CN110433467A - Picking up table tennis ball robot operation method and equipment based on binocular vision and ant group algorithm - Google Patents

Abstract

The invention discloses a method and equipment for picking up table tennis robots based on binocular vision and an ant colony algorithm. Coordinate position; obtain the depth data of the binocular camera, and obtain the distance between each table tennis ball and the camera; calculate the coordinate point set of the table tennis ball in the global coordinate system; use the ant colony algorithm to plan the shortest traversal path for the coordinate point set, and obtain a preliminary The ball-picking path; remove the ping-pong balls that have passed through many times from the preliminary ball-picking path to obtain a path without redundant routes. The invention uses a binocular camera to realize the detection and positioning of the table tennis ball at the same time, which is convenient and efficient, and at the same time ensures the accuracy, making the ball picking process more intelligent; the redundant route optimization is carried out on the path planned by the ant colony algorithm, which further reduces the number of robots. The walking path saves time.

Description

Operation method and equipment for table tennis picking robot based on binocular vision and ant colony algorithm

technical field

The invention relates to the technical field of robots, in particular to an operating method and equipment for a table tennis picking robot based on binocular vision and an ant colony algorithm.

Background technique

With the development of artificial intelligence technology and my country's social economy, service robots have gradually entered the homes of the public. As a national game, table tennis is very popular among the people. However, the subsequent work of picking up balls is very troublesome. At present, there is a special person in the table tennis club to be responsible for picking up the balls, and the labor cost is relatively high. And there is not a table tennis robot that can be popularized on the market, and the only research is also focused on academic research, and the actual engineering applications are seldom, and most of them only have scattered methods.

The existing detection and positioning technology is complex and cumbersome. It is difficult to use ordinary RGB cameras to complete the positioning task of table tennis. Therefore, most ball-picking robots cannot locate the ball, so it is impossible to realize the global traversal path planning and navigation of all balls. The method is very inefficient.

Ant colony algorithm is a bionic algorithm, mainly used for TSP problems, it can find a near optimal solution in a short time. However, when it is actually applied to the ball-picking robot, it is limited by the width of the robot and needs to be further optimized.

Contents of the invention

The purpose of the present invention is to overcome the deficiencies in the prior art, proposes a kind of table tennis picking robot operating method and equipment based on binocular vision and ant colony algorithm, realizes the detection and location of table tennis, and the path that goes out of planning Redundant route optimization is carried out to further reduce the walking path of the robot and save time.

For solving the problems of the technologies described above, the invention provides a kind of robot operation method of picking up table tennis based on binocular vision and ant colony algorithm, it is characterized in that, comprises following process:

Establish the global coordinate system of the table tennis training ground;

Obtain the RGB image of the scattered situation of the ground table tennis balls in the venue captured by the binocular camera, detect the table tennis balls from the RGB image and obtain the coordinate positions of each table tennis ball;

Obtain the depth data of the binocular camera, and get the distance between each table tennis ball and the camera;

According to the coordinate position of each table tennis ball and the distance from the camera, calculate the coordinate point set of the table tennis ball under the global coordinate system;

Use the ant colony algorithm to plan the shortest traversal path for the coordinate point set, and get the preliminary ball picking path;

Remove the ping-pong balls that have passed multiple times from the initial ball picking path to obtain a path without redundant paths.

Further, the specific process of detecting table tennis balls from the RGB image and obtaining the coordinate positions of each table tennis ball includes:

Convert an RGB image to an HSV image;

Convert the HSV image into a grayscale image for Gaussian blurring;

Use the canny operator for contour detection to obtain the contours of all objects in the picture;

By using Hough circle transformation to detect the circles in the picture, the table tennis balls in the field can be detected, and the coordinate positions of all table tennis balls in the picture can be calculated.

Further, according to the coordinate position of each table tennis ball and the distance from the camera, the coordinate point set of the table tennis ball under the global coordinate system includes the following process:

Convert the coordinate position of each table tennis ball and the distance from the camera to the three-dimensional coordinate system of the camera, and obtain the coordinate point set of the table tennis ball under the three-dimensional coordinate system of the camera;

Then transform the coordinates of the table tennis ball in the camera's three-dimensional coordinate system to the robot coordinate system;

Finally, transform the coordinates of the table tennis ball in the robot coordinate system to the global coordinate system, and obtain the coordinate point set of the table tennis ball in the global coordinate system.

Further, the process of converting the coordinate positions of each table tennis ball and the distance from the camera to the three-dimensional coordinate system of the camera includes:

Establish camera three-dimensional coordinate system XYZ, image plane coordinate system xy;

The mapping relationship between the point P(X, Y, Z) in the camera’s three-dimensional coordinate system and the point p(x, y) in the image plane coordinate system is:

Among them, f is the focal length of the camera, Z is the depth of the ping-pong ball obtained from the depth data, x represents the x-axis coordinate of the image plane coordinate system, X represents the X-axis coordinate of the camera's three-dimensional coordinate system; y represents the image plane coordinate system The y-axis coordinate of the camera, Y indicates the Y-axis coordinate in the camera's three-dimensional coordinate system; c _x is the coordinate of the main point on the x-axis, c _y is the coordinate of the main point on the y-axis, and f _x is the coordinate of f on the x-axis Pixel dimension, f _y is the pixel dimension of f on the y-axis;

The coordinate position of each table tennis ball is the coordinate in the image plane coordinate system, and the distance from the camera is the Z-axis value in the camera coordinate system. According to the above formula, the three-dimensional coordinates P(X, Y, Z) of the table tennis ball in the camera coordinate system are calculated. .

Further, the process of converting the coordinates of the table tennis ball in the robot coordinate system to the global coordinate system is:

At the beginning, the robot coordinate system is at the origin of the global coordinate system, the coordinates are (x ₀ , y ₀ , th ₀ )=(0, 0, 0), and the moving speed of the robot is V=(V _x , V _y , V _th ) , V _x refers to the speed along the x-axis direction, V _y refers to the speed along the y-axis direction, and V _th refers to the robot rotation speed; after the dt time, the coordinates of the robot coordinate system relative to the global coordinate origin are:

(x ₁ , y ₁ , th ₁ )=(x ₀ +(V _x *cos(th ₀ )-V _y *sin(th ₀ ))*dt, y ₀ +(V _x *sin(th ₀ )+ V _y *cos(th ₀ ))*dt, th ₀ +V _th *dt) (5)

According to the coordinate conversion relationship of the robot coordinate system obtained by formula (5), coordinate conversion can be carried out, and the target table tennis coordinates (X', Y') under the robot coordinate system are converted to the coordinates (X ", Y) under the global coordinate system ")for:

(X″, Y″)=(x ₁ +(X′*cos(th ₁ )-Y′*sin(th ₁ )),

y ₁ +(X′*sin(th ₁ )+Y′*cos(th ₁ )))

Complete the coordinate transformation of all table tennis balls, and obtain the coordinate point set of table tennis balls in the global coordinate system.

Further, removing the ping-pong balls that have passed through multiple times from the preliminary ball picking path, and obtaining a path without redundant routes includes the following process:

According to the preliminary ball picking path, assuming that the robot moves from the starting point to the first target table tennis ball,

Draw straight lines b ₁ , b ₂ and d ₁ , d ₂ , b ₁ , b ₂ are parallel to the starting point and the target table tennis point, and are tangent to the minimum circumscribed circle of the robot at the starting point; straight line d ₁ and The direction of the line connecting two points is perpendicular to and tangent to the tail of the smallest circumscribed circle of the robot at the starting point; the straight line d ₂ is perpendicular to the direction of the line connecting two points and tangent to the head of the smallest circumscribed circle of the robot at the target point; these four lines constitute The rectangle represents the area swept by the robot from the starting point to the target point, calculate and obtain all the table tennis coordinate points in the rectangle formed by four lines, delete these points from the coordinate point set,

Then use this target point as the starting point, select the next target point from the initial ball picking path, and perform the same operation above; The path of the route.

Correspondingly, the present invention also provides a computer-readable storage medium storing one or more programs, and the one or more programs include instructions, and the instructions, when executed by a computing device, cause the computing device to perform the above-mentioned method.

Correspondingly, the present invention also provides a computing device, including:

one or more processors, memory, and one or more programs, wherein one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs include Instructions for performing the methods described above.

Compared with the prior art, the beneficial effects achieved by the present invention are: the present invention uses the binocular camera to realize the detection and positioning of the table tennis ball at the same time, which is convenient and efficient, and at the same time ensures the accuracy, making the ball picking process more intelligent; The path planned by the group algorithm is optimized for redundant routes, which further reduces the walking path of the robot and saves time.

Description of drawings

Fig. 1 is a schematic flow sheet of the method of the present invention;

Fig. 2 is the schematic flow chart of table tennis detection and location;

Fig. 3 is a schematic diagram of a camera pinhole imaging model;

Figure 4 is a conversion diagram of the base_link and map coordinate systems;

Fig. 5 is a schematic flow chart of ant colony algorithm for traversal path planning;

Fig. 6 is a schematic diagram of redundant points.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings. The following examples are only used to illustrate the technical solution of the present invention more clearly, but not to limit the protection scope of the present invention.

A kind of picking table tennis robot operating method based on binocular vision and ant colony algorithm of the present invention, referring to shown in Figure 1, comprises the following steps:

Step 1. Obtain the map of the table tennis training venue and establish a global coordinate system.

First, use the lidar and the gmapping mapping program under the ROS (Robot Operating System) platform to obtain a static map of the table tennis training ground, and then denoise and modify the map to obtain an available global static grid map. Carry out navigation and obstacle avoidance. As a global coordinate system map, this coordinate system takes a corner point of the rectangular training field as the origin, and the length and width directions of the training field are used as the horizontal and vertical axes respectively. The ping-pong balls detected later will be converted to this coordinate system for display.

Step 2, obtain the RGB image of the binocular camera, detect the ping-pong ball from the RGB image and obtain its coordinate point in the figure.

The binocular camera uses kinect, which has an RGB color camera and a depth camera. The binocular camera is installed in the center of the front of the robot, with the angle of view looking straight ahead, and the height from the ground is preferably 15cm to 40cm. The robot takes pictures of the ping pong ball on the ground before the start of the ball picking task, and simultaneously receives the RGB data and depth data of the kinect at the same time in the ROS platform to ensure that the time stamps of the two data obtained are consistent.

As shown in Figure 2, take out the RGB image containing several table tennis balls captured by kinect at a certain time, convert the image into an HSV image, filter according to the HSV value threshold of the white table tennis ball, and take out the white part of the image, that is, take out Image of white objects including ping pong balls. Then convert the HSV image into a grayscale image and perform Gaussian blurring to make the edges of the image smoother. Then use the canny operator for contour detection, set the threshold to 100 to 200, and take out the contours of all objects in the picture. Finally, use the Hough circle transformation to detect the circle in the picture, and then the table tennis balls in the field (the coordinates of the table tennis balls in the picture) can be detected, and the coordinate point sets of all table tennis balls in the picture can be calculated, and these detected The table tennis balls are numbered from 1 to N (the number of table tennis balls) according to the order of detection.

Step 3, obtain the depth data of the binocular camera, obtain its corresponding depth information (that is, the horizontal distance between the table tennis and the camera) according to the two-dimensional coordinates of the table tennis ball, and calculate the distance of the table tennis ball under the robot coordinate system according to the depth information. coordinate.

According to the camera pinhole model, that is, the idealized pinhole imaging model, two coordinate systems are obtained, as shown in Figure 3. One is the camera's three-dimensional coordinate system XYZ, where the X axis is parallel to the ground and the image plane, and the Y axis is perpendicular to On the ground, the Z axis is perpendicular to the image plane; one is the image plane coordinate system xy, and the image plane is perpendicular to the ground; according to the intercept theorem in geometry, the midpoint P (X, Y, Z) of the camera measurement coordinate system and the image plane coordinates are obtained The mapping relationship between the midpoint p(x, y) of the system is:

Among them, f is the focal length of the camera, Z is the depth of the ping-pong ball obtained from the depth data, x represents the x-axis coordinate of the image plane coordinate system, X represents the X-axis coordinate of the camera's three-dimensional coordinate system; y represents the image plane coordinate system The y-axis coordinate of , Y represents the Y-axis coordinate in the camera's three-dimensional coordinate system.

According to the internal parameters of the camera, the above steps are improved:

Due to actual differences, the principal point of each camera (the origin of the image plane coordinate system) is not necessarily in the exact center of the image; the pixels in the x and y directions are also different, so the above formula (1) (2) is transformed into :

Among them, c _x is the coordinate of the main point on the x-axis, c _y is the coordinate of the main point on the y-axis, f _x is the pixel dimension of f on the x-axis, f _y is the pixel volume of f on the y-axis outline.

As shown in flow chart 2. First call the interface function to obtain the internal reference matrix of the camera, then calculate the three-dimensional coordinates P (X, Y, Z) of the target table tennis ball in the camera coordinate system according to the above formula (3) (4) (X, Y are obtained by the above calculation, Z is the depth of the point, which is the horizontal distance from the ping-pong ball to the camera). Since the camera's three-dimensional coordinate system and the robot coordinate system (named base_link, the robot's forward direction is the X axis, and the left is the Y axis), there is a conversion relationship. The final coordinates of the table tennis ball in the robot coordinate system are (X′, Y ')=(Z, X) (assuming that the table tennis balls are all on the ground, and the height coordinate is 0), and displayed in RVIZ (ROS visualization tool) under the ROS platform.

Repeat the above process to get the set of coordinate points in the robot coordinate system composed of all the table tennis balls on the ground in the image.

Step 4: According to the conversion relationship between the robot coordinate system and the global coordinate system, coordinate conversion is performed on the table tennis ball in the robot coordinate system to obtain the coordinate point set of the table tennis ball in the global coordinate system.

The global coordinate system is named map, and the robot coordinate system is named base_link. At the beginning, the base_link coordinate system is at the origin of the map coordinate system (the coordinates are (x ₀ , y ₀ , th ₀ )=(0, 0, 0), where x, y, and th refer to the x-axis coordinates and y-axis coordinates, respectively. coordinate system offset angle), let the robot moving speed V=(V _x , V _y , V _th ), V _x refers to the speed along the x-axis direction, V _y refers to the speed along the y-axis direction, V _th refers to Robot rotation speed. Then after the dt time, the coordinates of the base_link coordinate system relative to the origin of the global coordinates are:

(x ₁ , y ₁ , th ₁ )=(x ₀ +(V _x *cos(th ₀ )-V _y *sin(th ₀ ))*dt, y ₀ +(V _x *sin(th ₀ )+ V _y *cos(th ₀ ))*dt, th ₀ +V _th *dt) (5)

Coordinate conversion can be performed according to the coordinate conversion relationship of the base_link coordinate system obtained by formula (5). The coordinates in the global coordinate system can be calculated from the origin (x ₁ , y ₁ ) of the base_link coordinate system and the deflection angle th ₁ (shown in FIG. 4 ). The target table tennis coordinates (X′, Y′) in the robot coordinate system base_link are converted to the coordinates (X″, Y″) in the global coordinate system map as:

(X″, Y″)=(x ₁ +(X′*cos(th ₁ )-Y′*sin(th ₁ )),

y ₁ +(X′*sin(th ₁ )+Y′*cos(th ₁ )))

So far, all the coordinate transformation work has been completed.

Step 5: Use the ant colony algorithm to plan the shortest traversal path for the obtained table tennis coordinate point set, and obtain the preliminary order of picking up the ball.

First, according to the numbering sequence when detecting table tennis balls, calculate the distance between all table tennis balls and combine them into a two-dimensional adjacency matrix Graph. Calculation of the shortest path. The elements in the matrix are the distances between the table tennis balls, for example, row i and column j represent the distance between the i-th ball and the j-th ball. If there are N ping-pong balls, the number of elements in the matrix is N ² .

As shown in Figure 5, at the beginning, the parameters of the ant colony algorithm α (weighted value of pheromone), β (weighted value of visibility), ρ (evaporation rate of pheromone), Q (pheromone gain coefficient), initialization Pheromone τ ₀ , number of iterations N.

Then randomly select a ping-pong ball as the starting point for each ant, and add the starting point to the point set C that has already passed. Calculate the probability P=τ ^α *(γ) ^β from the current point to all the remaining points (set O) that have not been passed, where τ is the pheromone pheromone between two points, and γ is the distance between two points (distance from adjacent Obtained in the matrix graph) reciprocal. Using a roulette wheel approach, the next point is chosen for each ant. That is, take a random number q, calculate the probability of the first point, if it is less than q, continue to calculate the next point; if the cumulative sum of the probability of reaching a certain point is greater than q, then take this point. For example, q=0.4, p ₁ =0.2, p ₂ =0.3, p ₃ =0.5, (p _n represents the probability of selecting the nth point), then since p ₁ <0.4, and p ₁ +p ₂ >0.4, So take the second point as the next target. By analogy, move all ants to the next target point, and add this point to the set that has passed (set C).

Repeat the above steps until all the points are added to the already passed set. This completes one iteration.

Calculate the path length of each ant according to the selected path, that is, calculate the distance between the table tennis balls according to the path sequence and accumulate the summation, for example, the path sequence is 2-1-3-4 (the serial number has been calibrated in the previous step) , then the path length is the distance d ₁ from the No. 2 table tennis ball to the No. 1 table tennis ball, plus the distance d ₂ from the No. 1 ball to the No. 3 ball, plus the distance d ₃ from the No. 3 ball to the No. 4 ball, that is, L= d ₁ +d ₂ +d ₃ . Update the pheromone of each edge after all ants have finished walking: the pheromone increment between two points on each path is inversely proportional to the length of their path: add=Q/L _k (Q is generally 50-100), Among them, L _k represents the length of the kth path, and add represents the increase amount of pheromone. Therefore, the updated pheromone is τ=(1-ρ)*τ ₀ +Q/L _k .

Use the new pheromone to continue the next round of iteration; if (1) the number of iterations reaches the set value N (50 times) and all have been completed, or (2) the pheromone increment add is less than a certain threshold (such as 0.01), then the output traversal Approximate optimal solution (shortest path length) for all points.

Step 6: According to the shape of the robot itself and the planned path, remove the ping-pong balls that have passed many times, and further shorten the path length.

As shown in Figure 6 (in the global coordinate system map), according to the order obtained from the path planning in step 5, it is assumed that the first target table tennis ball moved by the robot is point 2 (the digital serial number is calibrated according to the order of table tennis balls identified in step 2, Point 0 represents the current coordinate point of the robot when it recognizes the table tennis ball, and point 2 is the coordinate point of the second recognized table tennis ball (the number has been calibrated when detecting the table tennis ball), which is used as the first coordinate point in path planning. target point), draw straight line b ₁ , b ₂ and d ₁ , d ₂ , b ₁ , b ₂ are parallel to the line connecting the two points of point 0 and point 2, and the minimum circumscribed line with the robot at point 0 The circle (because the shape of the robot may be irregular, the smallest circle that can completely surround the robot is called the minimum circumscribed circle) is tangent; the straight line d ₁ is perpendicular to the direction connecting the two points and touches the tail of the minimum circumscribed circle of the robot at the starting point Tangent; the straight line d ₂ is perpendicular to the direction connecting the two points and tangent to the head of the smallest circumscribed circle of the robot at point 2; the rectangle formed by these four lines represents the area swept by the robot from point 0 to point 2 ( For the sake of efficiency, the parts not covered by the four corners of the rectangle are ignored). Calculate and obtain all table tennis coordinate points in the rectangle formed by four lines, that is, points 4, 7, and 15 in Figure 4, and delete these points from the coordinate point set, because the robot carries a suction device or a roller picking device, so In the process of moving from point 0 to point 2, the robot will pass through these table tennis balls and pick them up, that is, the robot can pick up all the balls within the radius width of the smallest circumscribed circle. Deleting these points reduces redundant traversal points and improves operational efficiency.

Then use point 2 as the starting point, select the next target point from the path planning in step 5, and perform the same operation on these two points according to the operation process of point 0 and point 2. By analogy, repeat step 6 until all points on the path calculated in step 5 are processed, and finally a path without redundant routes is obtained. The efficiency of the robot picking up the ball according to this path is higher than that of only step 5 The path run in is efficient and saves time.

Correspondingly, the present invention also provides a computer-readable storage medium storing one or more programs, and the one or more programs include instructions, and the instructions, when executed by a computing device, cause the computing device to perform the above-mentioned method.

Correspondingly, the present invention also provides a computing device, including:

one or more processors, memory, and one or more programs, wherein one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs include Instructions for performing the methods described above.

Those skilled in the art should understand that the embodiments of the present application may be provided as methods, systems, or computer program products. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.

The present application is described with reference to flowcharts and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present application. It should be understood that each procedure and/or block in the flowchart and/or block diagram, and a combination of procedures and/or blocks in the flowchart and/or block diagram can be realized by computer program instructions. These computer program instructions may be provided to a general purpose computer, special purpose computer, embedded processor, or processor of other programmable data processing equipment to produce a machine such that the instructions executed by the processor of the computer or other programmable data processing equipment produce a An apparatus for realizing the functions specified in one or more procedures of the flowchart and/or one or more blocks of the block diagram.

These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing apparatus to operate in a specific manner, such that the instructions stored in the computer-readable memory produce an article of manufacture comprising instruction means, the instructions The device realizes the function specified in one or more procedures of the flowchart and/or one or more blocks of the block diagram.

These computer program instructions can also be loaded onto a computer or other programmable data processing device, causing a series of operational steps to be performed on the computer or other programmable device to produce a computer-implemented process, thereby The instructions provide steps for implementing the functions specified in the flow chart or blocks of the flowchart and/or the block or blocks of the block diagrams.

The above is only a preferred embodiment of the present invention, it should be pointed out that for those of ordinary skill in the art, without departing from the technical principle of the present invention, some improvements and modifications can also be made, these improvements and modifications It should also be regarded as the protection scope of the present invention.

Claims (8)

1. a kind of robot operation method of picking up table tennis based on binocular vision and ant colony algorithm, is characterized in that, comprises following process: Establish the global coordinate system of the table tennis training ground; Obtain the RGB image of the scattered situation of the ground table tennis balls in the venue captured by the binocular camera, detect the table tennis balls from the RGB image and obtain the coordinate positions of each table tennis ball; Obtain the depth data of the binocular camera, and get the distance between each table tennis ball and the camera; According to the coordinate position of each table tennis ball and the distance from the camera, calculate the coordinate point set of the table tennis ball under the global coordinate system; Use the ant colony algorithm to plan the shortest traversal path for the coordinate point set, and get the preliminary ball picking path; Remove the ping-pong balls that have passed multiple times from the initial ball picking path to obtain a path without redundant paths. 2. a kind of picking up table tennis robot operating method based on binocular vision and ant colony algorithm according to claim 1, is characterized in that, detects table tennis from RGB image and obtains the concrete process of the coordinate position of each table tennis include: Convert an RGB image to an HSV image; Convert the HSV image into a grayscale image for Gaussian blurring; Use the canny operator for contour detection to obtain the contours of all objects in the picture; By using Hough circle transformation to detect the circles in the picture, the table tennis balls in the field can be detected, and the coordinate positions of all table tennis balls in the picture can be calculated. 3. a kind of picking up table tennis robot operating method based on binocular vision and ant colony algorithm according to claim 1, is characterized in that, obtains under the global coordinate system according to the coordinate position of each table tennis and distance calculation from camera The coordinate point set of table tennis includes the following process: Convert the coordinate position of each table tennis ball and the distance from the camera to the three-dimensional coordinate system of the camera, and obtain the coordinate point set of the table tennis ball under the three-dimensional coordinate system of the camera; Then transform the coordinates of the table tennis ball in the camera's three-dimensional coordinate system to the robot coordinate system; Finally, transform the coordinates of the table tennis ball in the robot coordinate system to the global coordinate system, and obtain the coordinate point set of the table tennis ball in the global coordinate system. 4. a kind of picking up table tennis robot operation method based on binocular vision and ant colony algorithm according to claim 3, is characterized in that, the coordinate position of each table tennis ball and the distance from camera are converted to under the camera three-dimensional coordinate system The process includes: Establish camera three-dimensional coordinate system XYZ, image plane coordinate system xy; The mapping relationship between the point P(X, Y, Z) in the camera’s three-dimensional coordinate system and the point p(x, y) in the image plane coordinate system is: Among them, f is the focal length of the camera, Z is the depth of the ping-pong ball obtained from the depth data, x represents the x-axis coordinate of the image plane coordinate system, X represents the X-axis coordinate of the camera's three-dimensional coordinate system; y represents the image plane coordinate system The y-axis coordinate of the camera, Y indicates the Y-axis coordinate in the camera's three-dimensional coordinate system; c _x is the coordinate of the main point on the x-axis, c _y is the coordinate of the main point on the y-axis, and f _x is the coordinate of f on the x-axis Pixel dimension, f _y is the pixel dimension of f on the y-axis; The coordinate position of each table tennis ball is the coordinate in the image plane coordinate system, and the distance from the camera is the Z-axis value in the camera coordinate system. According to the above formula, the three-dimensional coordinates P(X, Y, Z) of the table tennis ball in the camera coordinate system are calculated. . 5. a kind of picking up table tennis robot operating method based on binocular vision and ant colony algorithm according to claim 3, is characterized in that, the process that the coordinate conversion of table tennis under the robot coordinate system to the global coordinate system is: : At the beginning, the robot coordinate system is at the origin of the global coordinate system, the coordinates are (x ₀ , y ₀ , th ₀ )=(0, 0, 0), and the moving speed of the robot is V=(V _x , V _y , V _th ) , V _x refers to the speed along the x-axis direction, V _y refers to the speed along the y-axis direction, and V _th refers to the robot rotation speed; after the dt time, the coordinates of the robot coordinate system relative to the global coordinate origin are: (x ₁ , y ₁ , th ₁ )=(x ₀ +(V _x *cos(th ₀ )-V _y *sin(th ₀ ))*dt, y ₀ +(V _x *sin(th ₀ )+ V _y *cos(th ₀ ))*dt, th ₀ +V _th *dt) (5) According to the coordinate conversion relationship of the robot coordinate system obtained by formula (5), coordinate conversion can be carried out, and the target table tennis coordinates (X', Y') under the robot coordinate system are converted to the coordinates (X ", Y) under the global coordinate system ")for: (X″, Y″)=(x ₁ +(X′*cos(th ₁ )-Y′*sin(th ₁ )), y ₁ +(X′*sin(th ₁ )+Y′*cos(th ₁ ))) Complete the coordinate transformation of all table tennis balls, and obtain the coordinate point set of table tennis balls in the global coordinate system. 6. a kind of picking up table tennis robot operating method based on binocular vision and ant colony algorithm according to claim 1, is characterized in that, removes the table tennis ball that passes through repeatedly from preliminary picking up ball path, obtains no redundancy A route's path includes the following procedures: According to the preliminary ball picking path, assuming that the robot moves from the starting point to the first target table tennis ball, Draw straight lines b ₁ , b ₂ and d ₁ , d ₂ , b ₁ , b ₂ are parallel to the starting point and the target table tennis point, and are tangent to the minimum circumscribed circle of the robot at the starting point; straight line d ₁ and The direction of the line connecting two points is perpendicular to and tangent to the tail of the smallest circumscribed circle of the robot at the starting point; the straight line d ₂ is perpendicular to the direction of the line connecting two points and tangent to the head of the smallest circumscribed circle of the robot at the target point; these four lines constitute The rectangle represents the area swept by the robot from the starting point to the target point, calculate and obtain all the table tennis coordinate points in the rectangle formed by four lines, delete these points from the coordinate point set, Then take this target point as the starting point, select the next target point from the preliminary ball picking path, and perform the same operation above; The path of the route. 7. A computer-readable storage medium storing one or more programs, the one or more programs comprising instructions which, when executed by a computing device, cause the computing device to perform any one of claims 1-6. A kind of operation method of picking table tennis robot based on binocular vision and ant colony algorithm described in the item. 8. A computing device comprising, One or more processors, memory and one or more programs, wherein one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs include The instruction for executing the operation method of a table tennis robot based on binocular vision and ant colony algorithm described in any one of claims 1-6.