CN114131629B

CN114131629B - Ground detection robot

Info

Publication number: CN114131629B
Application number: CN202111505376.4A
Authority: CN
Inventors: 胡子俊; 徐行; 曹宏宇; 崔振富; 孙本楠; 许浩
Original assignee: Beijing Dongfang Measurement and Test Institute
Current assignee: Beijing Dongfang Measurement and Test Institute
Priority date: 2021-12-10
Filing date: 2021-12-10
Publication date: 2024-02-06
Anticipated expiration: 2041-12-10
Also published as: CN114131629A

Abstract

The invention relates to a ground detection robot which comprises a carrier (1), a data acquisition module (2), a sensing module and a control system (4), wherein the carrier (1) comprises a frame body (11) and a travelling mechanism (12), and the data acquisition module (2) and the sensing module are arranged on the carrier (1). The invention can realize the fusion development of artificial intelligence and metering service.

Description

Ground detection robot

Technical Field

The invention relates to a ground detection robot.

Background

Along with the development of artificial intelligence technology in fields such as natural language processing, machine learning, robot vision, robot technology and the like, the metering field is gradually integrated with an intelligent manufacturing theory, so that industrial manufacturing is gradually converted into intelligent manufacturing by utilizing the artificial intelligence technology and electronic informatization technology, and the problems of insufficient labor force and excessive cost are solved.

In the prior art, the implementation of metering automation mainly depends on industrial automation technology. However, the metering service has the characteristics of complex flow, fine operation, multiple scenes and the like, so that the existing industrial automation technology suitable for pipeline operation is difficult to flexibly adapt. Especially in static detection business, the static-free ground grounding detection of places such as factory buildings, warehouses, laboratories and the like is mainly finished manually by means of a laser range finder, namely, manually drawing a plan view, calculating detection points, marking points, removing obstacle points, moving a heavy hammer, recording detection results, drawing a hand-drawn plan view into a Vis io diagram, and recording detection records in a metering system. The prior automatic technology is difficult to complete the series of complicated manual operations. It can be seen that artificial intelligence technology applications in the metering field are yet to be developed.

Disclosure of Invention

The invention aims to provide a ground detection robot.

In order to achieve the above object, the invention provides a ground detection robot, which comprises a carrier, a data acquisition module, a sensing module and a control system, wherein the carrier comprises a frame body and a travelling mechanism, and the data acquisition module and the sensing module are both arranged on the carrier.

According to one aspect of the invention, the running gear comprises a chassis, a front wheel and a rear wheel;

the front wheel and the rear wheel are arranged on the chassis and are Mecanum wheels;

the chassis is provided with wheel motors connected with the front wheels and the rear wheels, and the wheel motors are independently controlled by corresponding motion control cards;

the front wheel is flexibly connected with the chassis through an elastic connecting rod;

the frame body is made of aluminum alloy.

According to one aspect of the invention, the device further comprises a ground resistance meter, a lifting mechanism and a camera;

the ground resistance instrument comprises a control screen and a heavy hammer, and the control screen is arranged in the frame body;

the lifting mechanism comprises a lifting motor, a cantilever and a pulley assembly;

the lifting motor is arranged on the chassis, one end of the cantilever is hinged on the chassis, and the other end of the cantilever is connected with the frame body through the pulley component;

One end of the cantilever far away from the chassis is provided with a connecting ring, and the heavy hammer is arranged in the connecting ring;

the limit position of the heavy hammer is limited by a travel switch;

the camera is arranged on the frame body and used for collecting screen data of the ground resistance meter.

According to one aspect of the invention, when the carrier runs to a test point in a working state, the lifting motor drives the cantilever to turn over to put down the heavy hammer to be in contact with the ground, the camera reads the screen reading of the ground resistance meter, and the screen reading is stored in the control system after being identified by the text identification software.

According to one aspect of the invention, the camera is connected with the control system through a USB interface to receive a controlled instruction and transmit snapshot frame data;

the character recognition software carries out AI visual recognition on the image shot by the camera according to the code pattern as a training basis;

the snapshot frame data of the camera is 3 times of the real record data;

the camera determines the shape of a screen display image of the ground resistance meter by detecting dark and bright modes, the shape is translated into computer characters by a character recognition method, the characters are converted into a text format by a character recognition software, and the text format is edited and processed by a character processing software;

Defining a number and letter template by adopting a template matching method, and sliding and matching characters on the ground resistance meter by utilizing the number and letter template;

the character recognition software converts the image acquired by the camera into a black-and-white image through filter processing, cuts the black-and-white image into individual character images, compares the individual character images with potential images, and judges the optimal possibility value.

According to one aspect of the invention, the data acquisition module comprises a height-adjustable support platform, a laser ranging module and a rotating mechanism;

the laser ranging module comprises a laser ranging probe, a modulator, a distance display, a phase discriminator and a reflecting mirror;

the rotating mechanism comprises an RS232 circuit board, a stepping motor, an STM32 circuit board, an encoder, a motor driver, a connector and a collector ring;

two ends of the connector are respectively connected with the output end of the stepping motor and one end of the collecting ring,

the other end of the collecting ring is connected with the support of the laser ranging probe.

According to one aspect of the invention, two laser ranging probes are arranged and are arranged at an included angle of 90 degrees in the horizontal direction;

The laser ranging probe adopts a TOF ranging mode to carry out multipoint sampling measurement, and the maximum detection distance is within 30 meters;

the measuring precision of the data acquisition module is the straight line precision of five ten thousandths to one thousandth, the angle precision is 3-5mrad, the radio frequency is used for carrying out amplitude modulation on laser, the phase delay generated by the back and forth measuring line of modulated light once is measured, the distance represented by the phase delay is converted according to the wavelength of the modulated light, and the time required for the light to pass through the back and forth measuring line is measured;

if the modulation light angle frequency is omega, the phase delay generated by one round trip on the distance D to be measured is phi, and the corresponding time t is: t=Φ/ω, substituting into the distance D:

D＝1/2ct＝1/2c·φ/ω＝c/(4πf)(Nπ+Δφ)＝c/4f(N+ΔN)＝U(N+)；

wherein phi is the total phase delay generated by the signal round trip line at one time; ω is the angular frequency of the modulated signal, ω=2pi f; u is unit length, and the value is equal to 1/4 modulation wavelength; n is the number of modulation half-wavelengths contained in the measuring line; delta phi is the part of the signal round trip line where the phase delay is less than pi at a time; Δn is the fractional part of the modulated wave contained in the line less than half a wavelength, Δn=Φ/ω;

under the given modulation and standard atmospheric conditions, the frequency c/(4pi f) is a constant, and the distance is measured by the half-wavelength number N and the fractional part delta N less than half the wavelength contained in the measuring line;

If the edge length of the field is not more than 60 meters, drawing a plan view of the field at each sampling point at four corners;

if the side of the field is longer than 60 meters, on the basis of measuring four angle sampling points, more than one sampling point is added in the middle of each side for measurement;

setting a subdivision number control step angle of the stepping motor, wherein each step of operation of the stepping motor is paused for a period of time, the laser ranging probe finishes one ranging during the period, and the carrier automatically operates to the next sampling point;

during ranging, the laser ranging probe scans for obstacles on the travel path in real time.

According to one aspect of the invention, the data acquisition module performs positioning navigation and map construction by SLAM, and comprises: feature extraction, data association, state estimation, state update, and feature update;

the step of mapping the plant plan by SLAM includes:

optimizing the original data acquired by the data acquisition module, and eliminating problem data;

searching a corresponding position on an established map by using point cloud data of a current local environment, and matching and splicing the point cloud currently acquired by the data acquisition module into an original map;

And splicing the new data acquired by the data acquisition module in the round into the original map to finish updating the map.

According to one aspect of the invention, the sensing module comprises a laser ranging sensor, a multi-axis acceleration sensor and a geomagnetic sensor;

parameters of the laser ranging sensor comprise detection distance, length precision, angle precision, environmental influence, sampling frequency and data throughput rate;

the angle precision of the multi-axis acceleration sensor is X, Y static 0.05 degrees, dynamic 0.1 degrees and the Z axis is 1 degree; the baud rate is 2400-921600bps; the voltage is 9V-36V; the gyroscope is +/-2000 degrees/S;

the geomagnetic sensor and the laser ranging probe return data in real time, whether the running track of the carrier is deviated or not is obtained, and when the track of the carrier is deviated, the movement of the carrier is corrected by the wheel motor;

the geomagnetic sensor has a resolution of 13nt@200Cycle Count.

According to one aspect of the invention, the control system is arranged in the frame body and comprises an upper computer and a lower computer;

the upper computer is used for generating and displaying a map, generating an automatic patrol route according to the map and identifying the display content of the ground resistance meter;

The lower computer comprises an STM32 embedded system and a sensor main control board and is used for controlling actions of the carrier and the sensing module, collecting laser mapping data and preprocessing initial data;

the lower computer is provided with a corresponding instruction system and is used for converting a program instruction installation protocol into an instruction of the lower computer and analyzing the return information of the lower computer;

the upper computer passes through the data structure calibrated by the lower computer and carries out subsequent processing;

the subsequent processing comprises smooth generation of a plane graph, generation of an automatic measurement path and display and export of final data;

the construction principle of the data structure of the lower computer is that plane mapping data is stored in a polar coordinate form; storing side length data with the total side length exceeding 5000 meters by using flash, and automatically pushing the side length data to a memory of the upper computer for storage if the side length data exceeds the storage; the cache rom adopts an excessive large cache;

the distance and position data output in the measuring process are shared as input data in the control process, the state data output in the control process are shared as time node judgment basis in the measuring process, and the data output in the measuring process and the control process are unified into final output result data and log data for providing calibration and reference for the next measurement;

The lower computer is communicated with the data acquisition module by adopting a serial port, a UART or a TTL interface, the lower computer is communicated with the upper computer by adopting a network port, and the minimum transmission rate of control commands and acquired data of the camera is 0.8Mb/s-1.2Mb/s;

in the mapping stage and the displacement adjustment stage, the wireless remote control equipment is utilized to control the walking function;

the coordination communication of the data among the sensors is realized by directly calling the mutually compatible data;

the upper computer establishes an environment map of the track scanned and retrieved by the data acquisition module, optimizes the image acquired by the data acquisition module by using a position map, and sets a stable posture correction and full-right angle movement algorithm and an independent map building module;

the establishment of the environment map comprises the steps of special environment scanning map establishment, special drawing point measurement and special map management;

the specialized environment scanning mapping comprises:

the data acquisition module completes initial scanning and converts an image into a matrix image, judges a wall straight line appearing in the matrix image, adjusts an image angle according to the wall angle, enables the carrier to walk along an indoor wall for one circle in a clockwise direction, and further stops and generates a current image acquired by the data acquisition module before each carrier, compares the current image with a main map, judges a current coordinate, and blends the current image into the main map according to the current coordinate; wherein, a special right-angle action and a straight line deviation correction algorithm are adopted for drawing;

The specialized plot measurement includes:

generating a measuring point set in a full map range, judging and going to a first measuring point according to the current position of the carrier, determining a current coordinate after the carrier arrives, setting down the heavy hammer, shooting a screen of the ground resistance meter by using the camera, identifying a reading value, and if the reading value is correct, going to the next measuring point by the carrier; the distance between the measuring points is 1-5 meters, and the periphery of the map surrounds the wall;

the specialized map management includes:

and comprehensively managing map data, integrating the map scanned by the data acquisition module into a main map, and judging the included angle between the current image acquired by the data acquisition module and the main map by comprehensive geomagnetic information to obtain the coordinate positioning of the carrier.

According to the conception of the invention, an antistatic ground grounding detection robot is provided, electrostatic detection service is taken as a break, and the technologies of laser navigation, visual analysis, image recognition and the like are introduced. Based on the metering detection technology, the full-automatic antistatic ground detection is realized by combining intelligent elements such as a robot arm, machine vision, a sensor, logistics storage and the like, the fusion development of artificial intelligence and metering business is realized, and a set of intelligent detection equipment comprising a robot, a laser radar, a camera, a lifting arm and a central console is formed. The device can automatically draw a place map through the laser radar and the camera of the intelligent robot during primary training, the robot can automatically move according to a planned travel route and stay at each detection point, after detection is completed, numbers on the ground resistance testing instrument are read and recorded through image recognition, and then the robot continues to move until all detection points are finished.

According to one scheme of the invention, the travelling mechanism utilizes the aluminum alloy frame body, and can meet high strength and heavy load and reduce the dead weight of the body through reasonable structural design. The front and rear wheels all use Mecanum wheel technology, can realize the motion modes such as going forward, sideslip, oblique running, rotation and combination thereof, are suitable for operation in the environments with limited transfer space and narrow operation channels, can effectively improve the efficiency, increase the space utilization rate and reduce the labor cost. On the basis, the front wheels are flexibly connected with the chassis through the elastic connecting rods, so that when pits and protrusions are arranged on the running track of the carrier, the four wheels of the carrier can be ensured to land, and the stable running and accurate track of the carrier are ensured.

According to one scheme of the invention, the robot is provided with the lifting mechanism of the ground resistance measuring heavy hammer, so that the heavy hammer is driven by the lifting mechanism to fall into a test point for measurement when in measurement, and data is stored in the upper computer after OCR recognition. And the heavy hammer can be retracted when in idle so as to reduce the whole volume of the robot, thereby being convenient for moving.

According to one scheme of the invention, the motion driving of the robot is realized by utilizing a wheel driving direct current motor, a heavy hammer lifting direct current motor and a range finder rotating stepping motor, and all the motion motors are controlled by a 32-bit high-speed microprocessor. The four wheel motors are independently controlled by the four motion control cards, so that when the carrier is over-bent, the over-bending radius and speed can be accurately controlled by controlling the rotating speed of each motor. And even if the situation of hollowing, protruding and the like is encountered in the route, the carrier can also realize the purpose of automatically adjusting the rotating speed through the resistance feedback of each wheel, thereby enabling the carrier to stably pass. The four-wheel independent control mode is matched with the arrangement of the geomagnetic sensor and the laser ranging head, so that the accuracy of the motion trail of the carrier can be realized.

According to one scheme of the invention, the lifting mechanism of the heavy hammer is driven by the direct-current gear motor, so that when the carrier runs to a test point, the lifting motor can drive the cantilever to put down the heavy hammer, and after the heavy hammer is completely contacted with the ground, the reading of the measuring instrument is started to be read, and the upper limit position and the lower limit position of the heavy hammer are both ensured to stop by the travel switch.

According to one scheme of the invention, the rotation of the laser ranging probe is driven by adopting the stepping motor, so that the stepping angle of each step of motor operation can be controlled by setting the subdivision number of the stepping motor, the ranging motor can be suspended for a certain time every step of operation, and the laser head can conveniently finish ranging. Therefore, the motor rotates the laser head for one circle to finish one circle of measurement, so that the measurement of the sampling point is realized, and the carrier can automatically run to the next sampling point for measurement again.

According to one scheme of the invention, the motion of the carrier adopts a correction technology, the moving track of the carrier can be known whether to deviate or not through the geomagnetic sensor and the laser ranging probe which return data in real time, and when the carrier deviates, the motion of the carrier can be corrected by controlling the four driving wheel motors.

According to one scheme of the invention, an active obstacle avoidance technology is adopted, and an obstacle on a running path is scanned in real time through a laser ranging probe. Thus, when the vehicle approaches the obstacle, the obstacle can be avoided through motion control.

According to one scheme of the invention, aiming at the laser mapping part, the robot is provided with two high-speed laser ranging probes, and the two probes are arranged at an included angle of 90 degrees in the horizontal direction, so that measurement data in two directions can be obtained simultaneously, and the finally obtained plan view is more accurate. And a TOF ranging mode is adopted, namely ranging is performed aiming at the flight time difference of laser, and the characteristics of extremely short laser pulse duration, relatively concentrated energy in time and larger instantaneous power are utilized for ranging. The method has the advantages that a factory floor plan is drawn by adopting a multipoint sampling measurement mode, the detection distance of the laser ranging head is sacrificed, the maximum detection distance of the laser ranging probe is set to be 30 meters from the original 100 meters, and rapid measurement is realized by matching with a mode of increasing sampling points, so that the mapping time of each sampling point can be greatly reduced, the error in long-distance measurement is avoided, and the overall measurement speed and the overall measurement precision can be improved.

According to one scheme of the invention, for a field with the side length not more than 60 meters, a plan view of the field can be drawn only by selecting one sampling point at each of four corners at most; for a field with a side longer than 60 meters, not only four corners of data are measured, but also one or more sampling points are added in the middle of each side for measurement.

According to one scheme of the invention, aiming at an OCR image recognition part, a high-definition camera is adopted to shoot a ground resistance instrument screen, the shape of the ground resistance instrument screen is determined by detecting dark and bright modes, the shape is translated into computer characters by a character recognition method, and the characters in the image are converted into a text format by recognition software for further editing and processing by word processing software. As the fonts on the ground resistance meter are fewer (only Arabic numerals and individual English letters), and the fonts are unified and the definition is good, for the identification scene, a template matching method is adopted, firstly, a number template and a letter template are defined, and then the template is used for sliding and matching characters on the ground resistance meter, so that only a template library is needed to be made before identification, and the identification performance is improved. The character recognition software has a character cutting function, and characters to be recognized can be cut out according to the characters, so that the phenomenon of character adhesion and pen breakage caused by the limitation of photographing conditions is avoided, and the performance of a recognition system is improved.

Drawings

Fig. 1 is a schematic view showing a structure of a ground detection robot according to an embodiment of the present invention;

FIG. 2 schematically shows a Pytorch model training process diagram;

FIG. 3 schematically shows an effect diagram of final recognition and export of a ready-to-use excel document;

FIG. 4 is a schematic diagram of a laser ranging method according to an embodiment of the present invention;

FIG. 5 schematically illustrates a parameter diagram of a multi-axis acceleration sensor according to an embodiment of the present invention;

FIG. 6 is a schematic diagram showing parameters of a geomagnetic sensor according to an embodiment of the present invention;

FIG. 7 is a schematic view showing a system hardware and software architecture of a ground detection robot according to an embodiment of the present invention;

FIG. 8 is a schematic diagram showing the integration of hardware and software control of the upper computer and the lower computer of the ground detection robot according to an embodiment of the invention;

FIG. 9 schematically shows a pretreatment effect of an embodiment of the present invention;

FIG. 10 schematically illustrates a matching effect graph of an embodiment of the present invention;

FIG. 11 schematically illustrates a map fusion of an embodiment of the present invention;

FIG. 12 schematically illustrates a plan view of an area drawn during field testing of one embodiment of the present invention;

FIG. 13 is a schematic diagram of a vehicle movement sequence according to an embodiment of the present invention;

FIG. 14 schematically illustrates a measurement point of an embodiment of the present invention;

fig. 15 schematically shows a flow chart of a ground detection robot according to an embodiment of the present invention in comparison with a prior art manual measurement.

Detailed Description

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required to be used in the embodiments will be briefly described below. It is apparent that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained from these drawings without inventive effort for a person of ordinary skill in the art.

In describing embodiments of the present invention, the terms "longitudinal," "transverse," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," and the like are used in terms of orientation or positional relationship shown in the drawings for convenience of description and simplicity of description only, and do not denote or imply that the devices or elements in question must have a particular orientation, be constructed and operated in a particular orientation, so that the above terms are not to be construed as limiting the invention.

The present invention will be described in detail below with reference to the drawings and the specific embodiments, which are not described in detail herein, but the embodiments of the present invention are not limited to the following embodiments.

Referring to fig. 1, the antistatic ground detection robot of the invention comprises a carrier 1 (trolley), a data acquisition module 2, a sensing module and a control system 4 (or industrial control host/center console). The carrier 1 comprises a frame body 11 and a travelling mechanism 12, and the data acquisition module 2, the sensing module and the control system 4 are all arranged on the carrier 1. Therefore, the parts of the robot are arranged in a modularized form, so that the parts can be quickly communicated or built into a whole, and the work can be quickly performed. The frame body 11 is made of an aluminum alloy.

In the present invention, running gear 12 includes chassis 121, two front wheels 122, and two rear wheels 123. The front wheel 122 and the rear wheel 123 are arranged at the bottom of the chassis 121 and are Mecanum wheels. The bottom of the chassis 121 is provided with wheel motors 124 connected to each of the front wheels 122 and the rear wheels 123, and each wheel motor 124 is individually controlled by a corresponding motion control card. The front wheel 122 is flexibly connected to the chassis 121 by an elastic connecting rod.

The ground detection robot of the present invention further comprises a ground resistance meter 5, a lifting mechanism 6 (or tilting mechanism/escapement mechanism) and a camera. The earth resistance meter 5 includes a control panel provided in the frame body 11 and a weight 52. The lifting mechanism 6 includes a lifting motor 61, a cantilever 62 and a pulley assembly 63. The elevating motor 61 is provided on the chassis 121, and one end of the cantilever 62 is hinged to the chassis 121, and the other end is connected to the frame body 11 through the pulley assembly 63. The end of the cantilever 62 away from the chassis 121 has a connecting ring 621, and the weight 52 is disposed in the connecting ring 621. The limit position of the weight 52 is limited by the travel switch. The camera is arranged on the frame body 11 (A position) and is used for collecting screen data of the ground resistance meter 5, and the control system 4 is used for carrying out AI automatic identification so as to realize complete automatic measurement. In the working state, when the carrier 1 runs to the test point, the lifting motor 61 drives the cantilever 62 to turn over to put down the heavy hammer 52 to be in contact with the ground, the screen reading of the ground resistance meter 5 is read by the camera, and the screen reading is stored in the control system 4 after being identified by the text identification software.

Therefore, the present invention uses the small and miniature all-terrain traveling mechanism, i.e. the Medahm wheel, which is not commonly used for the four-wheel vehicle, because the load of the vehicle 1 is heavy and the weight 52 for escapement measurement needs to be accurate under the stable condition, so that the traveling mechanism 12 has the omnidirectional traveling chassis support, and is firm and durable. Meanwhile, the travelling mechanism 12 is large in load and stable in braking, and the movement deviation correction cannot travel an arc-shaped route, so that the travelling mechanism can return to a correct path in a straight line. Of course, after the mecanum wheel is reused, the motion algorithm of the carrier cannot be multiplexed with the normal four-wheel algorithm. Therefore, on the basis of the motion control and walking algorithm, the invention constructs a set of control algorithm suitable for the motion characteristics of the invention, the upper computer gives the action parameters, and the lower computer directly controls the high torque motor so as to realize the accurate control of the motion of the carrier. In addition, the walking mechanism 12 also performs full and accurate calculation, design and experiment in the aspects of material selection, physical strength, action reliability, correction accuracy, horsepower strength, power consumption loss and the like.

In the invention, the carrier 1 is responsible for bearing the lower computer, the corresponding mechanical transmission mechanism, the lifting mechanism and the like, so that the carrier needs to have certain anti-seismic firm characteristics. In the present embodiment, the load carrying capacity of the carrier 1 is set to 70-120 kg as described above. The robot of the present invention belongs to a precision mapping system, so the carrier 1 should have high stability. Of course, since the present invention has the laser ranging probe 221 rotating at a high speed, the carrier 1 needs to have not only stability in horizontal movement but also centrifugal stability in a vertical direction. In the rotation performance of the chassis 121, the present invention needs to design a gravity center immobilization technology such as rotation of four wheels around the center and double-track anti-phase rotation of the track chassis. Meanwhile, a timely deviation correcting algorithm is also made on the adopted inertial sensor data, so that the vehicle 1 is accurately controlled to walk.

Referring to fig. 2 and 3, the ground resistance meter 5 of the present invention is a hand-held meter, specifically a hand-held TREK152-1 surface resistance tester. The character recognition software performs AI visual recognition based on the code pattern as a training basis. Specifically, the camera determines the shape of the screen display image of the ground resistance meter 5 by detecting dark and bright modes, then translates the shape into computer characters by using a character recognition method, converts the characters into a text format by using a character recognition software, and finally edits and processes the characters by using a character processing software. The invention adopts a template matching method to define a number and letter template, and uses the number and letter template to slide and match the characters on the earth resistance meter 5.

In order to obtain a perfect fitting result, the main difficulty of the AI recognition technology is to collect massive screenshots for training and make parameter adjustment for a model for training the screenshots. Because the identified object (the ground resistance instrument 5) is handheld, the instrument generally has the characteristics of older model, no backlight, single color, unadjustable contrast and rapid jump, and the characteristics belong to environmental model parameters, and when the environmental model parameters exceed more than 3 types, AI training can be regarded as no solution. Therefore, the invention specifically improves the characteristics of simplification of the ground resistance meter 5, namely, simplifies the recognition mode based on the training of points, establishes the recognition mode based on the training of code patterns after a large number of experiments, and redevelens the recognition environment and adjusts the recognition algorithm.

The improvement is mainly characterized in that the image acquired by a camera is converted into a black-and-white image through filter processing by utilizing character recognition software (OCR.cs), the black-and-white image is cut into individual character images, and then the individual character images are compared with potential images to judge the optimal possibility value. The APIs in this way are: getScares (bool [, ] image) compares the target picture with the original picture of each character, and returns each comparison similarity value; getBustScore (bool [, ] image) compares the target character picture with the original picture of each character, and returns the most similar original picture and similarity score; list < Bitmap > gettext () cuts a black-and-white image into individual text images.

Besides the standardized development AI identification method and action, the invention also carries out customized parameter adjustment, algorithm adjustment and instrument cabin physical adjustment according to the special difficult items of the instrument such as liquid crystal screen reflection degree, definition, liquid crystal screen delay, display shake, screen backlight-free and the like, and finally achieves the aim that the identification accuracy is infinitely close to 100%.

Referring to fig. 4, the data acquisition module 2 of the present invention comprises a support platform, a laser ranging module 22 and a (mechanically controllable) rotation mechanism for acquiring a plant plane in a plant mapping need, and thus may also be referred to as a lidar/laser 2D planar mapping system. The laser ranging module 22 includes a laser ranging probe 221, a modulator 222, a distance display 223, a phase detector 224, and a mirror 225. The rotating mechanism comprises an RS232 circuit board, a stepping motor, an STM32 circuit board, an encoder, a motor driver, a connector and a collector ring. The two ends of the connector are respectively connected with the output end of the stepping motor and the left end of the collector ring, and the right end of the collector ring is connected with the bracket B of the laser ranging probe 221.

According to the invention, the height of the supporting platform is adjustable, so that the measuring height is adjustable, and the device is suitable for different painting of factory enclosing walls, and accidental factors influencing mapping, such as the height of windows in the walls, the door height, a fireproof box and the like. In addition, the laser ranging probes 221 are provided in two and are disposed at an angle of 90 degrees in the horizontal direction. Therefore, 2 ranging laser beams with 90-degree phase included angles can be formed, so that mapping accuracy and mapping speed are improved. The method can greatly improve the working efficiency and the precision on important indexes such as mapping time, mapping stability and the like. Of course, the number and arrangement of the laser beams can be customized according to the requirement, and the corresponding mechanical structural design and the coordination of the lower computer data acquisition algorithm are also adaptively adjusted. In addition, the laser ranging probe 221 of the present invention performs multi-point sampling measurement by using the TOF ranging method, and the maximum detection distance is within 30 meters.

Because the robot of the invention has special application fields, a laser device with higher physical precision and a laser ranging algorithm with a high-precision phase method are required. In contrast, in the invention, the measurement precision of the data acquisition module 2 is the straight line precision of five to one thousandth, and the angle precision expressed by polar coordinate data is 3-5mrad. Also, incorporating precision sensors (such as inertial and geomagnetic sensors, etc.) on the rotating algorithm in motion focuses on accuracy rather than fuzzy avoidance. The high-precision phase method uses radio frequency to carry out amplitude modulation on laser, and measures the phase delay generated by the back and forth line of modulated light once, and then converts the distance represented by the phase delay according to the wavelength of the modulated light, namely, the time required by the light passing through the back and forth line is measured by an indirect method.

If the modulation light angle frequency is omega, the phase delay generated by one round trip on the distance D to be measured is phi, and the corresponding time t is: t=Φ/ω, substituting this relationship into the distance D:

D＝1/2ct＝1/2c·φ/ω＝c/(4πf)(Nπ+Δφ)＝c/4f(N+ΔN)＝U(N+)。

wherein phi is the total phase delay generated by the signal round trip line at one time; ω is the angular frequency of the modulated signal, ω=2pi f; u is unit length, and the value is equal to 1/4 modulation wavelength; n is the number of modulation half-wavelengths contained in the measuring line; delta phi is the part of the signal round trip line with phase delay less than pi; Δn is the fractional part of the modulated wave contained in the line less than half a wavelength, Δn=Φ/ω; c is the propagation speed of light in the atmosphere; f is the hertz frequency. Under given modulation and standard atmospheric conditions, the frequency c/(4pi f) is a constant, where the measurement of distance becomes a measurement of the number of half wavelengths contained in the line and a measurement of the fraction of less than half wavelengths, i.e., N or ΔN. And because of using precision machining and radio phase measurement technology, the phi measurement precision is higher.

If the edge length of the field is not more than 60 meters, drawing a plan view of the field at each sampling point at four corners; if the field side is longer than 60 meters, more than one sampling point is added in the middle of each side for measurement on the basis of measuring four angular sampling points. By setting the subdivision number control step angle of the stepping motor, the laser ranging probe 221 completes one ranging during each step of the operation of the stepping motor for a period of time, and the carrier 1 automatically operates to the next sampling point. During ranging, the laser ranging probe 221 also scans for obstacles on the travel path in real time.

Referring to fig. 5 and 6, in the present invention, the sensing module includes a laser ranging sensor, a multi-axis acceleration sensor, and a geomagnetic sensor. Parameters of the laser ranging sensor include detection distance, length accuracy, angle accuracy, environmental impact, sampling frequency, and data throughput rate. Specifically, the model is SK-Z-5; the working voltage is DC5V; the working current is 0.5A; the data interface is a TTL serial port; the ranging mode is TOF; the applicable scene is indoor/outdoor; no reflector is provided; the light source is 780nm infrared laser; the measurement frequency is 100Hz (MAX); the dead zone is 6cm. The multi-axis acceleration sensor is mainly used for improving the measurement precision in motion, so that the situation that the measurement can only be carried out in static state can be avoided, the mapping time is greatly reduced, and the main parameters include that the angle precision (X, Y axis sensitivity) is X, Y static 0.05 degrees, the dynamic state is 0.1 degrees and the Z axis is 1 degree; the baud rate is 2400-921600bps; the voltage is 9V-36V; the gyroscope is +/-2000 degrees/S; other parameters should also be focused on the parameter selection in fig. 5, where the selection of angular accuracy and baud rate is of paramount importance for moving vehicles and high-speed rotating 2D laser mapping institutions. The geomagnetic sensor is used for improving measurement accuracy in motion, and can also avoid that the geomagnetic sensor can only measure in static state, so that mapping time is greatly reduced. The use scene of the invention is basically in an indoor environment with a roof, so that GPS satellite signals are not completely reliable in monitoring short-distance position change, and a dynamic mapping mode (mapping in walking) is needed, so that the linear walking of the carrier 1 becomes an important guarantee for improving mapping precision and reducing fitting calculation load. Therefore, the geomagnetic sensor with high accuracy and resolution is selected, the resolution is 13nt@200Cycle Count, and other parameters are shown in fig. 6. In this way, the geomagnetic sensor and the laser ranging probe 221 return data in real time, so that whether the running track of the carrier 1 is deviated or not can be known, and when the track of the carrier 1 is deviated, the wheel motor 124 can be controlled to correct the movement of the carrier 1.

Referring to fig. 7 and 8, in the present invention, the control system 4 is disposed in the frame body 11, and includes an upper computer and a lower computer, which are mechanically and electrically controlled by means of integrated software and hardware control. The upper computer comprises a PC (self-contained display) with an operating system, and is used for generating and displaying a mapping (plane) diagram, generating an automatic patrol route according to the mapping diagram, displaying contents of the AI-identified ground resistance meter 5 and other tasks needing large-scale intelligent processing. The lower computer comprises a high-speed STM32 embedded system and a related sensor main control board and is used for small-scale calculation tasks with relatively strong real-time performance, such as action control of the carrier 1 and the sensing module, acquisition of laser mapping data, preprocessing of initial data and the like.

The data structure and the communication protocol of the invention enable the data structure to be universal on the lower computer and the upper computer and finally output as a specified data structure. Specifically, the construction principle of the data structure of the lower computer is that plane mapping data is stored in a polar coordinate form, flash of the lower computer needs to be enough to store side length data with total side length exceeding 5000 meters, if the side length data exceeds the storage, the side length data is automatically pushed to a mass memory of the upper computer to be stored, and meanwhile, the cache rom adopts an excessive large cache to increase the measurement speed and stability. The specific mapping data are drawn to include ({ a rad, b length, c time, d ComdResp, burner STATE, edge STATE … }) and the like, wherein the latter bits are typical flag bits and STATE bits, and the post-processing of an upper computer can be greatly and accurately accelerated. The degree of freedom of the upper computer in data processing is greatly increased, and specifically, the upper computer passes through the data structure calibrated by the lower computer and performs corresponding subsequent processing. The subsequent processing of the invention comprises the smooth generation of the plane graph, the generation of the automatic measurement path and the display and the export of the final data.

Therefore, the invention adopts the integrated communication and instruction (lowermachine. Cs) of the lower computer (embedded control system). In addition, because the specific time sequence rule and data rule of the upper computer (PC software) part and other special requirements under the condition of data which are not seen, for example, the laser plate data of the via slip ring are transmitted to the singlechip core plate, and byte correction commands are required to be designed for eliminating the accidental jump interference of the via slip ring on digital signals, the embedded control system does not use a general communication command library, particularly customizes and unifies the communication and commands with the lower computer, and independently forms a set of command system penetrating through the upper computer and the lower computer. The system is mainly used for converting the program instruction installation protocol into the lower computer instruction and analyzing the return information of the lower computer. The API of this algorithm is Start (): starting up; stop (): shutting down; write (string strn): writing into a serial port; check (SerialPortport): checking whether a serial port is opened; processresponse (SerialPort port, string strn): processing the return data of the lower computer; startMovementCheckThread (lower computer Status action): starting a timeout detection thread; forward (int distance): advancing; backward (int distance): backing; leftward (int distance): left shift; rightward (int distance): right-shifting; turn (int degree): turning; setSpeed (int speed): setting a speed; updateAngle (): updating the angle; updateBattary (): updating the battery; drop (): placing a heavy hammer; lift (): lifting a heavy hammer; stopMoving (): the movement is stopped. The communication is set to have a serial baud rate of 115200, no parity, and 1 stop bit. The instruction table is shown in table 1 below:

TABLE 1

In the invention, the main tasks of the working stage of the robot are measuring and outputting data, and the tasks of the robot include walking control, action control, measurement calculation, displacement calculation, data preprocessing, screen recognition, data fusion, data conversion, data output and the like. In order to perfectly realize the input output quantity required by a user and the input output quantity of different task stages of the vehicle in a proper computing resource space, and accurately operate various complex operations such as wall-following ranging, fixed-point regression, parking measurement, video photographing, image identification confirmation, next-path point tracking and the like at the same time. The invention adopts the idea of integrating measurement, control and output. The key point is that the calculation resources are moved upwards, namely, the calculation capacity of the upper computer is used as much as possible to process various process quantities and final quantities, the lower computer does not participate in complex calculation as much as possible and is only responsible for simple movement, and the design architecture is also an important guarantee for providing reliability. Specifically, in this embodiment, distance and position data output in the measurement process are shared as input data in the control process, state data output in the control process is shared as a time node judgment basis in the measurement process, and various data output in the measurement process and the control process are unified into final output result data and log data. Wherein the log data is used to provide calibration and reference for the next measurement.

The communication system is arranged in such a way that interfaces such as a serial port, a UART or a TTL are selected for communication according to the load capacity of the lower computer, the sampling frequency and the communication rate of the collector on a communication interface between the lower computer (STM 32 is to be adopted) and the data acquisition module 2 between the lower computer and the data acquisition module 2. Between the lower computer and the upper computer, the lower computer comprises: the mapping position data, the sensor data, the control data, the self-checking data and the state data are communicated with the upper computer, so that a communication channel of the lower computer and the upper computer adopts a communication interface with a larger bandwidth, namely, a network port communication mode is adopted to realize data communication. Between the ground resistance meter 5 and the upper computer, the screen data of the ground resistance meter 5 is captured by the camera, so that the controlled instruction of the camera and the captured frame data are connected with the upper computer of the control system 4 by the USB interface to receive the controlled instruction and transmit the captured frame data. The snap frame data of the camera is 3 times of the real recorded data, and is determined by an intelligent identification algorithm of the picture AI, and is judged according to the screen pixels of the class LCD1602 of the handheld ground resistance meter 5. The minimum transmission rate of the fixed-focus USB camera containing control commands and collected data is 0.8Mb/s-1.2Mb/s. The invention uses wireless remote control equipment to control walking in mapping stage and displacement adjustment stage (including loading and unloading of the carrier 1, etc.), so the invention also carries wireless remote control equipment to control walking function, thus the invention also needs to correspondingly set the optional communication system of the remote control equipment. The invention carries various sensor data and has universal data bus interface, so that the compatible data can be directly fetched between different sensor devices by optimized algorithm, thereby realizing high-efficiency processing.

In the present invention, in the 2D laser mapping and plan view generation (grid map) mode of the data acquisition module 2, positioning navigation and map construction are performed by using SLAM (Simul taneous Localization And Mapping, synchronous positioning and map construction), including: feature extraction, data association, state estimation, state update, feature update, etc., a corresponding method may be selected for each portion. Therefore, the problems of positioning navigation and map construction when the mobile robot operates in an unknown environment can be solved by utilizing the SLAM mapping technology.

In the invention, a relatively accurate factory building plan is automatically mapped by utilizing a laser SLAM scanning mapping technology, thereby providing a foundation for subsequent automatic operation. Since the lidar can only acquire the environmental information of the location of the lidar at a certain moment, the lidar can only reflect a part of the environment of the device. Therefore, as shown in fig. 9, the 2D laser mapping of the present invention is first preprocessed, that is, the original data collected by the data collection module 2 is optimized, and some problematic data are removed, which may be called filtering. Then, matching can be performed, as shown in fig. 10, the point cloud data of the current local environment is found out to a corresponding position on the established map, and the point cloud data acquired by the data acquisition module 2 is matched and spliced into the original map. Finally, map fusion is carried out, as shown in fig. 11, new data acquired by the data acquisition module 2 in this round are spliced into an original map, and the update of the map is completed, and the process is accompanied with a 2D laser mapping process in real time. Therefore, 2D laser mapping is complex, more probability algorithms are needed, and fusion is performed in a filtering mode. The above process is then performed successively, and the desired grid map is finally generated, as shown in the area plan drawing at the time of the field test shown in fig. 12.

In the invention, the upper computer is used for realizing the automatic walking path generation of the patrol ground resistance, namely, an environment map of the track of the equipment is scanned and retrieved by using the data acquisition module 2 is established, and a series of images acquired by the data acquisition module 2 are optimized by using the position map, so that synchronous positioning and mapping algorithm is realized. The robot provided by the invention is provided with a special SLAM and a path generation algorithm, wherein the SLAM laser scanning algorithm is additionally provided with a stable posture correction and full-right-angle movement algorithm aiming at a large-area factory building. In addition, in order to ensure reliability, independent mapping modules are independently designed and developed so as to reduce the dependence on other computing resources of the system and the possibility of interference as much as possible. The environment map is built by special environment scanning and mapping, special drawing point measurement and special map management.

The functions of the specialized environment scanning profiler (mapnavigator. Cs) include: firstly, the data acquisition module 2 completes initial scanning and converts an image into a matrix image, a straight line of a wall appearing in the matrix image is judged, the image angle is adjusted according to the wall angle, the carrier 1 is made to walk along an indoor wall for one circle in a clockwise direction, the carrier 1 is further stopped before each time and generates a current image (namely a radar image) acquired by the data acquisition module 2, then the current image is compared with a main map (MapManager. Cs) and the current coordinate is judged, and the current radar image is fused into the main map according to the current coordinate. The moving sequence of the carrier 1 is shown in fig. 13, and in the robot drawing, the special right-angle action and straight line deviation correction algorithm matched with the Mecanum omnidirectional chassis is adopted to replace the common four-wheel movement overlap measurement algorithm with burr data to be used as the moving algorithm of the carrier 1 for drawing, so that the robot drawing has high efficiency and high accuracy. The API of this algorithm (mapnavigator. Cs) is checknext action (): checking the next action; getmatrix target (): finding the next moving target point in the current map type; the current angle is explored outwards (while not explored); a larger angle; move back (dead end); if no larger angle destination is found, returning to the point 0; the explored targets are not explored outwards; generateExplorationPoint (int diameter): generating all movable target points in the current map. Therefore, the special environment scanning and mapping module realizes a special scanning flow aiming at the field environment, and the accuracy of subsequent point-by-point measurement is realized when the laser acquires data, and unique algorithms such as coordinate comparison, new and old map fusion and the like are additionally added, so that the method has higher accuracy.

The functions of the special plot measurer (mapmeasure.cs) include: according to the user setting, a measuring point set is generated in the whole map range, the point position of a first measuring point is judged according to the current position of the carrier 1, the first measuring point is moved to, the current coordinate is determined after the carrier 1 reaches the target position, the heavy hammer 52 is put down, the screen reading of the ground resistance meter 5 is shot by a camera, the reading value is identified, and if the reading value is correct, the carrier 1 is moved to the next measuring point. The distribution of the measurement points is shown in fig. 14, the distance between the measurement points can be set according to the requirement of the user, preferably 1-5 meters, and the periphery of the map needs to surround the wall, so that the measurement points can be automatically generated in the map. The API of this algorithm is GetPoints (MapManager map): according to the user setting, generating a measuring point in the whole map range; getNextTarget (Target target): searching the next nearest measuring point according to the vehicle position; generateExplorePoint (int diameter, int size): a mobile point within the detection area is generated. In addition, because the application scene of the invention is not a process of sequentially mapping a pair of maps at a time, the cloud point mapping action is performed by logic triggering after other actions (such as heavy hammer measurement and screen identification) are matched in different scenes. Of course, except for the need of calling the drawing process in the whole drawing process, the patrol process, the starting point process and the deviation correcting process all need to be called again at random. Therefore, the drawing point measuring module is specially independent into the high-priority calling module in the laser cloud point mapping technology so as to be called at any time.

The functions of the specialized map manager (mapmanager. Cs) include: the map data are comprehensively managed, the map scanned by the data acquisition module 2 (namely, the small radar map) is integrated into the main map, and the included angle between the current image acquired by the data acquisition module 2 (namely, the current radar image) and the main map is judged by comprehensive geomagnetic information so as to provide coordinate positioning of the carrier 1. The API of this algorithm is AddFrame (Frame frame): the current image is merged into a main map; determine (List < PointLidar > points,) is: judging coordinate points of the vehicle in the map according to the current radar points; determineAccuratePosition (Frame, frame New): a more accurate coordinate point within a certain range is determined. Therefore, the special map manager can be used for adding factory building standard size rules at the later stage, and the method can be applied to other factory buildings built in accordance with aerospace specifications, so that the reusability and the high-efficiency applicability can be greatly improved by importing factory building specifications. And before the measurement work is carried out, the existing factory drawings can be imported and compared with the layers of the laser measurement drawings and corrected (automatic/manual), so that the drawing accuracy is greatly improved. These features are also specifically configured for the present invention to be directed to a map manager.

In summary, the invention forms a set of special intelligent ground detection robots by using the carrier, the laser radar, the camera, the lifting mechanism and the control system, integrates PC software, the embedded control system and the high-strength movement structural member, forms the unification of cross-service, and coordinates the overall framework, the linkage scene analysis and the unification action to the final data and log output.

Referring to fig. 15, the robot of the present invention can automatically draw a map of a location using an automatically navigated lidar and a camera during initial training, automatically calculate the positions of the detection points by setting the number of the detection points and the distance between the two points, and mark the positions on the map, and simultaneously draw a travel route according to the detection points. The robot can automatically move according to the planned travel route, stop at each detection point, read the numbers on the instrument through image recognition and record after the detection is completed, and then continue to move until all the detection points are completed. After the robot finishes acquisition, the measurement data is automatically written into a terminal management system, and a measurement result is automatically calculated by a program. The robot of the invention bears 50kg and continues to travel for 5-10 km for 2-4 hours.

The above description is only one embodiment of the present invention and is not intended to limit the present invention, and various modifications and variations of the present invention will be apparent to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. The ground detection robot is characterized by comprising a carrier (1), a data acquisition module (2), a sensing module, a control system (4), a ground resistance meter (5), a lifting mechanism (6) and a camera, wherein the carrier (1) comprises a frame body (11) and a travelling mechanism (12), and the data acquisition module (2) and the sensing module are arranged on the carrier (1);

the travelling mechanism (12) comprises a chassis (121), a front wheel (122) and a rear wheel (123);

the ground resistance meter (5) comprises a control screen and a heavy hammer (52), and the control screen is arranged in the frame body (11);

the lifting mechanism (6) comprises a lifting motor (61), a cantilever (62) and a pulley assembly (63);

the lifting motor (61) is arranged on the chassis (121), one end of the cantilever (62) is hinged on the chassis (121), and the other end of the cantilever is connected with the frame body (11) through the pulley assembly (63);

one end of the cantilever (62) far away from the chassis (121) is provided with a connecting ring (621), and the heavy hammer (52) is arranged in the connecting ring (621);

the limit position of the heavy hammer (52) is limited by a travel switch;

the camera is arranged on the frame body (11) and used for collecting screen data of the ground resistance meter (5);

When the carrier (1) runs to a test point, the lifting motor (61) drives the cantilever (62) to turn over to put down the heavy hammer (52) to be in contact with the ground, the camera reads the screen reading of the ground resistance meter (5), and the screen reading is stored in the control system (4) after being identified by the text identification software;

the control system (4) is arranged in the frame body (11) and comprises an upper computer and a lower computer;

the upper computer is used for generating and displaying a map, generating an automatic patrol route according to the map and identifying the display content of the ground resistance instrument (5);

the lower computer comprises an STM32 embedded system and a sensor main control board and is used for controlling actions of the carrier (1) and the sensing module, collecting laser mapping data and preprocessing initial data;

The construction principle of the data structure of the lower computer is that plane mapping data is stored in a polar coordinate form; storing side length data with the total side length exceeding 5000 meters by using flash, and automatically pushing the side length data to a memory of the upper computer for storage if the side length data exceeds the storage;

the lower computer is communicated with the data acquisition module (2) by adopting a serial port, a UART or a TTL interface, the lower computer is communicated with the upper computer by adopting a network port, and the minimum transmission rate of control commands and acquired data of the camera is 0.8Mb/s-1.2Mb/s;

the upper computer establishes an environment map of the track scanned and retrieved by the data acquisition module (2), optimizes the image acquired by the data acquisition module (2) by using a position map, and sets a stable posture correction, a full-right angle movement algorithm and an independent map establishment module;

the specialized environment scanning mapping comprises:

the data acquisition module (2) completes initial scanning and converts an image into a matrix image, a straight line of a wall in the matrix image is judged, the image angle is adjusted according to the wall angle, the carrier (1) walks along an indoor wall for one circle in a clockwise direction, the carrier (1) further stops and generates a current image acquired by the data acquisition module (2) before each time, the current image is compared with a main map, the current coordinate is judged, and the current image is fused into the main map according to the current coordinate; wherein, a special right-angle action and a straight line deviation correction algorithm are adopted for drawing;

the specialized plot measurement includes:

generating a measuring point set in a full map range, judging and going to a first measuring point according to the current position of the carrier (1), determining a current coordinate after the carrier (1) arrives, setting down the heavy hammer (52), shooting a screen of the ground resistance meter (5) by using the camera, identifying a reading value, and if the reading value is correct, going to a next measuring point by the carrier (1); the distance between the measuring points is 1-5 meters, and the periphery of the map surrounds the wall;

The specialized map management includes:

and comprehensively managing map data, integrating the map scanned by the data acquisition module (2) into a main map, and judging the included angle between the current image acquired by the data acquisition module (2) and the main map by comprehensive geomagnetic information to obtain the coordinate positioning of the carrier (1).

2. The floor inspection robot of claim 1, wherein,

the front wheel (122) and the rear wheel (123) are arranged on the chassis (121) and are Mecanum wheels;

the chassis (121) is provided with wheel motors (124) connected with the front wheels (122) and the rear wheels (123), and the wheel motors (124) are independently controlled by corresponding motion control cards;

the front wheel (122) is flexibly connected with the chassis (121) through an elastic connecting rod;

the frame body (11) is made of aluminum alloy.

3. The ground detection robot according to claim 2, characterized in that the camera is connected with the control system (4) through a USB interface to receive controlled instructions and transfer snapshot frame data;

the snapshot frame data of the camera is 3 times of the real record data;

The camera determines the shape of a screen display image of the ground resistance instrument (5) by detecting dark and bright modes, the shape is translated into computer characters by utilizing a character recognition method, the characters are converted into a text format by a character recognition software, and the text format is edited and processed by a character processing software;

defining a number and letter template by adopting a template matching method, and sliding and matching characters on the ground resistance instrument (5) by utilizing the number and letter template;

4. The ground detection robot according to claim 2, characterized in that the data acquisition module (2) comprises a height-adjustable support platform, a laser ranging module (22) and a rotation mechanism;

the laser ranging module (22) comprises a laser ranging probe (221), a modulator (222), a distance display (223), a phase discriminator (224) and a reflecting mirror (225);

Two ends of the connector are respectively connected with the output end of the stepping motor and one end of the collecting ring;

the other end of the collecting ring is connected with a bracket of the laser ranging probe (221).

5. The ground detection robot of claim 4, wherein the laser ranging probes (221) are arranged in two and are arranged at an included angle of 90 degrees in the horizontal direction;

the laser ranging probe (221) adopts a TOF ranging mode to carry out multipoint sampling measurement, and the maximum detection distance is within 30 meters;

the measuring precision of the data acquisition module (2) is straight line precision of five ten thousandths to one thousandth, the angle precision is 3-5mrad, the radio frequency is used for carrying out amplitude modulation on laser, the phase delay generated by the modulated light back and forth measuring line once is measured, and then the distance represented by the phase delay is converted according to the wavelength of the modulated light, so that the time required for the light to pass through the back and forth measuring line is measured;

D＝1/2ct＝1/2c·φ/ω＝c/(4πf)(Nπ+Δφ)＝c/4f(N+ΔN)＝U(N+)；

the step angle is controlled by setting the subdivision number of the step motor, each step of the step motor is suspended for a period of time, the laser ranging probe (221) finishes one ranging during the period, and the carrier (1) automatically operates to the next sampling point;

during ranging, the laser ranging probe (221) scans for obstacles on the travel path in real time.

6. The ground detection robot of claim 4, wherein the data acquisition module (2) performs positioning navigation and mapping using SLAM, comprising: feature extraction, data association, state estimation, state update, and feature update;

the step of mapping the plant plan by SLAM includes:

optimizing the original data acquired by the data acquisition module (2) and eliminating problem data;

Searching the point cloud data of the current local environment for a corresponding position on an established map, and matching and splicing the point cloud currently acquired by the data acquisition module (2) into an original map;

and splicing the new data acquired by the data acquisition module (2) in the current round into the original map to finish updating the map.

7. The ground detection robot of claim 4, wherein the sensing module comprises a laser ranging sensor, a multi-axis acceleration sensor, and a geomagnetic sensor;

the geomagnetic sensor and the laser ranging probe (221) return data in real time, whether the running track of the carrier (1) is deviated or not is known, and when the track of the carrier (1) is deviated, the wheel motor (124) corrects the movement of the carrier (1);

the geomagnetic sensor has a resolution of 13nt@200Cycle Count.