A rectangular recognition-based cluster robot position perception system and method

Abstract

The application relates to a rectangular recognition-based cluster robot position sensing system and method, and belongs to the field of robots. The system comprises a light sign module, a visual sensor module and a main control board, the main control board is connected with the light coding module and the visual sensor module. The method comprises the following steps: the main control board processes the image collected by the visual sensor, the processed image collected by the visual sensor is used to calculate the relative distance and angle of the neighbor robots and store the processed result into the storage unit of the main control board, the cluster algorithm program can be called, and the main control board controls power supply of the light sign module. The position information of the surrounding neighbors can be obtained by only loading a cheap camera module and a simple light sign module on the robot.

Description

Technical Field

[0001] This invention belongs to the field of robotics, specifically relating to a cluster robot position perception system and method based on rectangle recognition. Background Technology

[0002] Robotics, a crucial branch of artificial intelligence, has profoundly shaped all aspects of our lives. Among its applications, using multiple robots to collaborate on complex tasks is an effective way to improve efficiency. However, when the number of robots is large, traditional centralized control methods suffer from low fault tolerance and are difficult to maintain. The emergence of swarm robots offers a feasible solution to these problems.

[0003] Swarm robots aim to enable complex swarm behaviors in systems composed of a large number of simple robots through simple interaction rules among individuals. With in-depth research into self-organizing motion control algorithms for swarm robots, the verification of swarm robots has become a crucial step in engineering applications. To better align with engineering practice, robots should possess the ability to independently perceive their surroundings during swarm algorithm verification experiments. However, many robot systems employ a pseudo-distributed control approach, where global information is acquired first, and then processed as needed into local information required by the algorithm.

[0004] Swarm robot systems require robots with simple and low-cost construction. Therefore, these robots are typically not equipped with expensive and high-powered positioning devices. To better meet the practical needs of algorithms, it is necessary to build distributed swarm robot perception systems in a simple and low-cost manner. Summary of the Invention

[0005] The technical problem to be solved by this invention is: To address the bottleneck issue of desktop swarm robots struggling to acquire global perception information, this invention designs a swarm robot position perception system and method based on rectangle recognition. Relying on a vision sensor module and a light marker module, the robot quickly identifies rectangular bright markers and calculates the relative positions of its neighbors to itself using these rectangles for swarm algorithm computation.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A cluster robot position perception system based on rectangle recognition is characterized in that the system is installed on the top of the robot and includes a light marker module, a vision sensor module, a main control board, and a power supply module. The light marker module is used to generate a rectangular bright block as a marker for identifying neighboring individuals; The visual sensor module is used to collect image information of the individual's omnidirectional field of vision and transmit it to the main control board for calculation and analysis; The main control board connects the light sign module and the vision sensor module. It is used to process the images collected by the vision sensor, calculate the relative distance and angle of neighboring robots, and store the processed results in its storage unit for use by the cluster algorithm program. The power module is used to supply power to the main control board.

[0007] A further technical solution of the present invention: the light sign module comprises a cylindrical lampshade, three LED light strips and an equilateral triangular wooden strip; the three LED light strips are respectively fixed on the three sides of the equilateral triangular wooden strip; the cylindrical lampshade is made of frosted acrylic tube, and the upper and lower ends of the lampshade are provided with black opaque strip-shaped parts.

[0008] A further technical solution of the present invention: the visual sensor module consists of three cameras, which are oriented in multiple directions to obtain an omnidirectional field of view around the individual.

[0009] A further technical solution of the present invention: the main control board is a Raspberry Pi 4B, which can support its own control program calls or transmission with other terminals within a local area network.

[0010] A method for position perception of swarm robots based on rectangle recognition, characterized by comprising: S1: Convert the acquired multiple images to grayscale to generate grayscale images; perform Gaussian filtering on the grayscale images to form a grayscale histogram; set a threshold based on the grayscale histogram and perform binarization on the grayscale images. S2: Perform erosion and dilation processing on the binarized image; S3: Filter the bright spots in the image after erosion and dilation processing by size, and remove the bright spots that are too small; S4: Obtain the center coordinates and pixel height of the smallest inscribed rectangle of each bright patch in the image, and use the x-coordinate of the center point of the rectangular bright patch generated by the neighboring robot in the image as the position of the neighboring robot in the corresponding image; S5: Calculate the relative angle of neighboring robots using the horizontal axis; S6: Calculate the relative distance to neighboring robots using the pixel height of the rectangular bright blocks in the image.

[0011] A further technical solution of the present invention includes a robot for filtering out repeated calculations in multiple images.

[0012] A further technical solution of the present invention: the filtering specifically involves removing bright spots with fewer than 200 pixels.

[0013] A further technical solution of the present invention: Calculating the relative angle of neighboring robots using the horizontal axis specifically involves:

[0014] in, M * N For image resolution, Let x be the x-coordinate of the midpoint of the line connecting the centers of the rectangular bright blocks, and let y be the field of view angle of the camera. ~ ,if For negative values, the above formula needs to be adjusted. Take the remainder.

[0015] A further technical solution of the present invention: Calculating the relative distance to neighboring robots based on the height of rectangular bright blocks in an image specifically involves:

[0016] in, When calibrating the camera, the pixel height of the rectangular bright blocks of neighboring robots in the image captured by the camera is determined. To calibrate the distance between the robot and its neighboring robots for the camera, The pixel height of the rectangular bright block on the image captured by the neighboring robot.

[0017] A further technical solution of the present invention is to use OpenCV, a cross-platform computer vision and machine learning software library released under the Apache 2.0 license, for processing.

[0018] The beneficial effects of this invention are as follows: This invention proposes a position perception system and method for swarm robots based on rectangle recognition. The robot only needs an inexpensive camera module and a simple light marker module to acquire the position information of its neighbors. The image shape presented by the same light marker module is consistent from any direction, with no directional differences, reducing recognition interference and facilitating mutual recognition among distributed swarm robots. Calculating relative angles using the x-coordinate of the rectangle's center point is convenient, quick, and less susceptible to interference, and also facilitates filtering out duplicates from multiple visual sensors. Calculating relative distances using the rectangle's height is simple, reducing computation time on the main control board, improving robot response speed, and shortening the runtime of each step in the swarm algorithm, making the distributed swarm robot more closely aligned with the real-time requirements of swarm algorithms. Attached Figure Description

[0019] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts.

[0020] Figure 1 This is a system structure diagram of the present invention.

[0021] Figure 2 This is a schematic diagram of the structure of the present invention.

[0022] Figure 3 This is a schematic diagram of the camera numbering of the present invention.

[0023] Figure 4 This is a schematic diagram of the relative angle and relative distance calculation method of the present invention.

[0024] The components are: 1-base plate; 2-LED light strip; 3-equilateral triangular wooden strip; 4-cylindrical lampshade; 5-camera a; 6-camera b; 7-camera c; 8-pre-drilled connection holes on the base plate. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0026] This invention provides a swarm robot position perception system based on rectangle recognition. The system, mounted on top of each robot, includes a light marker module, a vision sensor module, a main control board, and a power supply module. Based on vision sensing technology, each robot has a light marker module on its top, which generates a rectangular bright area as a marker for identifying neighboring robots. Each robot perceives its surroundings through its vision sensor, identifies marker information in the images captured by the sensor, and calculates the position of its neighbors relative to itself. The main control board connects to the vision sensor module and the light marker module, processes the images captured by the vision sensor, calculates the relative distances and angles of neighboring robots, and stores the processed results in its storage unit for use by the swarm algorithm program. The main control board is connected to the light marker module to provide power. This invention requires only a simple structure and a common camera to identify the position of other swarm robots relative to itself, facilitating the transfer experiments of distributed swarm algorithms.

[0027] The lighting sign module consists of a connecting base plate 1, a cylindrical lampshade 4, three LED light strips 2, and an equilateral triangular wooden strip 3. The connecting base plate 1 is 3D printed to fix the various parts of the lighting sign module and has three evenly distributed slots for fixing three cameras. The base plate 1 connects to the robot below, and its size can be flexibly adjusted according to the type of robot to be adapted. The cylindrical lampshade 4 is a frosted acrylic tube with a length of 150mm and a diameter of 20mm. The bottom of the cylindrical lampshade 4 is wrapped with 30mm of black opaque tape, and the top is wrapped with 20mm of black opaque tape, making a lampshade with a light-transmitting part of 100mm. The LED light strips 2 are three 150mm long and 5mm wide linear COB high-brightness adhesive LED light strips. The equilateral triangular wooden strip 3 is a 150mm long and 5mm wide equilateral triangular wooden strip. The three light strips 2 are each pasted on one side of the triangular wooden strip 3. The three light strips 2 are connected in parallel and powered by the main control board's USB interface. The visual sensor module consists of three USB cameras connected to the main control board. These cameras are positioned below the cylindrical lampshade on three sides to obtain an omnidirectional field of view around the individual. The acquired image information is then transmitted to the main control board via USB for analysis. The power module outputs a 5V voltage to power the main control board. The main control board is a Raspberry Pi 4B. Its input is image information captured by three cameras. Based on the image information, it calculates the relative distance and angle with neighboring robots and writes it into the motherboard storage. It can support its own control program to call or transmit with other terminals within the local area network.

[0028] This invention also provides a method for position perception of swarm robots based on rectangle recognition, comprising the following steps: Step 1: Acquire image information from three cameras. The image obtained by camera a is image a, the image obtained by camera b is image b, and the image obtained by camera c is image c. Step 2: Convert the three images acquired by the vision sensor module to grayscale to generate grayscale images. Perform Gaussian filtering on the grayscale images to form a grayscale histogram. Set a threshold based on the grayscale histogram and perform binarization on the grayscale images. Step 3: Use OpenCV, a cross-platform computer vision and machine learning software library released under the Apache 2.0 license (open source), to perform erosion and dilation processing on the binarized image obtained in Step 2; Step 4: Use OpenCV to filter the bright spots in the image processed in Step 3 by size and remove bright spots with fewer than 200 pixels; Step 5: Use OpenCV to obtain the minimum inscribed rectangle of each bright patch in the image processed in Step 4, and use the x-coordinate of the center point of the rectangular bright patch generated by the neighboring robot in the image as the position of the corresponding robot in the corresponding image; Step 6: Remove duplicate robot markers from the three images and calculate the relative angle of the corresponding robot using the horizontal axis; Step 7: Calculate the relative distance to the corresponding robot based on the height of the rectangular bright blocks in the image.

[0029] The calculation method for the relative angle of the corresponding robot in step 6 is as follows: A polar coordinate system for the robot's platform is established with the centroid of an equilateral triangle formed by the robot's three cameras as the pole, the midline of the field of view of camera a as the polar axis, and the counterclockwise direction as the positive direction. The images captured by the three cameras have a resolution of 1920*1080, and the field of view of each camera is 140°. The field of view of image a captured by camera a corresponds to an angle of 290° to 70°, image b captured by camera b corresponds to an angle of 50° to 190°, and image c captured by camera c corresponds to an angle of 170° to 310°. A Cartesian coordinate system is established in each image, with the lower left corner of the image as the origin. Any point on each of the three images is then represented as follows: , and .

[0030] Taking the calculation of the relative angle between neighbors in image a acquired by any robot camera a as an example, assuming the coordinates of the center point of the rectangular bright patch generated by the neighboring robot in image a are... The horizontal coordinate of the robot's position in image a is The relative angle between the neighboring robots can then be expressed as: ; The relative angles between neighboring robots in any robot image b can be represented as: ; The relative angles between neighboring robots in any robot image c can be expressed as: .

[0031] The method for calculating the relative distance between the corresponding robots in step 7 is as follows: First, the cameras are calibrated. Assume the distance between any robot and its neighboring robots is... At that time, the pixel height of the rectangular bright block of the neighboring robot in the image captured by the camera is 1. When the neighboring robot captures a rectangular bright block in the image with a pixel height of... When the relative distance between neighboring robots is reached, it can be expressed as: . Detailed implementation method: Figure 1 This is a structural diagram of a cluster robot position perception system based on rectangle recognition. The system includes a light marker module, a vision sensor module, a main control board, and a power supply module. The light marker module generates a rectangular bright area, allowing the robot to identify the relative position and distance to neighboring robots. The vision sensor module consists of three USB cameras connected to the main control board, positioned below the cylindrical light cover on three sides to obtain an omnidirectional field of view around the robot. The acquired image information is transmitted to the main control board via USB for analysis. The main control board is a Raspberry Pi 4B. Its input is the image information captured by the three cameras. Based on the image information, it calculates the relative distance and angle with surrounding neighboring robots and writes it to the mainboard storage. This allows for calls to its own control program or transmission with other terminals within a local area network. The power supply module provides a stable 5V voltage to the main control board.

[0033] Figure 2 This is a schematic diagram of the structure of the lighting sign module of the present invention. The lighting sign module consists of a connecting base plate, a cylindrical lampshade, three LED light strips, and an equilateral triangular wooden strip. The three LED light strips are respectively fixed on the three sides of the equilateral triangular wooden strip, and the three light strips are connected in parallel and powered through the USB interface of the main control board. The cylindrical lampshade is made of frosted acrylic tube, and the top and bottom ends of the lampshade have black opaque strip-shaped parts. The lighting sign module is assembled and fixed by the connecting base plate and connected to the robot below, and three evenly distributed slots are provided to fix three cameras.

[0034] Figure 3 This is a schematic diagram of the camera numbering of the present invention.

[0035] Figure 4 This diagram illustrates the method for calculating the relative angle and relative distance between neighboring robots according to the present invention. The determination of the relative angle depends on the identification of the center of the rectangular bright patch in the acquired image, and the determination of the relative distance depends on the identification of the height of the rectangular bright patch in the acquired image. An OpenCV-based recognition library can be used. OpenCV is a cross-platform computer vision and machine learning software library released under a BSD license (open source). It consists of a series of C functions and a small number of C++ classes. It also provides interfaces for languages ​​such as Python, Ruby, and MATLAB, implementing many general-purpose algorithms in image processing and computer vision.

[0036] First, we implement the identification of the center and height of the rectangular bright block: 1. Convert the three images acquired by the vision sensor module to grayscale to generate grayscale images; perform Gaussian filtering on the grayscale images to form a grayscale histogram; set a threshold based on the grayscale histogram and perform binarization on the grayscale images.

[0037] 2. Perform erosion and dilation processing on the binarized image obtained in step 1.

[0038] 3. Filter the bright spots in the image after step 2 by size and remove bright spots with fewer than 200 pixels.

[0039] 4. Call the `minAreaRect` function in OpenCV to obtain the center coordinates of the smallest inscribed rectangle of each bright patch in the image processed in step 3. and rectangle pixel height The robot markers that were counted repeatedly in the three images were filtered out.

[0040] Then, calculate the relative angle and relative distance of the neighboring robots according to the calculation method of relative angle and relative distance.

[0041] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the scope of the technology disclosed in the present invention, and such modifications or substitutions should all be covered within the scope of protection of the present invention.

Claims

1. A cluster robot position perception system based on rectangle recognition, characterized in that... The system, located on top of the robot, includes a light and marking module, a vision sensor module, a main control board, and a power supply module. The light-marking module is used to generate a rectangular bright block as a neighboring individual identification mark; the light-marking module consists of a cylindrical lampshade, three LED light strips and an equilateral triangular wooden strip; the three LED light strips are respectively fixed on the three sides of the equilateral triangular wooden strip; the cylindrical lampshade is made of frosted acrylic tube, and the upper and lower ends of the lampshade have black opaque strip-shaped parts; The visual sensor module is used to collect image information of the individual's omnidirectional field of vision and transmit it to the main control board for calculation and analysis; The main control board connects the light sign module and the vision sensor module. It is used to process the images collected by the vision sensor, calculate the relative distance and angle of neighboring robots, and store the processed results in its storage unit for use by the cluster algorithm program. The power module is used to supply power to the main control board.

2. The cluster robot position perception system based on rectangle recognition according to claim 1, characterized in that, The visual sensor module consists of three cameras, which are oriented in multiple directions to obtain an omnidirectional field of view around the individual.

3. The cluster robot position perception system based on rectangle recognition according to claim 1, characterized in that, The main control board is a Raspberry Pi 4B, which can support its own control program calls or transmission with other terminals within a local area network.

4. A method for swarm robot position perception based on rectangle recognition, implemented using the swarm robot position perception system based on rectangle recognition as described in any one of claims 1-3, characterized in that, include: S1: Convert the acquired multiple images to grayscale to generate a grayscale image; A Gaussian filter is applied to the grayscale image to form a grayscale histogram. A threshold is set based on the grayscale histogram, and the grayscale image is then binarized. S2: Perform erosion and dilation processing on the binarized image; S3: Filter the bright spots in the image after erosion and dilation processing by size, and remove the bright spots that are too small; S4: Obtain the center coordinates and pixel height of the smallest inscribed rectangle of each bright patch in the image, and use the x-coordinate of the center point of the rectangular bright patch generated by the neighboring robot in the image as the position of the neighboring robot in the corresponding image; S5: Calculate the relative angle of neighboring robots using the horizontal axis; S6: Calculate the relative distance to neighboring robots using the pixel height of the rectangular bright blocks in the image.

5. The method for position perception of a swarm robot based on rectangle recognition according to claim 4, characterized in that, It also includes robots that filter out duplicate calculations in multiple images.

6. The method for position perception of a swarm robot based on rectangle recognition according to claim 4, characterized in that, The filtering specifically involves removing bright spots with fewer than 200 pixels.

7. The method for position perception of a swarm robot based on rectangle recognition according to claim 4, characterized in that, The relative angle between neighboring robots is calculated using the horizontal axis as follows: in, M * N For image resolution, Let x be the x-coordinate of the midpoint of the line connecting the centers of the rectangular bright blocks, and let y be the field of view angle of the camera. ~ ,if For negative values, the above formula needs to be adjusted. Take the remainder.

8. The method for position perception of a swarm robot based on rectangle recognition according to claim 4, characterized in that, The relative distance to neighboring robots is calculated using the height of the rectangular bright blocks in the image as follows: in, When calibrating the camera, the pixel height of the rectangular bright blocks of neighboring robots in the image captured by the camera is determined. When calibrating the camera, the distance between the robot and its neighboring robots is... The pixel height of the rectangular bright block on the image captured by the neighboring robot.

9. A method for position perception of a swarm robot based on rectangle recognition according to claim 4, characterized in that, The process was performed using OpenCV, a cross-platform computer vision and machine learning software library released under the Apache 2.0 license.

Images