A robot navigation and positioning method, system, device and storage medium

Through the combination of RFID and visual positioning technology, absolute and relative position information are integrated, and terrain mapping is established using stereo cameras and radars, which solves the problem of robot navigation and positioning in complex environments, achieving high-precision and robust navigation effects .

CN112629520BActive Publication Date: 2024-12-03BEIJING JIGUANGTONGDA TECH CO LTD
View PDF 9 Cites 0 Cited by

Patent Information

Application Number
CN202011344552.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-11-25
Publication Date
2024-12-03
Estimated Expiration
2040-11-25

AI Technical Summary

Technical Problem

There are difficulties in robots achieving accurate navigation and positioning in complex and changing environments, especially when there is a lot of external interference, the existing technology is difficult to effectively solve the positioning problem under high-precision and sparse label distribution.

Method used

RFID technology is used to combine particle filtering algorithms and visual positioning, and fuse data to achieve high-precision positioning by obtaining the absolute and relative position information of the robot, and establish topographic mapping through wide baseline stereo cameras, narrow baseline stereo cameras and panoramic radars to combine obstacles object height information for navigation and obstacle avoidance.

Benefits of technology

It realizes high-precision positioning and precise navigation of the robot, improves navigation accuracy and robustness in complex environments, and can effectively avoid obstacles.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN112629520B_ABST
    Figure CN112629520B_ABST
Patent Text Reader

Abstract

The present application relates to a robot navigation and positioning method, system, device and storage medium, and to the field of robot navigation technology. The method adopts RFID positioning technology to obtain the absolute position information of the robot, obtains the relative position information through visual positioning, and fuses the data of two different information sources, absolute positioning information and relative positioning information, to achieve the effect of complementary advantages and realize the high-precision positioning of the robot. The environmental image around the robot is collected, and the navigation map and obstacle map are obtained in combination with the precise position information of the robot; the navigation map and the obstacle map are fused to obtain a topographic map marked with obstacle height information. The obstacle height information in the topographic map is compared with the preset obstacle threshold, so as to output the control command. The robot realizes precise navigation and effective obstacle avoidance through the topographic map, with high accuracy, good robustness, and easy implementation.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of robot navigation technology, and in particular to a robot navigation and positioning method, system, device and storage medium. Background Art

[0002] With the rapid development of robotics technology, more and more intelligent robots are appearing in our daily lives. In recent years, intelligent robots have become a representative strategic goal in the field of high technology. The emergence and development of robotics technology has not only fundamentally changed the traditional industrial production landscape, but also has a profound impact on human social life.

[0003] Autonomous positioning and navigation is one of the prerequisites for robots to achieve intelligence, and is a key factor in giving robots the ability to perceive and act. When a robot is performing a task, when the working environment changes, the robot needs to know its position in the changing environment so that it can accurately perform the task. However, due to the complexity of the living environment, the environment changes at any time, and is subject to external influences and interference. People have an urgent need for location services, so the precise navigation and positioning of robots has become a major difficulty. Summary of the invention

[0004] In order to facilitate the accurate navigation and positioning of a mobile robot, the present application provides a robot navigation and positioning method, system, device and storage medium.

[0005] In a first aspect, the present application provides a robot navigation and positioning method, which adopts the following technical solution.

[0006] A robot navigation and positioning method, comprising:

[0007] Obtain the absolute position information and relative position information of the robot;

[0008] The absolute position information and relative position information of the robot are integrated to obtain accurate position information of the robot;

[0009] Collecting images of the robot's working environment, and combining the precise location information of the robot to obtain a navigation map and an obstacle map;

[0010] The navigation map and the obstacle map are integrated to obtain a topographic map with obstacle height information marked;

[0011] The robot is judged whether to climb over or turn to avoid the obstacle according to the obstacle height information in the terrain survey map.

[0012] By adopting the above technical solution, the absolute position information and relative position information of the robot are obtained, and the data of the two different information sources, absolute positioning information and relative positioning information, are integrated together to achieve the effect of complementary advantages and realize the high-precision positioning of the robot. A topographic map is obtained based on the navigation map and the obstacle map. The topographic map is marked with the height information of all obstacles. During the navigation process, the obstacle height information in the topographic map is compared with the preset obstacle threshold to determine whether the robot should flip or avoid obstacles, thereby outputting control instructions to realize the precise navigation of the robot.

[0013] Optionally, the process of acquiring the absolute position information of the robot is:

[0014] Establish an initial particle set and set the initial position and posture of each particle, where one particle posture corresponds to one robot posture;

[0015] For each particle’s position at time t-1, predict the position of each particle at time t;

[0016] Construct the relationship between the total weight of particles at time t and the readability weight and phase difference weight to obtain the total weight of each particle at time t;

[0017] The total weight of each particle at time t and the particle posture are used to calculate the posture of the robot at time t, and the absolute position information of the robot is obtained.

[0018] By adopting the above-mentioned technical solution, the RFID technology used in this application combines highly sensitive phase difference information and readability, and combines the particle filter positioning algorithm and the phase difference-based observation model to achieve the absolute positioning of the robot, thereby obtaining the robot's cluster position information, thereby solving the technical problem of robot positioning under high-precision and sparse tag distribution conditions.

[0019] Optionally, the process of acquiring the relative position information of the robot is:

[0020] Acquire a target object image and an expected target object image, and obtain a real-time acquired target object edge contour and an expected target object edge contour through an active contour;

[0021] Controlling the movement of the robot in real time according to the difference between the expected edge profile of the target object and the edge profile of the target object acquired in real time until the difference is minimized;

[0022] The relationship between the edge motion of the target object image and the edge motion of the image acquisition device is calculated according to the movement speed of the robot, so as to obtain the relative position information of the robot.

[0023] By adopting the above technical solution, the robot's movement is controlled based on active contour analysis to make it reach the correct position, and the relative position information of the robot relative to the target object is obtained according to the robot's movement, which facilitates the robot's accurate position information to be obtained in combination with the robot's absolute position information.

[0024] Optionally, the collecting of the environment image around the robot and obtaining the navigation map and obstacle map in combination with the precise location information of the robot specifically includes:

[0025] Collecting an environment image of the robot's feet to obtain a narrow baseline image, collecting an environment image other than the robot's feet to obtain a wide baseline image, matching the wide baseline image with the narrow baseline image, and obtaining a navigation map with the current precise location information of the robot as a navigation starting point based on the matching result and in combination with the precise location information of the robot;

[0026] Obtain a panoramic image, and scan obstacles in the surrounding environment through a laser radar to obtain a scan map containing several obstacles; measure obstacles in the surrounding environment to obtain obstacle height information, and mark the obstacle height information on the scan map to obtain an obstacle map.

[0027] By adopting the above technical solution, the environment image around the robot's feet and the environment image outside the feet are obtained respectively, and then the two images are matched to obtain a navigation map with the navigation starting point. The navigation map can clearly show the robot's overall working environment. By scanning and measuring the panorama and obstacles, an obstacle map marked with obstacle height information is obtained. The obstacle map can display all obstacles in the robot's overall working environment and their obstacle height information.

[0028] Optionally, the navigation map and the obstacle map are merged to obtain a topographic map marked with obstacle height information, and the specific process is as follows:

[0029] Extracting feature points from the navigation map and the obstacle map, and then matching the extracted feature points to obtain matching point pairs;

[0030] Using an affine transformation method, a rotation matrix, a scale variable and a translation vector between the navigation map and the obstacle map are obtained according to the matching point pairs, and a corresponding affine transformation matrix is ​​calculated;

[0031] The rotation matrix, scale variable, translation vector and affine transformation matrix are used to fuse the navigation map and the obstacle map, thereby obtaining a terrain survey map with obstacle height information marked.

[0032] By adopting the above technical solution, by fusing the navigation map and the obstacle map, the obtained terrain map can display the overall working environment of the robot and the height information of all obstacles in the environment, so as to facilitate the subsequent comparison of the obstacle height information in the terrain map with the pre-set obstacle threshold, so as to output control instructions to control the movement of the robot.

[0033] In a second aspect, the present application provides a robot navigation and positioning system, which adopts the following technical solution.

[0034] A robot navigation and positioning system, comprising:

[0035] Positioning module, used to obtain the absolute position information and relative position information of the robot;

[0036] A data fusion module, used to fuse the absolute position information and relative position information of the robot to obtain the precise position information of the robot;

[0037] An environmental perception module is used to collect images of the robot's working environment and obtain a navigation map and an obstacle map in combination with the robot's precise location information;

[0038] A map fusion module, used to fuse the navigation map and the obstacle map to obtain a topographic map marked with obstacle height information;

[0039] An obstacle avoidance navigation module, used to determine whether the robot should climb over or turn to avoid obstacles based on obstacle height information in the terrain survey map;

[0040] a controller, configured to output a control instruction according to a determination result of the obstacle avoidance navigation module; and

[0041] A driver is used to drive the robot to move according to the control instructions output by the controller.

[0042] By adopting the above technical solution, by combining the absolute position information and relative position information of the robot obtained by the positioning module, the effect of complementary advantages is achieved, and high-precision positioning is realized. The navigation map and obstacle map are obtained through the environmental perception module, and the navigation map and obstacle map are merged to obtain a topographic map with obstacle height information marked. The obstacle height information in the topographic map is compared with the preset obstacle threshold to determine whether the robot should climb over or turn to avoid, thereby controlling the movement of the robot.

[0043] Optionally, the positioning module includes:

[0044] RFID submodule, used to obtain the absolute position information of the robot by reading the RFID tag;

[0045] The visual positioning submodule is used to obtain the relative position information of the robot.

[0046] By adopting the above technical solution, RFID positioning technology is an absolute positioning method that can obtain accurate three-dimensional space coordinate information, and visual positioning is a relative positioning method that can obtain accurate relative position information. Therefore, fusing the data of two different information sources together can achieve complementary advantages.

[0047] Optionally, the environment perception module includes:

[0048] A narrow baseline image acquisition submodule, used for acquiring a narrow baseline image of the environment of the robot's feet in real time;

[0049] A wide baseline image acquisition submodule, used to acquire wide baseline images acquired by collecting the environment other than the robot's feet in real time;

[0050] An image fusion submodule is used to match the wide baseline image with the narrow baseline image, and obtain a navigation map with the precise location information as the navigation starting point based on the matching result and the precise location information of the robot;

[0051] A panoramic scanning submodule, used to obtain a panoramic image;

[0052] The obstacle scanning submodule is used to scan obstacles in the robot's working environment and obstacle height information; and

[0053] The obstacle marking submodule is used to mark the obstacles and their height information in the panoramic image to obtain an obstacle map.

[0054] By adopting the above technical solution, the narrow baseline image acquisition submodule and the wide baseline image acquisition submodule are used to obtain the robot's foot environment and the environment outside the foot respectively, and the navigation map is obtained by the image fusion submodule. The panoramic image and the height information of all obstacles are obtained by the panoramic scanning submodule and the obstacle scanning submodule, so as to obtain the information that can display the overall working environment of the robot and the height information of all obstacles in the environment.

[0055] On the third aspect, the device for robot navigation and positioning provided in this application adopts the following technical solution.

[0056] A computer device comprises a memory and a processor, wherein the memory stores a computer program that can be loaded and executed by the processor, and when the processor executes the computer program, the following is achieved:

[0057] Obtain the absolute position information and relative position information of the robot;

[0058] The absolute position information and relative position information of the robot are integrated to obtain accurate position information of the robot;

[0059] Collecting images of the robot's working environment, and combining the precise location information of the robot to obtain a navigation map and an obstacle map;

[0060] The navigation map and the obstacle map are integrated to obtain a topographic map with obstacle height information marked;

[0061] The step of determining whether the robot should climb over or turn to avoid the obstacle based on the obstacle height information in the terrain survey map.

[0062] By adopting the above technical solution, the processor in the computer device can implement the robot navigation and positioning method according to the relevant computer program stored in the memory, thereby achieving accurate navigation and positioning of the robot.

[0063] In a fourth aspect, the storage medium for robot navigation and positioning provided in the present application adopts the following technical solution.

[0064] A computer-readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor when it is run:

[0065] Obtain the absolute position information and relative position information of the robot;

[0066] The absolute position information and relative position information of the robot are integrated to obtain accurate position information of the robot;

[0067] Collecting images of the robot's working environment, and combining the precise location information of the robot to obtain a navigation map and an obstacle map;

[0068] The navigation map and the obstacle map are integrated to obtain a topographic map with obstacle height information marked;

[0069] The step of determining whether the robot should climb over or turn to avoid the obstacle based on the obstacle height information in the terrain survey map.

[0070] By adopting the above technical solution, the robot navigation and positioning method can be stored in a computer-readable storage medium, so that the computer program of the robot navigation and positioning method stored in the computer-readable storage medium can be executed by the processor, thereby realizing accurate navigation and positioning of the robot.

[0071] In summary, this application includes the following beneficial technical effects:

[0072] 1. RFID positioning technology is an absolute positioning method that can obtain coordinate information in three-dimensional space, and visual positioning is a relative positioning method. This application uses RFID positioning technology and visual fusion to fuse the data of two different information sources, absolute positioning information and relative positioning information, to achieve the effect of complementary advantages and realize high-precision positioning of the robot.

[0073] 2. A terrain map is created by processing with a wide-baseline stereo camera, a narrow-baseline stereo camera, and a panoramic radar. During navigation, the obstacle height information in the terrain map is compared with the preset obstacle threshold to output control instructions. The robot uses the terrain map to achieve precise navigation and effective obstacle avoidance with high accuracy, good robustness, and easy implementation. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 This is a flow chart of a robot navigation and positioning method according to one embodiment of the present application.

[0075] Figure 2 This is a schematic diagram of robot navigation and positioning logic according to one embodiment of the present application.

[0076] Figure 3 This is a flow chart of obtaining the absolute position information of a robot according to one embodiment of the present application.

[0077] Figure 4 It is a structural block diagram of a robot navigation and positioning system according to one embodiment of the present application. DETAILED DESCRIPTION

[0078] In order to make the objectives, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments.

[0079] This specific embodiment is merely an explanation of the present application and is not a limitation of the present application. After reading this specification, those skilled in the art may make modifications to the present embodiment without any creative contribution as needed, but such modifications are protected by the patent law as long as they are within the scope of the claims of the present application.

[0080] The present application embodiment discloses a robot navigation and positioning method, such as Figure 1 and Figure 2 As shown, the method includes the following steps.

[0081] S100, obtaining the absolute position information and relative position information of the robot.

[0082] (1) Obtain the robot’s absolute position information through RFID positioning technology.

[0083] The robot is equipped with a reader and a reader antenna, and a passive UHF RFID tag is installed on the ground as a reference tag. The EPC of all reference tags and their corresponding positions in the world coordinate system are stored in the robot as known information. During the continuous movement of the robot, the robot controls the RFID system to continuously read the reference tags, and locates the robot through the readability and phase-related information of the reference tags extracted from the RFID system.

[0084] like Figure 3 As shown, the specific steps of the method include:

[0085] S101, establishing an initial particle set, wherein a particle corresponds to the robot, and a particle posture corresponds to a robot posture.

[0086] The initial position of the particle is set as shown in formula (1):

[0087] (1)

[0088] in, Xi 0 is the first particle in the initial moment. i The initial position of each particle; R 1, R 2, R 3 is a random number between 0 and 1; T j = T 1, T 2, ..., T m , T j is the reference label read at the initial moment, x Tj is the horizontal coordinate of the reference label read at the initial time, y Tj is the ordinate of the reference label read at the initial moment, r is the maximum read range of the reference tag.

[0089] S102, for each particle t -1 The position of each particle is predicted using the dead reckoning algorithm. t The position of the moment.

[0090] The key to the dead reckoning (DR) positioning technology is to measure the distance traveled by the robot per unit time interval and the change in the robot's heading during this time interval. The rotation rate and acceleration rate are measured using a tug and an accelerometer, respectively, and the measurement results are integrated to solve the distance the robot has moved and the change in heading, and then the robot's position and posture are obtained based on the dead reckoning algorithm.

[0091] S103, build t The relationship between the total particle weight at a given moment and the readability weight and phase difference weight is obtained t The total weight of each particle at that moment.

[0092] First, the particle readability weight is evaluated: the particle posture judgment condition is set, and the rationality of the robot posture represented by each particle in the particle set is determined by the particle posture judgment condition. The readability weight of the particle that passes the rationality test is set as ω 1,i,t =1, the readability weight of particles that fail the rationality test is set to ω 1,i,t =0.

[0093] Calculate the phase difference weight of the particle. The specific formula is as follows:

[0094] (2)

[0095] (3)

[0096] (4)

[0097] in, ω 2,i,t No. i Particles in t The phase difference weight at the moment, △ θ t The antennas are respectively t -1 moment and t The difference in observed phase of the same RFID reference tag measured at each moment. θ i,t It is i The predicted phase difference of each particle is σ is the standard deviation of the phase measurement noise, △ d t yes t The distance difference in time, d 1 is in t -1 is the distance from the antenna to the RFID reference tag, d 2 is in t The distance from the antenna to the RFID reference tag at the moment, λ Represents the wavelength of a radio signal.

[0098] The total weight of the particle ω i,t It is expressed as:

[0099] (5)

[0100] in, ω i,t-1 It is i Particles in t The total weight at the moment -1, ω 1,i,t It is i Particles in t The readability weight of the moment, ω 2,i,t It is i Particles in t The phase difference weight at the moment.

[0101] S104, calculating the posture of the robot at time t using the total weight of each particle and the particle posture, so as to achieve absolute positioning of the robot.

[0102] By estimating the weighted geometric center of the particle set, the robot's position and posture are calculated to achieve robot positioning:

[0103] (6)

[0104] in, x j is the horizontal coordinate of the predicted pose, y j is the ordinate of the predicted pose, th j is the attitude angle of the predicted pose, x i,t yes t Moment i The horizontal coordinate of a particle in the world coordinate system, y i,t yes t Moment i The ordinate of a particle in the world coordinate system, th i,t yes t Moment i The attitude angle of a particle in the world coordinate system, N is the number of particles in the particle set.

[0105] (2) Obtain the relative position information of the robot through the camera.

[0106] In this embodiment, the camera will be installed on the robot's end effector. At this time, the target object image obtained by the RGB camera directly reflects the relative position and posture relationship between the target object and the robot's end effector. Before tracking the target object, offline learning is performed to control the robot to the correct position to obtain the desired target object image and use the active contour to obtain the desired target object edge contour. v * :

[0107] (7)

[0108] in, x*j , y*j Respectively represent the desired target object edge contour j The horizontal and vertical coordinates of the point, n is the number of points on the edge contour of the desired target object.

[0109] When working online, the RGB camera collects the target object image in real time, and obtains the edge contour of the target object in real time through the active contour. v :

[0110] (8)

[0111] in, x*j , y*j Respectively represent the edge contours of the target object acquired in real time j The horizontal and vertical coordinates of the point.

[0112] The difference between the expected target object edge contour and the target object edge contour obtained in real time e = v - v * The robot's movement is controlled in real time until the difference is eliminated or minimized, at which point the robot has reached the correct position.

[0113] The relationship between image edge motion and RGB camera edge motion:

[0114] (9)

[0115] in, τ c is the motion speed of the RGB camera: τ c =( T c , W c ), T c =( T cx , T cy , T cz ) is the translational velocity, W c =( W cx , W cy , W cz ) is the rotation speed, LIt reflects the three-dimensional coordinates of the target object ( X j , Y j , Z j ) and the coordinates of the corresponding point in the image coordinates ( x j , y j ) relationship, considering perspective projection, x j = X j / Z j , y j = Y j / Z j ,but L Determined by the following formula:

[0116] (10)

[0117] In order to make calculation more convenient, in actual work Z j Desirable Z*j (i.e. expected value).

[0118] Through the relationship matrix L Get the three-dimensional coordinates of the target object ( X j , Y j , Z j ), thereby obtaining the relative position information of the robot.

[0119] S200, fusing the absolute position information and relative position information of the robot to obtain accurate position information of the robot.

[0120] This embodiment adopts a method of fusing two positioning technologies to perform precise positioning, and uses visual relative position information for auxiliary positioning.

[0121] The specific fusion methods are:

[0122] The world coordinate system used in the RFID positioning process is used as the global coordinate system, and the relative position information calculated by the vision is converted to the world coordinate system through spatial transformation. A hierarchical neural network is constructed, and the two position coordinates are input into the neural network for data fusion.

[0123] In this embodiment, a three-layer feedforward network model based on the BP algorithm is adopted, and the established three-layer feedforward network model is first trained offline. The training algorithm uses the perception information represented by the binary vector as the input vector, and the corresponding decision command binary vector as the target vector, where the sample data is the perception information category within the scope of prior knowledge. After the model training is completed, the absolute position information and the relative position information after the spatial transformation are used as the input of the three-layer feedforward network model, and a fusion strategy is generated according to the pre-set fusion parameters, thereby completing the fusion of the absolute position information and the relative position information of the robot, and obtaining the precise position information of the robot.

[0124] The neural network used in this embodiment has strong fault tolerance and self-organization, self-learning and self-adaptation capabilities, and can realize complex mapping; the neural network also has strong nonlinear processing capabilities, and can well meet the requirements of data fusion technology.

[0125] S300, collecting the working environment image of the robot, and obtaining a navigation map and an obstacle map in combination with the precise position information of the robot.

[0126] (1) The current precise position information of the robot is used as the starting position of navigation. A narrow baseline stereo camera is used to take real-time photos of the environment near the robot's feet to obtain a narrow baseline image. A wide baseline stereo camera is used to take real-time photos of the distant environment outside the robot's feet to obtain a wide baseline image.

[0127] According to the optical characteristic information (grayscale value, color value, etc.), geometric characteristic information (shape, size, etc.), spatial position information (position, direction, etc. in the image), and feature information (position information of edges and corner points) represented by the wide baseline image and the narrow baseline image, the structural features in these information are combined to form a stable feature vector. The feature vector that maintains a certain stability for factors such as geometric deformation and illumination change of the image is used as an invariant. The wide baseline image data and the narrow baseline image data are matched. According to the nearest neighbor Euclidean distance ratio criterion, the most stable matching extreme feature region pairs in the two images are extracted. The sequential sampling consistency algorithm is used to eliminate the mismatched feature region pairs, and the outer pole geometric relationship of the two image data is estimated to obtain the matching result. According to the matching result and combined with the current precise position information of the robot, a navigation map with the current precise position information as the navigation starting point is obtained.

[0128] (2) A panoramic camera is used to capture the surrounding environment to obtain a panoramic image, and obstacles in the surrounding environment are scanned by several laser radars to obtain a scan map containing several obstacles. Then, obstacle height information is obtained by measuring obstacles in the surrounding environment, and the obstacle height information is annotated on the scan map to obtain an obstacle map.

[0129] The process of measuring obstacle height information by laser radar is as follows: assuming the ground is a plane and the height angle of the obstacle is α The slant distance between the radar and the top of the obstacle can be calculated based on the propagation speed of the radio wave and the time it takes to travel back and forth. d , then the height of the obstacle h for:

[0130] (11)

[0131] The calculated obstacle height information is marked on the panoramic scanning map to obtain an obstacle map.

[0132] S400: Fusing the navigation map and the obstacle map to obtain a topographic map with obstacle height information marked.

[0133] Step 401 , extracting feature points from the navigation map and the obstacle map, and then matching the extracted feature points to obtain matching point pairs.

[0134] Step 402: Using an affine transformation method, the rotation matrix, scale variable and translation vector between the navigation map and the obstacle map are obtained according to the matching point pairs, and the corresponding affine transformation matrix T is calculated.

[0135] (12)

[0136] Among them, Φ represents the rotation angle of the obstacle map compared to the navigation map, s represents the fusion ratio, δ x , δ y They represent the difference between the obstacle map and the navigation map. x Direction and y The offset in direction.

[0137] Step 403, using the rotation matrix, scale variable, translation vector and affine transformation matrix, the navigation map and the obstacle map are fused to obtain a topographic map with obstacle height information marked.

[0138] S500: Determine whether the robot should climb over or turn to avoid the obstacle based on the obstacle height information in the terrain survey map.

[0139] The obstacle height information in the topographic map is compared with the preset obstacle threshold so that the robot navigation can determine whether to climb over or turn to avoid. When the obstacle height information is less than the preset obstacle threshold, it means that the robot can climb over the obstacle, and the robot will climb over the obstacle. When the obstacle height information is not less than the preset obstacle threshold, it means that the robot cannot climb over the obstacle, and the robot will turn to avoid the obstacle.

[0140] The present application also discloses a robot navigation and positioning system. Figure 2 and Figure 4 As shown, the system comprises:

[0141] (1) Positioning module, used to obtain the absolute and relative position information of the robot.

[0142] In this embodiment, the positioning module includes:

[0143] The RFID submodule is used to obtain the absolute position information of the robot by reading the RFID tag. The RFID submodule includes: an RFID tag, a reader and a reader antenna. The reader and the reader antenna are installed on the robot, and a passive UHF RFID tag is installed on the ground as a reference tag. The reader reads / writes the information in the RFID tag. The reader antenna transmits the radio frequency signal between the RFID tag and the reader through the communication base station.

[0144] The visual positioning submodule is used to obtain the relative position information of the robot. The visual positioning submodule in this embodiment uses an RGB camera, which is installed on the end joystick of the robot. The object image obtained in this way directly reflects the relative position and posture relationship between the target object and the robot end manipulator, thereby obtaining the relative position information of the robot.

[0145] (2) A data fusion module, used to fuse the absolute position information and relative position information of the robot to obtain the precise position information of the robot.

[0146] (3) An environmental perception module, which is used to collect images of the robot's working environment and obtain a navigation map and an obstacle map based on the robot's precise location information.

[0147] In this embodiment, the environment perception module specifically includes:

[0148] The narrow baseline stereo camera is used to obtain a narrow baseline image of the environment of the robot's feet in real time.

[0149] The wide baseline stereo camera is used to obtain wide baseline images of the remote environment outside the robot's feet in real time.

[0150] The image fusion submodule is used to match the wide baseline image with the narrow baseline image, and obtain a navigation map with the precise position information as the navigation starting point based on the matching result and the precise position information of the robot.

[0151] The panoramic scanning submodule is used to obtain a panoramic image. The panoramic scanning submodule may be a panoramic camera.

[0152] The obstacle scanning submodule is used to scan obstacles in the robot's working environment and obtain obstacle height information. The obstacle scanning submodule may be a laser scanning radar.

[0153] The obstacle marking submodule is used to mark the measured obstacle height information into the panoramic scanning map to obtain an obstacle map.

[0154] (4) A map fusion module, used to fuse the navigation map and the obstacle map to obtain a topographic map with obstacle height information marked on it.

[0155] (5) An obstacle avoidance navigation module, which is used to determine whether the robot should climb over or turn to avoid the obstacle based on the obstacle height information in the terrain survey map.

[0156] (6) A controller, configured to output control instructions according to the determination result of the obstacle avoidance navigation module.

[0157] (7) A driver, used to drive the robot to move according to the control instructions output by the controller.

[0158] The present application also discloses a storage medium for robot navigation and positioning, which adopts the following technical solution:

[0159] A computer-readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor when it is run:

[0160] Obtain the absolute position information and relative position information of the robot;

[0161] The absolute position information and relative position information of the robot are integrated to obtain accurate position information of the robot;

[0162] Collecting images of the robot's working environment, and combining the precise location information of the robot to obtain a navigation map and an obstacle map;

[0163] The navigation map and the obstacle map are integrated to obtain a topographic map with obstacle height information marked;

[0164] The step of determining whether the robot should climb over or turn to avoid the obstacle based on the obstacle height information in the terrain survey map.

[0165] The present application also discloses a robot navigation and positioning device, which adopts the following technical solution:

[0166] A computer device comprises a memory and a processor, wherein the memory stores a computer program that can be loaded and executed by the processor, and when the processor executes the computer program, the following is achieved:

[0167] Obtain the absolute position information and relative position information of the robot;

[0168] The absolute position information and relative position information of the robot are integrated to obtain accurate position information of the robot;

[0169] Collecting images of the robot's working environment, and combining the precise location information of the robot to obtain a navigation map and an obstacle map;

[0170] The navigation map and the obstacle map are integrated to obtain a topographic map with obstacle height information marked;

[0171] The step of determining whether the robot should climb over or turn to avoid the obstacle based on the obstacle height information in the terrain survey map.

[0172] The above are all preferred embodiments of the present application, and the protection scope of the present application is not limited thereto. Therefore, any equivalent changes made according to the structure, shape, and principle of the present application should be included in the protection scope of the present application.

Claims

1. A robot navigation and positioning method, characterized in that: include: Obtain the absolute position information and relative position information of the robot; The absolute position information and relative position information of the robot are fused to obtain the precise position information of the robot, wherein the world coordinate system used by the absolute position information is used as the global coordinate system, and the relative position information is converted to the world coordinate system through spatial transformation, a hierarchical neural network is constructed, and the two position information are input into the hierarchical neural network for data fusion, wherein the training process of the hierarchical neural network includes: establishing a three-layer feedforward network model of the BP algorithm, and performing offline training on the three-layer feedforward network model, during the training, the perception information represented by the binary vector is used as the input vector, and the corresponding decision command binary vector is used as the target vector, wherein the sample data is the perception information category within the scope of prior knowledge; Collecting images of the robot's working environment and combining them with the robot's precise location information to obtain a navigation map and an obstacle map; fusing the navigation map and the obstacle map to obtain a topographic map with obstacle height information marked; Determining whether the robot should climb over or turn to avoid the obstacle based on the obstacle height information in the terrain survey map; The collecting of the working environment image of the robot and obtaining the navigation map and obstacle map in combination with the precise position information of the robot specifically include: Collecting an environment image of the robot's feet to obtain a narrow baseline image, collecting an environment image other than the robot's feet to obtain a wide baseline image, matching the wide baseline image with the narrow baseline image, and obtaining a navigation map with the current precise location information of the robot as a navigation starting point based on the matching result and in combination with the precise location information of the robot; Obtain a panoramic image and scan obstacles in the surrounding environment through a laser radar to obtain a scan map containing several obstacles; Obstacles in the surrounding environment are measured to obtain obstacle height information, and the obstacle height information is marked on the scan map to obtain an obstacle map; The step of matching the wide baseline image with the narrow baseline image includes: According to the optical characteristic information, geometric characteristic information, spatial position information and structural features in the feature information represented by the wide baseline image and the narrow baseline image, a stable feature vector is formed, and the wide baseline image data and the narrow baseline image data are matched, so as to extract the most stable matching extreme feature region pairs in the two images according to the nearest neighbor Euclidean distance ratio criterion, and the wrong matching feature region pairs are eliminated by using the sequential sampling consistency algorithm, and the geometric relationship between the two image data is estimated to obtain the matching result; Obstacles in the surrounding environment are measured to obtain obstacle height information, including: By using several laser radars to measure the height of obstacles in the surrounding environment, the obstacle height information corresponding to each of the several obstacles is obtained; the calculation formula of the obstacle height information is: h = dsinα, where h is the obstacle height information, d is the inclined distance between the radar and the obstacle vertex, and α is the height angle of the obstacle; The process of obtaining the relative position information of the robot is as follows: Before tracking the target object, offline learning is performed to control the robot to the correct position, obtain the target object image and the expected target object image, and obtain the expected target object edge contour through the active contour Wherein, x*j, y*j represent the horizontal coordinate and vertical coordinate of point j on the desired edge contour of the target object, respectively, and n is the number of points on the desired edge contour of the target object; When working online, the RGB camera is used to collect the target object image in real time, and the edge contour of the target object v = {(x j ,y j ), j = 1, 2, ..., n}, xj, yj represent the horizontal coordinate and vertical coordinate of point j on the edge contour of the target object acquired in real time, respectively; Controlling the movement of the robot in real time according to the difference between the expected edge profile of the target object and the edge profile of the target object acquired in real time, until the difference is minimized, at which time the robot reaches a correct position; The relationship between the edge motion of the target object image and the edge motion of the image acquisition device is calculated according to the movement speed of the robot, so as to obtain the relative position information of the robot, wherein the relationship between the edge motion of the target object image and the edge motion of the image acquisition device is u=Lτ c , τ c is the motion speed of the RGB camera: τ c =(T c ,W c ), T c =(T cx ,T cy ,T cz ) is the translational velocity, W c =(W cx ,W cy ,W cz ) is the rotation speed, L is the three-dimensional coordinate of the target object (X j ,Y j ,Z j ) and the coordinates of the corresponding point in the image coordinates (x j ,y j ) matrix, considering perspective projection, x j =X j / Z j ,y j =Y j / Z j , then L is determined by the following formula: The three-dimensional coordinates (X j ,Y j ,Z j ).

2. The robot navigation and positioning method according to claim 1, characterized in that: The process of obtaining the absolute position information of the robot is as follows: Establish an initial particle set and set the initial position and posture of each particle, where one particle posture corresponds to one robot posture; For each particle’s position at time t-1, predict the position of each particle at time t; Construct the relationship between the total weight of particles at time t and the readability weight and phase difference weight to obtain the total weight of each particle at time t; The total weight of each particle at time t and the particle posture are used to calculate the posture of the robot at time t, and the absolute position information of the robot is obtained.

3. The robot navigation and positioning method according to claim 1, characterized in that: The navigation map and the obstacle map are merged to obtain a terrain survey map with obstacle height information marked thereon. The specific process is as follows: Extracting feature points from the navigation map and the obstacle map, and then matching the extracted feature points to obtain matching point pairs; Using an affine transformation method, a rotation matrix, a scale variable and a translation vector between the navigation map and the obstacle map are obtained according to the matching point pairs, and a corresponding affine transformation matrix is ​​calculated; The rotation matrix, scale variable, translation vector and affine transformation matrix are used to fuse the navigation map and the obstacle map, thereby obtaining a terrain survey map with obstacle height information marked.

4. A robot navigation and positioning system, characterized in that: include: Positioning module, used to obtain the absolute position information and relative position information of the robot; A data fusion module is used to fuse the absolute position information and relative position information of the robot to obtain the precise position information of the robot, wherein the world coordinate system used by the absolute position information is used as the global coordinate system, the relative position information is converted to the world coordinate system through spatial transformation, a hierarchical neural network is constructed, and the two position information are input into the hierarchical neural network for data fusion, wherein the training process of the hierarchical neural network includes: establishing a three-layer feedforward network model of the BP algorithm, and performing offline training on the three-layer feedforward network model, during training, the perception information represented by the binary vector is used as the input vector, and the corresponding decision command binary vector is used as the target vector, wherein the sample data is the perception information category within the scope of prior knowledge; An environmental perception module is used to collect images of the robot's working environment and obtain a navigation map and an obstacle map in combination with the robot's precise location information; A map fusion module, used to fuse the navigation map and the obstacle map to obtain a topographic map marked with obstacle height information; An obstacle avoidance navigation module, used to determine whether the robot should climb over or turn to avoid obstacles based on obstacle height information in the terrain survey map; a controller, configured to output a control instruction according to a determination result of the obstacle avoidance navigation module; and A driver, used for driving the robot to move according to the control instructions output by the controller; The environment perception module comprises: A narrow baseline image acquisition submodule, used for acquiring a narrow baseline image of the environment of the robot's feet in real time; The wide baseline image acquisition submodule is used to acquire a wide baseline image obtained by collecting the environment other than the robot's feet in real time; the image fusion submodule is used to match the wide baseline image with the narrow baseline image, and obtain a navigation map with the precise location information as the navigation starting point according to the matching result and in combination with the precise location information of the robot; wherein matching the wide baseline image with the narrow baseline image includes: According to the optical characteristic information, geometric characteristic information, spatial position information and structural features in the feature information represented by the wide baseline image and the narrow baseline image, a stable feature vector is formed, and the wide baseline image data and the narrow baseline image data are matched, so as to extract the most stable matching extreme feature region pairs in the two images according to the nearest neighbor Euclidean distance ratio criterion, and the wrong matching feature region pairs are eliminated by using the sequential sampling consistency algorithm, and the geometric relationship between the two image data is estimated to obtain the matching result; A panoramic scanning submodule, used to obtain a panoramic image; The obstacle scanning submodule is used to scan obstacles and obstacle height information in the robot's working environment. The obstacle height information corresponding to each obstacle is obtained by measuring the height of the obstacles in the surrounding environment through a number of laser radars. The calculation formula of the obstacle height information is: h = dsinα, where h is the obstacle height information, d is the inclined distance between the radar and the obstacle vertex, and α is the height angle of the obstacle; and, An obstacle marking submodule, used for marking the obstacles and their height information into the panoramic image to obtain an obstacle map; The process of obtaining the relative position information of the robot is as follows: Before tracking the target object, offline learning is performed to control the robot to the correct position, obtain the target object image and the expected target object image, and obtain the expected target object edge contour through the active contour Where x*j, y*j represent the horizontal and vertical coordinates of point j on the desired edge contour of the target object, respectively, and n is the number of points on the desired edge contour of the target object; When working online, the RGB camera is used to collect the target object image in real time, and the edge contour of the target object v = {(x j ,y j ), j = 1, 2, ..., n}, xj, yj represent the horizontal coordinate and vertical coordinate of point j on the edge contour of the target object acquired in real time, respectively; Controlling the movement of the robot in real time according to the difference between the expected edge profile of the target object and the edge profile of the target object acquired in real time, until the difference is minimized, at which time the robot reaches a correct position; The relationship between the edge motion of the target object image and the edge motion of the image acquisition device is calculated according to the movement speed of the robot, so as to obtain the relative position information of the robot, wherein the relationship between the edge motion of the target object image and the edge motion of the image acquisition device is u=Lτ c , τ c is the motion speed of the RGB camera: τ c =(T c ,W c ), T c =(T cx ,T cy ,T cz ) is the translational velocity, W c =(W cx ,W cy ,W cz ) is the rotation speed, L is the three-dimensional coordinate of the target object (X j ,Y j ,Z j ) and the coordinates of the corresponding point in the image coordinates (x j ,y j ) relationship, considering perspective projection, x j =X j / Z j ,y j =Y j / Z j , then L is determined by the following formula: The three-dimensional coordinates (X j ,Y j ,Z j ).

5. The robot navigation and positioning system according to claim 4, characterized in that: The positioning module comprises: RFID submodule, used to obtain the absolute position information of the robot by reading the RFID tag; The visual positioning submodule is used to obtain the relative position information of the robot.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program that can be loaded and executed by the processor, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 3 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, wherein the computer program executes the steps of the method according to any one of claims 1 to 3 when executed by a processor.

Citation Information

Patent Citations

  • Single-line laser radar device

    CN105699985A

  • Obstacle detection method and device of mobile robot and mobile robot

    CN106680832A

  • Robot autonomous navigation system based on double single-line laser radars

    CN107167141A

  • Robot vision servo control method based on image mixing moment

    CN107901041A

  • Visual servo decoupling control method based on contour edge and internally- surrounded circle features

    CN108621167A