WO2019104741A1

WO2019104741A1 - Method and system for measuring compensating coefficient for odometer of visual robot cleaner

Info

Publication number: WO2019104741A1
Application number: PCT/CN2017/114340
Authority: WO
Inventors: 王声平; 周毕兴
Original assignee: 深圳市沃特沃德股份有限公司
Priority date: 2017-12-01
Filing date: 2017-12-01
Publication date: 2019-06-06

Abstract

A method and system for measuring a compensating coefficient for an odometer of a visual robot cleaner. The method comprises: calculating an actual displacement of the visual robot cleaner in a designated environment according to pictures photographed during the movement, and calculating the displacement by the odometer, and then calculating the compensating coefficient according to a formula, so as to improve the speed and accuracy of calculation.

Description

\¥0 2019/104741 卩(:17 \2017/114340

Method and system for measuring compensation coefficient of visual sweeping robot odometer

[0001] The present invention relates to the field of factory inspection of visual sweeping robots, and more particularly to a method and system for measuring the compensation coefficient of a visual sweeping robot odometer.

Background technique

[0002] The odometer is a key component of a visual sweeping robot used to measure the displacement and moving speed of a visual sweeping robot. The odometer is typically provided with an encoder that calculates the distance traveled based on the angle at which the encoder rotates and the circumference of the wheel. Due to environmental factors or structural factors, the calculated distance between the odometer and the actual distance is different. Generally, the odometer is calibrated before leaving the factory, and the compensation coefficient is calculated to compensate the distance calculated by the odometer. The detected distance is accurate.

[0003] Nowadays, the visual sweeping robot is manually manually calibrated to calculate the compensation coefficient before leaving the factory, so that the measured speed is relatively slow and the accuracy is not very high.

technical problem

[0004] A primary object of the present invention is to provide a method for quickly and automatically calibrating a visual sweeping robot odometer.

Problem solution

Technical solution

[0005] The present invention provides a method for measuring a compensation coefficient of a visual sweeping robot odometer. The ground and ceiling of the measurement environment are both horizontally set. The method includes:

[0006] The visual sweeping robot moves a distance within the test environment while the camera captures the ceiling image

[0007] calculating the displacement of the actual movement of the visual sweeping robot according to the acquired image! _^ 2;

[0008] obtaining the displacement 0 of the sweeping robot recorded by the odometer;

[0009] The compensation coefficient is calculated by dividing the displacement of the actual movement of the cleaning robot by 1 _^ 2 by the displacement of the cleaning robot recorded by the odometer 1 _^ 1.

[0010] Further, the step of calculating the displacement 1 _^ 2 of the actual movement of the visual sweeping robot according to the acquired image comprises: \¥0 2019/104741 卩(:17 \2017/114340

[0011] according to the picture collected by the visual sweeping robot, the feature points in the picture are spatially reconstructed three-dimensionally, and the three-dimensional space coordinates of the feature points are obtained, thereby obtaining three-dimensional information of the image;

[0012] calculating a displacement 2 of the actual movement of the visual cleaning robot according to the three-dimensional information of the image.

[0013] Further, the step of calculating the displacement 1 _^ 2 of the actual movement of the visual cleaning robot according to the three-dimensional information of the image comprises:

[0014] obtaining a moving distance of the visual sweeping robot in a three-dimensional space of the number 11;

[0015] calculating scale information corresponding to each scale according to a preset rule

[0016] Multiplying the moving distance of the visual sweeping robot by the scale number 11 in the three-dimensional space by the scale information ₈ to calculate the displacement 1 _^ 2 of the actual movement of the visual sweeping robot.

[0017] Further, the step of calculating the scale information ₈ corresponding to each scale according to a preset rule includes:

[0018] obtaining a camera to ceiling plane number of 2;

[0019] calculating the actual height of the camera from the ceiling based on the known environmental information and the parameters of the visual sweeping robot itself

[0020] The scale information 8 is calculated by dividing the actual height of the camera from the ceiling by the number of scales 12 from the camera to the ceiling plane.

[0021] Further, the obtaining the number of scales 12 of the camera to the ceiling plane comprises the steps of:

[0022] Estimating the ceiling plane using RANSAC plane fitting to obtain normal 11;

[0023] Calculate the number of scales from the camera to the ceiling plane according to the formula 12=11 «^, where

The vector representing the normal 11 in the reconstruction space, \ represents the coordinates of the camera in the three-dimensional space.

[0024] Further, after the calculating the compensation coefficient, the method includes the following steps:

[0025] Calculating a plurality of compensation coefficients a plurality of times and calculating an average value of the plurality of compensation coefficients.

[0026] Further, the step of repeatedly calculating a plurality of compensation coefficients multiple times includes:

[0027] The visual sweeping robot calculates a plurality of compensation coefficients after moving on the ground of different materials.

[0028] The present invention also provides a system for measuring a compensation coefficient of a visual sweeping robot odometer. The ground and ceiling of the measurement environment are both horizontally set. The system includes:

[0029] a mobile device, configured for the visual sweeping robot to move a distance within the test environment, and the camera captures a ceiling image; \¥0 2019/104741 卩(:17 \2017/114340

[0030] an actual displacement device, configured to calculate a displacement of the actual movement of the visual sweeping robot according to the collected picture

1 _^ 2;

[0031] odometer displacement device, used to obtain the displacement recorded by the odometer 1 _^ 1;

[0032] The compensation coefficient device is configured to calculate the compensation coefficient by dividing the displacement of the actual movement of the cleaning robot by 1 _^ 2 by the displacement of the cleaning robot recorded by the odometer by 1 _^ 1.

[0033] Further, the actual displacement device includes:

[0034] a building module, configured to perform spatial three-dimensional reconstruction of the feature points in the image according to the image captured by the visual sweeping robot, to obtain a three-dimensional space coordinate of the feature point, and thereby obtain three-dimensional information of the image;

[0035] a calculation module, configured to calculate, according to the three-dimensional information of the image, a displacement of the actual movement of the visual cleaning robot by 1 _^ 2.

[0036] Further, the calculating module includes:

[0037] a moving scale sub-module, configured to acquire a scale number of the moving distance of the visual sweeping robot in the three-dimensional space 11

[0038] a length information sub-module, configured to calculate scale information corresponding to each scale according to a preset rule

[0039] The calculation formula sub-module is configured to multiply the scale number _{8 of the} three-dimensional space by the moving distance of the visual sweeping robot by the scale information ₈ to calculate the displacement 1 _^ 2 of the actual movement of the visual sweeping robot.

[0040] Further, the length information submodule includes:

[0041] obtaining a scale unit, the number of scales used to obtain the camera to the ceiling plane is 2;

[0042] calculating a height unit for calculating the actual height of the camera from the ceiling based on the known environmental information and the parameters of the visual sweeping robot itself

[0043] A calculation formula unit for calculating the scale information 8 based on the actual height of the camera from the ceiling divided by the scale number 12 of the camera to the ceiling plane.

[0044] Further, the obtaining the scale unit comprises:

[0045] a normal subunit for estimating the ceiling plane using RANSAC plane fitting to obtain a normal! 1 _;

[0046] calculating a distance subunit for calculating a camera-to-ceiling plane scale number 2 according to the formula 12=!! «X, wherein the normal line is represented by a vector in the reconstruction space, where X indicates that the camera is in the The coordinates in 3D space.

[0047] Further, the system for measuring a compensation coefficient of a visual sweeping robot odometer further includes: \¥0 2019/104741 卩(:17 \2017/114340

[0048] A plurality of calculation means for calculating the distance between the normal line !1 and the camera of the sweeping visual robot, and 2 is obtained.

[0049] Further, the multiple calculation device includes:

[0050] The ground module is configured to calculate a plurality of compensation coefficients after the visual sweeping robot moves on the ground of different materials.

Advantageous effects of the invention

Beneficial effect

Compared with the prior art, the beneficial effects of the present invention are: using the sensor of the visual sweeping robot to automatically calculate the compensation coefficient, improve the calibration efficiency of the visual sweeping robot, reduce the time for manual calibration, and provide trademarks. Precision.

Brief description of the drawing

DRAWINGS

1 is a schematic diagram showing the steps of a method for measuring a compensation coefficient of a visual sweeping robot odometer according to an embodiment of the present invention;

2 is a schematic diagram showing the steps of a method for measuring a compensation coefficient of a visual sweeping robot odometer according to an embodiment of the present invention;

3 is a schematic diagram showing the steps of a method for measuring a compensation coefficient of a visual sweeping robot odometer according to an embodiment of the present invention;

4 is a schematic diagram showing the steps of a method for measuring a compensation coefficient of a visual sweeping robot odometer according to an embodiment of the present invention;

5 is a schematic diagram showing the steps of a method for measuring a compensation coefficient of a visual sweeping robot odometer according to an embodiment of the present invention;

6 is a schematic structural view of a system for measuring a compensation coefficient of a visual sweeping robot odometer according to an embodiment of the present invention;

7 is a schematic structural diagram of an actual displacement device of a system for measuring a compensation coefficient of a visual sweeping robot odometer according to an embodiment of the present invention;

8 is a schematic structural diagram of a calculation module of a system for measuring a compensation coefficient of a visual sweeping robot odometer according to an embodiment of the present invention;

9 is a length letter of a system for measuring a compensation coefficient of a visual sweeping robot odometer according to an embodiment of the invention. \¥0 2019/104741 卩 (: 17 \2017/114340 structure diagram of the sub-module;

10 is a schematic structural diagram of a calculation height unit of a system for measuring a compensation coefficient of a visual sweeping robot odometer according to an embodiment of the present invention;

11 is a schematic structural view of a system for measuring a compensation coefficient of a visual sweeping robot odometer according to an embodiment of the present invention.

[0063] The implementation, functional features, and advantages of the present invention will be further described in conjunction with the embodiments.

BEST MODE FOR CARRYING OUT THE INVENTION

The specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0065] Referring to FIG. 1, a method for measuring a compensation coefficient of a visual sweeping robot odometer is proposed. The ground and ceiling of the measurement environment are horizontally set, and the distance between the two planes is known. The measuring method includes the following steps:

[0066] 1. The visual sweeping robot moves a distance within the test environment, and the camera captures the ceiling image;

[0067] 2. Calculate the displacement of the actual movement of the visual sweeping robot according to the collected picture! _^ 2;

[0068] 33, obtain the displacement 0 recorded by the odometer;

[0069] 4. The displacement coefficient is calculated by dividing the displacement of the actual movement of the cleaning robot by 1 _^ 2 by the displacement of the sweeping robot recorded by the odometer by 1 _^ 1 .

[0070] In this embodiment, the visual sweeping robot moves within the environment, and the floor and the ceiling are horizontally disposed, that is, the floor and the ceiling are flat planes, without slope, curvature, and inclination, the visual sweeping robot can move normally, and the visual sweeping is performed. The machine has a vision system, and the vision system uses the machine instead of the human eye to make measurements and judgments. The visual system refers to the conversion of the ingested target into an image signal by a machine vision product (ie, an image capturing device, divided into CMOS and CCD), and transmitted to a dedicated image processing system, and converted into information according to pixel distribution, brightness, color, and the like. Digitized signals; The vision system performs various operations on these signals to extract features of the target, and then controls the behavior of the device in the field based on the result of the discrimination. It can be used to measure distances. The visual sweeping robot is equipped with a vision system that can measure the distance of obstacles in front and obtain map information of the cleaning environment. In this embodiment, the acquisition map is based on the measurement environment and the vision system. \¥0 2019/104741 卩(:17 \2017/114340 piece, measure the actual movement displacement L2 of a distance moved by the visual sweeping robot, and then read the displacement L1 calculated by the odometer of the visual sweeping robot at this distance, By dividing L2 by L1, the compensation factor can be calculated.

[0071] Referring to FIG. 2, further, the step S2 includes:

[0072] S21. Perform spatial three-dimensional reconstruction of the feature points in the image according to the image captured by the visual sweeping robot.

Obtaining a three-dimensional space coordinate of the feature point, thereby obtaining three-dimensional information of the image;

[0073] S22. Calculate a displacement L2 of the actual movement of the visual cleaning robot according to the three-dimensional information.

[0074] In the embodiment, when the visual sweeping robot cleans, the visual system takes a picture, extracts feature points according to the photographed photos, performs spatial three-dimensional reconstruction according to the feature points, and converts the captured two-dimensional image into a three-dimensional image, The camera calibration is used to establish an effective imaging model, and the internal and external parameters of the camera are solved, so that the three-dimensional point coordinates in the space can be obtained by combining the matching results of the image, thereby achieving the purpose of three-dimensional reconstruction, constructing a three-dimensional space, and then moving the visual sweeping robot. The front position and the moved position are marked in the three-dimensional space, and the actual moving distance L2 of the visual cleaning robot is calculated in the coordinate system. The three-dimensional space is established, and the position of the visual sweeping robot is more accurately obtained, and the calculated actual moving distance L2 is more accurate.

[0075] Referring to FIG. 3, further, step S22 includes:

[0076] S221. Obtain a scale number T1 of a moving distance of the visual sweeping robot in a three-dimensional space;

[0077] S222, calculating scale information s corresponding to each scale according to a preset rule;

[0078] S223. Multiply the scale number T1 of the three-dimensional space by the moving distance of the visual sweeping robot by the scale information s, and calculate the displacement L2 of the actual movement of the visual sweeping robot.

[0079] In this embodiment, in the three-dimensional space, each dimension has a scale at intervals, and the distance corresponding to the scale refers to the scale information s. In the three-dimensional coordinate system, the scale number T1 of the movement of the visual sweeping robot is acquired, and then L2 is calculated based on the length information s of each scale number based on the two quantities. Specifically, L2 is equal to the product of T1 and s. The size of s can be calculated according to preset rules and some parameters preset by the user.

[0080] Referring to FIG. 4, the step of calculating the length information s S222 corresponding to each scale according to a preset rule includes:

[0081] S2221. Obtain a scale number T2 of the camera to the ceiling plane; \¥0 2019/104741 卩(:17 \2017/114340

[0220] S2222, calculating the actual height H of the camera from the ceiling according to the known environmental information and the parameters of the visual sweeping robot itself;

[0083] S2223, dividing the actual height H of the camera from the ceiling by the number of scales from the camera to the ceiling plane T

2. Calculate the scale information s.

In this embodiment, in the three-dimensional space, the camera is a point, and the ceiling is in a plane, and the number of scales of the camera and the ceiling plane is calculated. In addition, the actual distance value between the camera and the ceiling can be calculated according to the known environment. That is, the actual distance value corresponding to the scale number can be calculated, and then the actual length value corresponding to each scale can be obtained obviously. The s can be calculated by dividing the actual height H of the camera from the ceiling by the number of scales T2 from the camera to the ceiling plane. In the known environment, the height m of the ceiling from the ground is known. Among the built-in parameters of the visual sweeping robot, the height H2 of the camera on the visual sweeping robot is known, then the visual sweeping robot is placed on the ground. When it is up, the actual distance between the camera and the ceiling is H=H1-H2. Then divide H by T2 and calculate the scale information s.

[0085] Referring to FIG. 5, further, step S2221 includes:

[0086] S22211, using the RANSAC plane fitting to estimate the ceiling plane to obtain a normal n;

[0087] S22212, obtaining T2 according to the formula T2=n ^T *x, obtaining a scale number T2 of the camera to the ceiling plane, where n ^T represents a vector of the normal line _n in the reconstruction space, and x represents a camera in the three-dimensional space. coordinate.

[0088] In this embodiment, RANSAC is an abbreviation of Random Sample Consensus, which is an algorithm for calculating valid mathematical sample parameters based on a set of sample data sets containing abnormal data, and obtaining valid sample data. The RANSAC algorithm is often used in computer vision to solve the matching point problem of a pair of cameras and the calculation of the basic matrix in the field of stereo vision. The camera has a certain angle. The established three-dimensional space contains many points on the ceiling plane. By fitting with RANSAC, the plane of the ceiling plane in the straight three-dimensional space can be calculated. The plane is the normal of the ceiling plane in three-dimensional space. Use n to indicate. Then, the distance from the ceiling to the camera of the visual sweeping robot is calculated according to the normal in the three-dimensional space, that is, the distance T2 between the camera and the ceiling in the three-dimensional space, and the T2 is the scale number in the three-dimensional coordinate system, instead of the actual distance value. . Specifically, the correspondence between the normal line n and the camera-to-ceiling distance T2 in the three-dimensional space is: T2=n ^T * _x , where x is the coordinate of the camera in the three-dimensional space, and n ^T represents the normal. The vector in this three-dimensional space. \¥0 2019/104741 卩(:17 \2017/114340

[0089] Further, after the calculating the compensation coefficient, the method includes the following steps:

[0090] 85. Calculate a plurality of compensation coefficients repeatedly, and calculate an average value of the plurality of compensation coefficients.

[0091] In this embodiment, after the compensation coefficient is calculated, the compensation coefficient inevitably has an error. Moreover, the error is a random error. To further reduce the error, the best method is to perform multiple experimental calculations. Repeat the above steps of 31^4 to obtain a plurality of compensation coefficients, and calculate an average value of the plurality of compensation coefficients. The average value is used as the final compensation coefficient to improve the accuracy of the detection distance of the visual sweeping robot.

[0092] Further, the step of repeatedly calculating the plurality of compensation coefficients multiple times includes:

[0093] The visual sweeping robot calculates a plurality of compensation coefficients after moving on the ground of different materials.

[0094] In the present embodiment, since the visual sweeping robot has to face different families, various factors must be comprehensively considered when measuring the compensation coefficient, so it is necessary to consider different friction coefficients. The general household floor includes various materials such as concrete floor, tile, wooden floor, and carpet. After the compensation coefficient is measured on the floor with different friction coefficients of different materials, the average is calculated, and a standard compensation coefficient is finally obtained.

6, the present invention further provides a system for measuring a compensation coefficient of a visual sweeping robot odometer, wherein the ground and ceiling of the measurement environment are both horizontally arranged, and the system includes:

[0096] the mobile device 1 is configured to move the visual sweeping robot within the test environment for a distance while the camera captures the ceiling image;

[0097] The actual displacement device 2 is configured to calculate a displacement of the actual movement of the visual sweeping robot by 1 _^ 2 according to the acquired picture;

[0098] odometer displacement device 3, used to obtain the displacement calculated by the odometer 1 _^ 1;

[0099] The compensation coefficient device 4 is configured to calculate the compensation coefficient by dividing the displacement 1 _^ 2 of the actual movement of the cleaning robot by the displacement 1 _^ 1 of the cleaning robot recorded by the odometer.

[0100] In this embodiment, the visual sweeping robot moves within the environment, and the floor and the ceiling are horizontally disposed, that is, the floor and the ceiling are flat planes, and the sweeping robot can move normally without slope, curvature, and tilt, and the visual sweeping machine With a vision system, the vision system uses machines instead of the human eye to make measurements and judgments. The visual system refers to the conversion of the ingested target into an image signal by a machine vision product (ie, an image capturing device, divided into CMOS and CCD), and transmitted to a dedicated image processing system, and converted into information according to pixel distribution, brightness, color, and the like. Digital signal; the visual system carries out various operations on these signals \¥0 2019/104741 卩(:17 \2017/114340 Calculate the characteristics of the target, and then control the action of the device according to the result of the discrimination. Can be used to measure the distance, the mobile device 1 of the visual sweeping robot controls the visual sweeping When the robot moves, the camera is simultaneously controlled to take a picture, and a picture of the ceiling is collected. In this embodiment, the actual displacement device 2 collects a picture according to the measurement environment and the vision system, and measures the actual movement displacement L2 of a distance moved by the visual sweeping robot, and then The odometer displacement device 3 reads the displacement L1 calculated by the odometer of the visual sweeping robot at this distance, and the compensation coefficient device 4 divides L2 by L1 to calculate the compensation coefficient.

[0101] Referring to FIG. 7, further, the actual displacement device 2 includes:

[0102] The building module 21 is configured to perform spatial three-dimensional reconstruction according to the image captured by the visual sweeping robot, and obtain three-dimensional spatial coordinates of the feature point, thereby obtaining three-dimensional information of the image;

[0103] The calculating module 22 is configured to calculate a displacement L2 of the actual movement of the visual cleaning robot according to the three-dimensional information of the image.

[0104] In this embodiment, when the visual cleaning robot is in the cleaning, the visual system takes a picture, and the building module 21 extracts the feature points according to the photographed photos, performs spatial three-dimensional reconstruction according to the feature points, and converts the captured two-dimensional images into three-dimensional images. Through the camera calibration to establish an effective imaging model, the internal and external parameters of the camera are solved, so that the three-dimensional point coordinates in the space can be obtained by combining the matching results of the image, thereby achieving the purpose of three-dimensional reconstruction, constructing a three-dimensional space, and then sweeping the vision. The position before the movement of the robot and the position after the movement are marked in the three-dimensional space, and the calculation module 22 calculates the actual moving distance L2 of the visual cleaning robot in the coordinate system. The three-dimensional space is established, and the position of the visual sweeping robot is more accurately obtained, and the calculated actual moving distance L2 is more accurate.

[0105] Referring to FIG. 8, further, the calculating module 22 includes:

[0106] The moving scale sub-module 221 is configured to acquire the scale number of the movement of the visual sweeping robot in three-dimensional space T1

[0107] The length information sub-module 222 is configured to calculate scale information _S corresponding to each scale according to a preset rule _;

[0108] The calculation formula sub-module 223 is configured to calculate the scale information L2 by multiplying the scale number T 1 of the three-dimensional space by the moving distance of the visual sweeping robot by the scale information s.

[0109] In this embodiment, in a three-dimensional space, each dimension has a scale at intervals, and the distance corresponding to the scale refers to the scale information s. In the three-dimensional space, the moving scale sub-module 221 acquires the scale number T1 of the movement of the visual sweeping robot, and then the length information sub-module 222 calculates the length s corresponding to each scale, \¥0 2019/104741 卩 (: 17 \2017/114340 calculation formula sub-module 223 according to the length information _{8 of} each scale number, according to these two quantities can be calculated 1 _^ 2. Specifically, 1 _^ 2 is equal 11 and ₈ and the product. The size of 8 can be calculated according to preset rules and some parameters preset by the user.

[0110] Referring to FIG. 9, further, the length information sub-module 222 includes:

[0111] obtaining a scale unit 2221, the scale for obtaining the camera to the ceiling plane is 2;

[0112] a calculation height unit 2222, configured to calculate the actual height of the camera from the ceiling according to the known environmental information and the parameters of the visual cleaning robot itself

[0113] The calculation formula unit 2223 is configured to calculate the scale information 8 by dividing the actual height 11 of the camera from the ceiling by the scale number 12 of the camera to the ceiling plane.

[0114] In this embodiment, in the three-dimensional coordinate space, the camera is a point, the ceiling is located in a plane, and the calculation scale unit 2221 calculates the number of scales of the camera and the ceiling plane. Additionally, the calculated height unit 2222 can calculate the actual distance value between the camera and the ceiling, according to known circumstances. That is, the actual distance value corresponding to the scale number can be calculated, and then the actual length value corresponding to each scale can be obtained obviously. The calculation formula unit 2223 divides the actual height 11 of the camera from the ceiling by the number of scales of the camera to the ceiling plane by 2, that is, ₈ can be calculated. In known environments, the height of the ceiling from the ground is known. Among the built-in parameters of the visual sweeping robot, the height 112 of the camera on the visual sweeping robot is known, then the visual sweeping robot is placed on the ground. When the actual distance between the camera and the ceiling is

Then divide by 2 and calculate the scale information 8.

[0115] Referring to FIG. 10, further, the calculating the scale unit 2222 includes:

[0116] 22221 normal subunit, using RANSAC estimation ceiling plane fitting plane, the normal line to give _1!;

[0117] Calculate the distance sub-unit 22222 for obtaining 12 according to the formula 12=!! «X, obtaining a scale of 2 from the camera to the ceiling plane, where the normal line is represented by 1 in the vector of the reconstruction space, and X indicates that the camera is at The coordinates in the three-dimensional space.

[0118] In this embodiment

Abbreviation, which is an algorithm for obtaining valid sample data based on a set of sample data sets containing abnormal data, mathematical model parameters of the data. The RANSAC algorithm is often used in computer vision to solve the matching point problem of a pair of cameras and the calculation of the basic matrix in the field of stereo vision. The camera has a certain angle. The established three-dimensional space contains many points on the ceiling plane. The normal subunit 22221 is fitted with RANSAC. \¥0 2019/104741 卩(:17 \2017/114340 to calculate the plane of the ceiling plane in the straight three-dimensional space, the plane is represented in the three-dimensional space n. Then calculate the distance sub-unit 22222 in the three-dimensional space according to The normal calculates the distance from the ceiling to the camera of the visual sweeping robot, which is the distance T2 from the camera to the ceiling in three dimensions, which is the number of scales in the three-dimensional coordinate system, rather than the actual distance value. Specifically, normal n The correspondence relationship between the camera-to-ceiling distance T2 in the three-dimensional space is: T2=n ^T * _x , where x is the coordinate of the camera in the three-dimensional space, and n ^T represents the normal in the three-dimensional space. vector.

[0119] Referring to FIG. 11, further, the system for measuring a visual sweeping robot odometer compensation coefficient further includes

[0120] The multi-calculation device 5 is configured to calculate the distance between the normal n and the camera of the sweeping visual robot to obtain T2.

[0121] In this embodiment, after the compensation coefficient is calculated, the compensation coefficient inevitably has an error. Moreover, the error is a random error. To further reduce the error, the best method is to perform multiple experimental calculations. The multiple calculation device repeatedly repeats the above device operation calculation to obtain multiple compensation coefficients, and calculates the average of multiple compensation coefficients. The average value is used as the final compensation coefficient to improve the accuracy of the detection distance of the visual sweeping robot.

[0122] Further, the multiple calculation device 5 includes:

[0123] The ground module 51 is configured to calculate a plurality of compensation coefficients after the visual sweeping robot moves on the ground of different materials.

[0124] In the present embodiment, since the visual sweeping robot has to face different families, when measuring the compensation coefficient, various factors must be comprehensively considered, and thus different friction coefficients need to be considered. The general household floor includes various materials such as concrete floor, tile, wooden floor, and carpet. The ground module 51 measures the compensation coefficients on the floor with different friction coefficients of different materials, and then averages them to obtain a standard compensation coefficient.

The above description is only a preferred embodiment of the present invention, and is not intended to limit the scope of the invention, and the equivalent structure or equivalent process transformations made by the description of the present invention and the drawings are used directly or indirectly. Other related technical fields are equally included in the scope of patent protection of the present invention.

Claims

\¥0 2019/104741 卩(:17 \2017/114340 Claims

[Claim 1] A method of measuring a compensation coefficient of a visual sweeping robot odometer, characterized in that the ground and ceiling of the measurement environment are both horizontally arranged, the method comprising:

The visual sweeping robot moves a distance within the test environment while the camera captures the image of the ceiling;

Calculate the displacement of the actual movement of the visual sweeping robot based on the acquired image! _^ 2;

Obtain the displacement of the sweeping robot recorded by the odometer 1 _^ 1 ;

The compensation coefficient is calculated by dividing the displacement of the actual movement of the sweeping robot by 1 _^ 2 by the displacement of the sweeping robot recorded by the odometer 1 _^ 1.

[Claim 2] The method of visual odometry compensation coefficient cleaning robot according to claim 1, wherein, according to said image acquisition, the cleaning robot calculates the actual movement of the visual displacement step 2 comprises _^!:

According to the image captured by the visual sweeping robot, the feature points in the image are spatially reconstructed three-dimensionally, and the three-dimensional space coordinates of the feature points are obtained, thereby obtaining three-dimensional information of the image; and the three-dimensional information of the image is used to calculate the The displacement of the visual sweeping robot actually moves 2.

[Claim 3] The method for measuring a compensation coefficient of a visual cleaning robot odometer according to claim 2, wherein the calculating a displacement of the actual movement of the visual cleaning robot according to the three-dimensional information of the image is 1 _^ The steps of 2 include:

Obtaining the scale number of the moving distance of the visual sweeping robot in three-dimensional space II;

Calculating the scale information corresponding to each scale according to a preset rule, multiplying the scale number of the visual sweeping robot by the scale number 11 in the three-dimensional space by the scale information ₈

Calculate the displacement of the actual movement of the visual sweeping robot by 1 _^ 2 .

[Claim 4] The method for measuring the compensation coefficient of the visual sweeping robot odometer according to claim 3, wherein the step of calculating the scale information ₈ corresponding to each scale according to a preset rule comprises:

Obtained a scale of 2 from the camera to the ceiling plane;

Calculate the camera distance based on the known environmental information and the parameters of the visual sweeping robot itself. \¥0 2019/104741 卩(:17 \2017/114340 The actual height from the ceiling

The scale information is calculated by dividing the actual height of the camera from the ceiling by the number of scales from the camera to the ceiling plane.

[Claim 5] The method for measuring a visual sweeping robot odometer compensation coefficient according to claim 4, wherein the obtaining the camera-to-ceiling plane scale number 12 comprises the steps of:

Plane fitting estimates the ceiling plane to get the normal 11;

The scale number 12 of the camera-to-ceiling plane is calculated according to the formula 12=11 «X, where 11 represents the vector of the normal 11 in the reconstruction space, and X represents the coordinates of the camera in the three-dimensional space.

[Claim 6] The method for measuring a compensation coefficient of a visual cleaning robot odometer according to claim 1, wherein the calculating the compensation coefficient comprises the following steps:

A plurality of compensation coefficients are calculated repeatedly and the average of the plurality of compensation coefficients is calculated.

[Claim 7] The method for measuring a compensation coefficient of a visual cleaning robot odometer according to claim 6, wherein the step of repeatedly calculating a plurality of compensation coefficients comprises: the visual cleaning robots are respectively in different materials Calculate multiple compensation coefficients after moving on the ground

[Claim 8] A system for measuring a compensation coefficient of a visual sweeping robot odometer, wherein a ground and a ceiling of a measurement environment are horizontally disposed, the system comprising: a mobile device for a visual sweeping robot in the test Move a distance within the environment, while the camera captures the ceiling image;

The actual displacement device is configured to calculate the displacement of the actual movement of the visual sweeping robot according to the acquired picture! _^ 2;

An odometer displacement device for obtaining a displacement of the odometer recorded by 1 _^ 1 ;

The compensation coefficient device is used to calculate the compensation coefficient by dividing the displacement of the actual movement of the cleaning robot by 1 _^ 2 by the displacement of the cleaning robot recorded by the odometer 1 _^ 1.

[Claim 9] The system for measuring a calibrating coefficient of a glider according to claim 8, wherein the actual displacement device comprises:

a building module for taking pictures according to a visual sweeping robot, and characterizing the points in the picture \¥0 2019/104741 卩(:17 \2017/114340 performs spatial three-dimensional reconstruction, obtains the three-dimensional space coordinates of the feature points, and then obtains three-dimensional information of the image;

And a calculating module, configured to calculate, according to the three-dimensional information of the image, a displacement of the actual movement of the visual cleaning robot by 1 _^ 2.

[Claim 10] The system for measuring a calibrating coefficient of a scoping robot according to claim 9, wherein the calculating module comprises:

a moving scale sub-module for obtaining a scale number of the moving distance of the visual sweeping robot in three-dimensional space 11;

a length information sub-module, configured to calculate scale information corresponding to each scale according to a preset rule;

The calculation formula sub-module is used to multiply the scale number of the moving distance of the visual sweeping robot in the three-dimensional space by the scale information 8 to calculate the displacement of the actual movement of the visual sweeping robot 1 _^ 2

[Claim 11] The system for measuring a compensation coefficient of a visual sweeping robot odometer according to claim 10, wherein the length information sub-module comprises:

Obtaining a scale unit for obtaining a scale of 2 from the camera to the ceiling plane;

Calculate the height unit, which is used to calculate the actual height of the camera from the ceiling based on the known environmental information and the parameters of the visual sweeping robot itself.

The calculation formula unit is used to calculate the scale information 8 based on the actual height 11 of the ceiling from the ceiling divided by the number of scales 12 of the camera to the ceiling plane.

[Claim 12] The system for measuring a compensation coefficient of a visual sweeping robot odometer according to claim 11, wherein the obtaining the scale unit comprises:

Normal subunit, used to estimate the ceiling plane using RANSAC plane fitting, to get the normal

II;

Calculate the distance subunit, used to formulate

The number of scales from the camera to the ceiling plane is calculated to be 2, where 11 ⁷ represents the vector of the normal 11 in the reconstruction space, and X represents the coordinates of the camera in the three-dimensional space.

[Claim 13] A system for measuring a compensation coefficient of a finder oscilloscope according to claim 8, \¥0 2019/104741 卩(:17 \2017/114340 The levy is, it also includes:

A multi-calculation device is used to calculate the distance between the normal line and the camera of the sweeping visual robot.

[Claim 14] The system for measuring an odometer compensation coefficient of a visually-swept robot according to claim 13, wherein the plurality of calculation means comprises:

The ground module is used to calculate a plurality of compensation coefficients after the visual sweeping robot moves on the ground of different materials.