Represents an orthogonal matrix The third column of Step 605B, based on the geometric principle of camera imaging and the rotation matrix and the translation vector Construct the left camera stereo rectification transformation matrix separately and the right camera stereo rectification transformation matrix , where the left camera stereo rectification transformation matrix is The formula is: ; In the formula, Indicates the coordinates of the principal point of the left view. represents the focal length of the left camera, Indicates that the left camera is Translation matrix in direction; Right camera stereo rectification transformation matrix The formula is: ; In the formula, Indicates the coordinates of the principal point of the right view; represents the focal length of the right camera, Indicates that the right camera is Translation matrix in direction; Step 606B, using the left camera stereo correction transformation matrix and the right camera stereo rectification transformation matrix The stereoscopic correction operation is performed on the distortion-corrected left view and the right view respectively.

[0031] The remapping operation of the stereo-rectified left and right views specifically includes: Take the left camera stereo rectification transformation matrix and the right camera stereo rectification transformation matrix As the key parameter, the remapping matrix is ​​generated through the cv::initUndistortRectifyMap function, and the cv::remap function is used to remap the pixels of the left and right views so that the corresponding pixels of the left and right views are in the same horizontal direction.

[0032] Step 7, calculating the depth information of the vehicle occupants and the vehicle seats based on the remapped left view and right view; specifically including Step 701, use a deep learning-based image segmentation algorithm to segment the foreground and background in the remapped left view and right view; this specific embodiment uses a deep learning-based image segmentation algorithm to segment the foreground (target objects such as human bodies and seats) and the background (in-vehicle environment, scenery outside the vehicle, etc.) in the image to reduce the interference of complex background textures and objects on stereo matching. For occlusion situations, use multi-frame image information for processing. In the time series, analyze the changes in human body posture and position between adjacent frames, and predict the depth information of the occluded area through motion estimation and compensation algorithms, and fusion and optimization with the visible part of the current frame. In the stereo matching process, when searching for the most similar pixel block on the corresponding epipolar line of the right image, if occlusion occurs and an accurate match cannot be found, interpolation calculation or model-based depth estimation is performed in combination with the matching results and depth trends of the surrounding unoccluded areas to fill in the depth value of the occluded area;

[0033] Step 702, using stereo matching method, the pixel blocks of the driver and the car seat in the foreground of the left view are searched for the most similar pixel blocks on the corresponding epipolar line of the right view through the SSD algorithm, thereby generating a disparity map; in order to accurately locate the object, the regional matching method is used, the core of which is to carefully determine the matching window size according to the image resolution characteristics, configure a larger window for high-resolution images to capture rich details, and use a smaller window for low-resolution images to focus on key features. Carefully select pixel blocks in the left image, and then search for the most similar pixel blocks on the corresponding epipolar line of the right image according to the SSD algorithm, thereby generating a disparity map;

[0034] Step 703: Use the median filter algorithm to process the disparity map, remove noise interference, and smooth the disparity data; its core advantage is that it can effectively remove noise interference, smooth the disparity data, and at the same time excellently protect the image edge detail features. On this basis, a threshold is reasonably set according to the depth range of a specific scene, and the processed disparity map is strictly screened based on this threshold to accurately remove abnormal disparity data, effectively avoid the introduction of erroneous depth values, and ensure the high accuracy of depth calculation;

[0035] Step 704: Calculate the depth estimation value of the driver and the car seat in the foreground based on the disparity map ; The calculation formula is: ; Indicates the focal length of the binocular camera; Indicates the baseline distance of the binocular camera; Indicates the disparity between corresponding points in the left and right views.

[0036] Step 8: Construct a human body posture model, such as Figure 2 As shown, the three-dimensional coordinates of the key points of each part of the driver and passenger are calculated in combination with the depth information of the driver and passenger; specifically, the following are included: Step 801, using a Kinect V2 device to collect key points of various parts of the driver and passenger on the car seat and number them, the key points of various parts of the driver and passenger include: the focus of the head, the neck and the left and right shoulders, the right shoulder, the right elbow, the right wrist, the left shoulder, the left elbow, the left wrist, the right hip point, the right knee, the right ankle, the right foot, the left hip point, the left knee, the left ankle, and the left foot; Step 802: Calculate the depth estimate based on the lens distortion characteristics of the binocular camera, the thickness of the driver's clothing, and the lighting environment. The error rate with the real depth distance is corrected to obtain accurate depth information of the driver and the car seat; The error correction formula is as follows: η is the compensation coefficient of each error term, which is displayed as correction term, original term, distortion term, clothing term, and illumination term; Step 803, constructing a human body posture model, and calculating the three-dimensional coordinates of key points of various parts of the human body based on the depth information. The human body posture model is: ; ; In the formula, Respectively represent the focal length of the binocular camera lens in the x-direction and the y-direction; Represents the coordinates of the principal point of the binocular camera in the pixel coordinate system; and Respectively represent the pixel coordinates of the left view and the right view; and ; Represents depth information.

[0037] Step 9, constructing a car seat adjustment model, inputting the three-dimensional coordinates of key points of various parts of the driver and passenger, and outputting the target joint angles of the driver and passenger and the car seat adjustment parameters; specifically including: Step 901, based on the three-dimensional coordinates of the key points of each part of the driver and passenger and the matrix formula of the size of each part, the size of each part of the driver and passenger is calculated; including the torso L1, the height from eyes to shoulders L2, the length of the left upper arm L3, the length of the right upper arm L4, the length of the left forearm L5, the length of the right forearm L6, the length of the left thigh L7, the length of the right thigh L8, the length of the left calf L9, the length of the right calf L10, the length of the left foot L11, the length of the right foot L12, and the height of the eyes L3; The matrix formula for each part size is: ; and Represents the three-dimensional coordinates of key points of adjacent parts; Step 902, constructing a car seat adjustment model, inputting the three-dimensional coordinates of key points of each part, and outputting the target joint angle and car seat adjustment parameters; the car seat adjustment model is: In the formula, Indicates the initial angle of the joint angles of each part of the driver and passenger; represents the target joint angle; represents the square of the length of one of the limb segments adjacent to the joint, represents the square of the length of the other limb segment adjacent to the joint, Indicates The comfortable angle value of each joint, represents the total number of joint angles; Indicates the adjustment parameters of the car seat, including the car seat back tilt angle, car seat height, and car sliding adjustment distance; Indicates the calibration scale factor, including the car seat back inclination factor , Car seat height coefficient and vehicle slip adjustment coefficient ; Indicates offset, including car seat offset , Car seat height offset and car slide adjustment offset .

[0038] Step 10, setting the priority of the car seat adjustment, and adjusting the car seat based on the priority and the car seat adjustment parameter, specifically including: Step 1001, preset the optimal height threshold of the driver's eyes, take the set optimal height threshold as the starting point, and calculate the coordinates of the preset hip point in combination with the driver's trunk tilt comfort angle value, the relative distance between the steering wheel and the body, and the trunk length ; The calculation formula is: ; ; ; In the formula, Indicates the length of the car seat, Indicates the width of the car seat, Indicates the height of the car seat, Indicates the leg length of the driver and passenger. Indicates the tilt angle of the car seat back; Step 1002, setting the position of the car seat as the first priority; adjusting the position of the car seat based on the preset hip point; The parameters adjusted in this process are specifically the front-to-back sliding distance of the car seat, that is, the horizontal position, in order to make the horizontal coordinate (x direction) of the hip point conform to the preset value, so that the hip point of the driver and passenger matches the preset x coordinate, ensuring the rationality and safety of the basic driving posture; Step 1003, setting the sliding adjustment distance and lifting height of the car seat as the second priority; based on the preset hip point and the fixed heel point, the hip point is moved twice by using the cosine theorem combined with the comfortable angle value of the knee joint and the length of the thigh and calf to obtain the coordinates of the ideal hip point, and the coordinate values ​​of the ideal hip point and the preset hip point are compared to obtain the sliding adjustment distance and height lifting value of the car seat; Step 1004, setting the tilt angle of the car seat back as the third priority; adjusting the angle of the car seat back based on the comfortable tilt angle of the driver's body and the initial value of the driver's body tilt.

[0039] Among them, the adjustment parameter S outputted in step 9 is a comprehensive parameter, including the global optimization results of the seat back tilt angle, height and sliding adjustment distance, while the sliding adjustment distance and height lifting value of step 1003 are the local geometric optimization results based on the secondary movement of the hip point. According to the priority setting of step 10 (position> sliding / height> backrest angle), the adjustment value of step 1003 will supplement the corresponding parameters in step 9 (sliding and height parts) to ensure that the basic sitting posture (hip point position) meets the ergonomic requirements first, that is, the adjustment parameter S of step 9 is used to provide preliminary adjustment suggestions. Step 1003 corrects the sliding and height parameters through geometric calculation of the secondary movement of the hip point. The final adjustment value is a combination of the two. The specific calculation process is as follows: Based on preset hip points and fixed heel point , combined with the comfortable angle of the knee joint , Thigh length and calf length , calculate the ideal hip point using the law of cosines : Next, calculate the sliding adjustment distance : Then calculate the height rise and fall value : The final adjusted values ​​are integrated through priority logic: .

[0040] In order to specifically reflect the automatic adjustment of the car seat, the angle adjustment of the car seat backrest, the sliding distance adjustment of the car seat and the height adjustment of the car seat in this specific embodiment rely on motors respectively; In addition, the intelligent car system also integrates voice, fatigue detection, and audio systems. Drivers and passengers can transmit adjustment information to the processor through voice, control the fine-tuning of the car seat, and provide feedback through the audio system; the fatigue detection system detects that the driver is tired and sleepy, and transmits the information to the feedback and adjustment module to adjust the seat to improve driving efficiency.

[0041] In the description of the embodiments of the present application, it should be noted that in the description of the present application, terms such as "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings. This is only for the convenience of description, and does not indicate or imply that the device or component must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present application.

[0042] In the description of the present application, the description with reference to the terms "one embodiment", "some embodiments", "in the present embodiment", "specific example", or "some examples" etc. means that the specific features, mechanisms, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, mechanisms, materials or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0043] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. An intelligent adjustment method for a car seat by sensing the posture of a driver and passenger, characterized in that: include: Step 1: The driver and passengers sit in the car seat and make preliminary adjustments; Step 2: Use binocular camera to collect A checkerboard image of a driver or passenger holding a checkerboard calibration plate; Step 3, perform checkerboard calibration, checkerboard corner point detection, binocular camera internal parameter matrix and external parameter matrix calculation, and set distortion coefficient on the checkerboard image; Step 4, calculating the reprojection error of the checkerboard corner points based on the checkerboard corner points and the distortion coefficients, and selecting the internal parameter matrix and the external parameter matrix corresponding to the checkerboard image with the smallest average reprojection error as the optimal internal parameter matrix and the optimal external parameter matrix; Step 5: adjust the relative position and posture of the binocular camera based on the optimal external parameter matrix; and collect the real scene inside the car through the adjusted binocular camera to obtain the left view and the right view; Step 6, performing distortion correction, stereo correction and remapping operations on the left view and the right view; Step 7, calculating the depth information of the vehicle occupants and the vehicle seats based on the remapped left view and right view; Step 8, constructing a human body posture model, and calculating the three-dimensional coordinates of key points of various parts of the driver and passenger in combination with the depth information of the driver and passenger; Step 9, constructing a car seat adjustment model, inputting the three-dimensional coordinates of key points of various parts of the driver and passenger, and outputting the target joint angles of the driver and passenger and the car seat adjustment parameters; Step 10, setting the priority of the car seat adjustment, and adjusting the car seat based on the priority and the car seat adjustment parameters.

2. The method for intelligently adjusting a car seat by sensing the posture of a driver and passenger according to claim 1, characterized in that: The step 1 specifically includes: Step 101, a pressure sensor is arranged on the car seat, the binocular camera is arranged at the intersection of the car roof and the A-pillar, and a millimeter wave radar is arranged on the car steering wheel; Step 102: The driver and passenger sit on the car seat, and the weight parameter, height parameter, gender parameter, and distance parameter between the driver and passenger and the steering wheel are obtained according to the pressure sensor, binocular camera, and millimeter wave radar; Step 103, constructing an initial height adjustment model of the car seat: ; In the formula, Indicates the initial height of the car seat; Indicates the weight parameter of the driver and passengers; and are coefficients that vary in direct proportion to the weight parameter; the initial height of the car seat is preliminarily adjusted according to the weight parameter; Step 104, constructing a car seat back tilt angle adjustment model based on the sex parameter sex of the driver and passenger: ； In the formula, Represents the height parameter of male, Represents the height parameter of female; and represents the coefficient that varies in direct proportion to the height parameter of males; and represents the coefficient that changes in direct proportion to the height parameter of women; Indicates the tilt angle of the car seat back for male occupants; Indicates the tilt angle of the car seat back when the driver and passenger are female; preliminarily adjusts the tilt angle of the car seat back according to the gender parameter and the height parameter; Step 105, constructing a car seat position adjustment model based on the distance parameter between the driver and the steering wheel: ； In the formula, Indicates the amount of adjustment of the car seat position; Indicates the distance parameter between the driver and the steering wheel; Indicates the preset safety distance threshold; Represents the proportional coefficient; calculates the distance parameter between the driver and the steering wheel, calculates the car seat position adjustment amount and preliminarily adjusts the position of the car seat.

3. The method for intelligently adjusting a car seat by sensing the posture of a driver and passenger according to claim 1, characterized in that: The step 2 specifically includes: Step 201, setting the parameters of the binocular camera to adjust the resolution and frame rate of the binocular camera; Step 202, selecting a checkerboard calibration plate according to the resolution and frame rate of the binocular camera; Step 203: The driver or passenger holds a checkerboard calibration plate in the field of view of the binocular camera and uses the binocular camera to capture the checkerboard calibration plate by changing the posture and distance of the checkerboard calibration plate. Checkerboard image.

4. The method for intelligently adjusting a car seat by sensing the posture of a driver and passenger according to claim 1, characterized in that: The step 3 specifically includes: Step 301, import all chessboard images into MATLAB software; Step 302: setting the distortion coefficient according to the distortion characteristics of the binocular camera, wherein the distortion coefficient includes the radial distortion coefficient. and tangential distortion coefficient ; Step 303, using MATLAB software to detect all the checkerboard corner points in each checkerboard image, and calculate the internal parameter matrix and external parameter matrix of the binocular camera corresponding to each checkerboard image, wherein the internal parameter matrix includes the focal length of the binocular camera , principal point coordinates , radial distortion coefficient and tangential distortion coefficient ; The external parameter matrix includes a rotation matrix for describing the rotation direction of the binocular camera And the translation vector T used to describe the translation distance of the binocular camera.

5. The method for intelligently adjusting a car seat by sensing the posture of a driver and passenger according to claim 1, characterized in that: The step 4 specifically includes: Step 401, calculating the reprojection error of each checkerboard corner point based on the checkerboard corner points in each checkerboard image and the set distortion coefficient by using MATLAB software; Step 402, calculating the average reprojection error of the checkerboard corner points in each checkerboard image; Step 403 , selecting the checkerboard image with the smallest average reprojection error as the best calibration object, and taking the internal parameter matrix and the external parameter matrix of the binocular camera corresponding to the best calibration object as the best internal parameter matrix and the best external parameter matrix.

6. The method for intelligently adjusting a car seat by sensing the posture of a driver and passenger according to claim 1, characterized in that: The distortion correction of the left view and the right view in step 6 specifically includes: Step 601A, combining the best internal parameter matrix and the distortion correction matrix to calculate and generate the distortion correction parameters, the distortion correction matrix expression is: ； In the formula, Represents the coordinates of the pixel points on the left view and the right view; Represents the coordinates of the pixel after distortion correction; Indicates the lens distortion of the binocular camera Distortion value in direction; Indicates the distortion value of the binocular camera's lens distortion in the Y direction; Step 602A, using the undistort Image function in MATLAB software to perform distortion correction on the pixels in the left view and the right view according to the distortion correction parameters and the optimal internal parameter matrix of the binocular camera to eliminate the influence of lens distortion. The distortion correction model is: ； In the formula, Indicates the distance from the pixel point on the left view and right view image plane to the center point of the image plane; ; represents the radial distortion coefficient; represents the tangential distortion coefficient; Indicates the coordinates of the pixel before distortion correction; Represents the coordinates of the pixel after distortion correction; The stereoscopic correction of the left view and the right view in step 6 specifically includes: Step 601B: The pixel coordinates after distortion correction are To perform coordinate transformation, the transformation formula is: ； In the formula, Indicates the translation amount of the binocular camera in three-dimensional space based on the translation vector. represents the scaling factor; Thus, we can get multiple sets of corresponding coordinates in the left view and the right view. and ; Step 602B, coordinate and Transform to the camera coordinate system through the optimal intrinsic parameter matrix: ； In the formula, and Represents the coordinates of corresponding points in the left view and the right view in the camera coordinate system; represents the optimal intrinsic parameter matrix of the left camera; represents the optimal intrinsic parameter matrix of the right camera; according to the epipolar relationship, there is an essential matrix , so that any pair of corresponding points in the left view and the right view and satisfy: , Represents the coordinate vector of the pixel point in the right view The transpose of Step 603B, based on , bring it into Get ; Solved by multiple groups of corresponding pixel points ; Used to adjust the essential matrix in epipolar relation ; Step 604B, using singular value decomposition method to solve the essential matrix ; The essential matrix Decompose into ; Represents the translation vector The opposition formed into a matrix, is the rotation matrix; The essential matrix Perform singular value decomposition into ; represents an orthogonal matrix, represents a diagonal matrix, , ; Based on the above decomposition, the rotation matrix of the left camera in the binocular camera is obtained and the rotation matrix of the right camera ; and the translation vector , , where Represents an orthogonal matrix The third column of Step 605B, based on the geometric principle of camera imaging and the rotation matrix and the translation vector Construct the left camera stereo rectification transformation matrix separately and the right camera stereo rectification transformation matrix , where the left camera stereo rectification transformation matrix is The formula is: ； In the formula, Indicates the coordinates of the principal point of the left view. represents the focal length of the left camera, Indicates that the left camera is Translation matrix in direction; Right camera stereo rectification transformation matrix The formula is: ； In the formula, Indicates the coordinates of the principal point of the right view; represents the focal length of the right camera, Indicates that the right camera is Translation matrix in direction; Step 606B, using the left camera stereo correction transformation matrix and the right camera stereo rectification transformation matrix Perform stereo correction operations on the distortion-corrected left and right views respectively; The remapping operation of the left view and the right view after stereoscopic correction in step 6 specifically includes: Take the left camera stereo rectification transformation matrix and the right camera stereo rectification transformation matrix As the key parameter, the remapping matrix is ​​generated through the cv::initUndistortRectifyMap function, and the cv::remap function is used to remap each pixel of the left view and the right view.

7. The method for intelligently adjusting a car seat by sensing the posture of a driver and passenger according to claim 1, characterized in that: The step 7 specifically includes: Step 701, using a deep learning-based image segmentation algorithm to segment the foreground and background in the remapped left view and right view; Step 702, using a stereo matching method, searching for the most similar pixel blocks of the driver and the car seat in the foreground of the left view on the corresponding epipolar lines of the right view through an SSD algorithm, thereby generating a disparity map; Step 703, using a median filter algorithm to process the disparity map, remove noise interference, and smooth the disparity data; Step 704: Calculate the depth estimation value of the driver and the car seat in the foreground based on the disparity map ; The calculation formula is: Indicates the focal length of the binocular camera; Indicates the baseline distance of the binocular camera; Indicates the disparity between corresponding points in the left and right views.

8. The method for intelligently adjusting a car seat by sensing the posture of a driver and passenger according to claim 1, characterized in that: The step 8 specifically includes: Step 801, using a Kinect V2 device to collect key points of various parts of the driver and passenger on the car seat and number them, the key points of various parts of the driver and passenger include: the focus of the head, the neck and the left and right shoulders, the right shoulder, the right elbow, the right wrist, the left shoulder, the left elbow, the left wrist, the right hip point, the right knee, the right ankle, the right foot, the left hip point, the left knee, the left ankle, and the left foot; Step 802: Calculate the depth estimate based on the lens distortion characteristics of the binocular camera, the thickness of the driver's clothing, and the lighting environment. The error rate with the real depth distance is corrected to obtain accurate depth information of the driver and the car seat; Step 803, constructing a human body posture model, and calculating the three-dimensional coordinates of key points of various parts of the human body based on the depth information. The human body posture model is: ； ； In the formula, Respectively represent the focal length of the binocular camera lens in the x-direction and the y-direction; Represents the coordinates of the principal point of the binocular camera in the pixel coordinate system; and Respectively represent the pixel coordinates of the left view and the right view; and ; Represents depth information.

9. The method for intelligently adjusting a car seat by sensing the posture of a driver and passenger according to claim 1, characterized in that: The step 9 specifically includes: Step 901, based on the three-dimensional coordinates of the key points of each part of the driver and passenger and the matrix formula of the size of each part, the size of each part of the driver and passenger is calculated; including the torso L1, the height from eyes to shoulders L2, the length of the left upper arm L3, the length of the right upper arm L4, the length of the left forearm L5, the length of the right forearm L6, the length of the left thigh L7, the length of the right thigh L8, the length of the left calf L9, the length of the right calf L10, the length of the left foot L11, the length of the right foot L12, and the height of the eyes L3; The matrix formula for each part size is: ; and Represents the three-dimensional coordinates of key points of adjacent parts; Step 902, constructing a car seat adjustment model, inputting the three-dimensional coordinates of key points of each part, and outputting the target joint angle and car seat adjustment parameters; the car seat adjustment model is: In the formula, Indicates the initial angle of the joint angles of each part of the driver and passenger; represents the target joint angle; represents the square of the length of one of the limb segments adjacent to the joint, represents the square of the length of the other limb segment adjacent to the joint, Indicates The comfortable angle value of each joint, represents the total number of joint angles; Indicates the adjustment parameters of the car seat, including the car seat back tilt angle, car seat height, and car sliding adjustment distance; Indicates the calibration scale factor, including the car seat back inclination factor , Car seat height coefficient and vehicle slip adjustment coefficient ; Indicates offset, including car seat offset , Car seat height offset and car slide adjustment offset .

10. The method for intelligently adjusting a car seat by sensing the posture of a driver and passenger according to claim 1, characterized in that: The priorities set in step 10 include: Step 1001, preset an optimal height threshold for the driver's eyes, and use the preset optimal height threshold as a starting point to calculate the coordinates of the preset hip point in combination with the driver's torso tilt comfort angle value, the relative distance between the steering wheel and the torso, and the torso length. ; The calculation formula is: ； ； ； In the formula, Indicates the length of the car seat, Indicates the width of the car seat, Indicates the height of the car seat, Indicates the leg length of the driver and passenger. Indicates the tilt angle of the car seat back; Step 1002, setting the position of the car seat as the first priority; adjusting the position of the car seat based on the preset hip point; Step 1003, setting the sliding adjustment distance and lifting height of the car seat as the second priority; based on the preset hip point and the fixed heel point, the hip point is moved twice by using the cosine theorem combined with the comfortable angle value of the knee joint and the length of the thigh and calf to obtain the coordinates of the ideal hip point, and the coordinate values ​​of the ideal hip point and the preset hip point are compared to obtain the sliding adjustment distance and height lifting value of the car seat; Step 1004, setting the tilt angle of the car seat back as the third priority; adjusting the angle of the car seat back based on the comfortable tilt angle of the driver's body and the initial value of the driver's body tilt.

Images