Customizable ping pong robot ball hitting method, apparatus, and storage medium

Abstract

This invention discloses a customizable table tennis robot hitting method, device, and storage medium. The method includes: acquiring user-defined target hitting parameters, or selecting target hitting parameters corresponding to a specific hitting technique, wherein the target hitting parameters include target racket posture and target racket speed; identifying and tracking the position of the table tennis ball to obtain its predicted trajectory; selecting a desired receiving position; and controlling the robot to perform a hitting action at the desired receiving position. Through customization, the table tennis robot can execute specific hitting techniques while retaining a certain degree of randomness. Reproducible hitting techniques allow humans to train countermeasures against different playing styles, while the randomness enhances the fun of human training and practice. This invention also allows users to participate in the formulation of hitting strategies, increasing user engagement in human-machine competition and allowing for personalized customization of hitting strategies.

Description

Technical Field

[0001] This invention relates to the field of table tennis robot control technology, and in particular to a customizable table tennis robot hitting method, device, and storage medium. Background Technology

[0002] A table tennis robot is an assistive device designed for table tennis, enabling it to play against the user. Compared to other devices like table tennis ball machines, it enhances training enjoyment, and users can improve their skills by playing against the robot. For a table tennis robot, the hitting strategy is the core of its control system. Currently, the hitting strategies for table tennis robots include the following:

[0003] (1) A hitting strategy model that can control the landing position of a ping-pong ball is obtained by training based on physical model analysis or through data-driven methods such as reinforcement learning, and then the strategy model is transferred to reality. The strategy model takes data such as the real-time position of the ping-pong ball, the real-time speed of the ping-pong ball, and the expected landing position of the ping-pong ball as input, and the hitting time, hitting position, hitting posture, and racket speed (or directly output joint speed) as output. Then, robotics is used to realize the hitting action of the strategy output. A typical example is the KUKA ping-pong robot from the University of Tübingen.

[0004] (2) Collect data on human-robot matches, recording data such as the maximum and minimum ball speed, ball landing point, and initial ball position. Based on this data, simulate the human's hitting trajectory extensively in a virtual environment, and train the robot in the virtual environment using the simulated hitting methods. After training, transfer the training results from the virtual environment to the real environment and repeat the above process repeatedly to maximize the number of rounds of human-robot matches. A typical example of this hitting strategy is the Google eight-DOF ping-pong robot, which uses a two-DOF guide rail + a six-DOF ABB serial robot.

[0005] (3) By combining reinforcement learning with self-play, the robot can repeatedly practice against itself in a virtual environment, thereby enhancing its competitive ability. A typical example of this ball-striking strategy is the ping-pong robot trained in the Google PyBullet simulation environment.

[0006] While the aforementioned hitting strategies allow the robot to perform relatively good hitting actions, controlling the landing point of the ping-pong ball, adapting to human hitting methods, and using self-play to control the robot's hitting deprives humans of the ability to customize the racket posture and racket speed when the robot hits the ball. Furthermore, since various hitting techniques (such as push, attack, chop, and topspin) are all determined by the racket posture and racket speed at the time of hitting, this ultimately prevents humans from customizing the hitting techniques used by the robot. The applicant's prior application CN113650010A provides a motion control method, system, and storage medium for a ping-pong robot, which presents a kinematic solution method for the ping-pong robot. However, the following problems still exist:

[0007] (1) Humans cannot train against specific hitting techniques;

[0008] (2) It is impossible to make the robot maintain a fixed rhythm of human-machine interaction, such as continuously using techniques such as reducing power and slow chopping to achieve a slow rhythm of low intensity, or continuously using techniques such as attacking and smashing to achieve a fast rhythm of high intensity.

[0009] (3) As human ball-striking skills gradually improve, people’s demand for robot ball-striking skills will also change, but the above methods cannot meet this demand. Summary of the Invention

[0010] Purpose of the invention: In order to overcome the shortcomings of the existing technology, the present invention provides a customizable table tennis robot hitting method, device and storage medium that can be customized to allow humans to train countermeasures against different playing styles.

[0011] Technical solution: To achieve the above objectives, the present invention provides a customizable table tennis robot hitting method, the method comprising:

[0012] (a) Identify and track the position of a ping-pong ball to obtain its predicted trajectory;

[0013] (b) Select the expected receiving position;

[0014] (c) Control the robot to perform a hitting action at the intended receiving position;

[0015] Prior to step (a), the method further includes:

[0016] (d) Obtain user-defined target hitting parameters, or select target hitting parameters corresponding to a specific hitting technique, wherein the target hitting parameters include target racket posture and target racket speed;

[0017] Step (b) specifically includes:

[0018] The expected receiving position is selected based on the predicted trajectory and the target hitting parameters;

[0019] Step (c) specifically involves controlling the ping-pong robot to complete the shot based on the target shot parameters and the expected receiving position.

[0020] Furthermore, the step of selecting the expected receiving position based on the predicted trajectory and the target hitting parameters specifically includes:

[0021] A rationality analysis is performed on the target hitting posture and target racket speed;

[0022] When the target hitting posture and / or target racket speed are reasonable, the target hitting parameters are directly used as the execution parameters;

[0023] When the target hitting posture and / or target racket speed are unreasonable, the unreasonable parameters will be adjusted to reasonable values, and the adjusted values ​​will be used as execution parameters.

[0024] The expected receiving position is selected based on the predicted trajectory and the execution parameters.

[0025] Furthermore, controlling the operation of the table tennis robot to complete the shot based on the target hitting parameters and the expected receiving position specifically includes:

[0026] Based on the racket posture and racket speed in the execution parameters, the joint parameters of each joint in the table tennis robot are calculated based on robot kinematics, and the joint parameters include target position and target velocity.

[0027] Based on the joint parameters, spatial trajectory planning is performed for each joint;

[0028] Based on the joint parameters and spatial trajectory, each joint is driven to move to the expected receiving position to complete the shot.

[0029] Furthermore, in the process of controlling the operation of the table tennis robot based on the target hitting parameters and the expected receiving position to complete the hitting, random errors are introduced into the parameters of specific joints so that the actual receiving position of the racket when the table tennis robot receives the ball differs from the expected receiving position.

[0030] Furthermore, the table tennis robot includes a dual-axis translational module and a robotic arm. The robotic arm includes a base, an upper arm, a lower arm, and an end effector, with the racket mounted on the end effector. The random error is applied to the rotational joint between the upper arm and the base.

[0031] Furthermore, if a suitable expected receiving position cannot be selected during the process, the hitting parameters corresponding to the alternative hitting technique are changed, and the expected receiving position is selected again.

[0032] A customizable table tennis robot hitting device includes:

[0033] The trajectory prediction module is used to identify and track the position of the ping-pong ball, thereby obtaining the predicted trajectory of the ping-pong ball.

[0034] The position selection module is used to select the expected receiving position;

[0035] An execution module, which controls the robot to perform a hitting action at the expected receiving position;

[0036] Also includes:

[0037] The definition module is used to obtain user-defined target hitting parameters or select target hitting parameters corresponding to a specific hitting technique. The target hitting parameters include target racket posture and target racket speed.

[0038] The location selection module is used to implement:

[0039] The expected receiving position is selected based on the predicted trajectory and the target hitting parameters;

[0040] The execution module is used to implement:

[0041] Based on the target hitting parameters and the expected receiving position, the table tennis robot is controlled to operate and complete the hitting.

[0042] The storage medium contains an executable program that, when executed by a processor, enables the implementation of the customizable table tennis robot hitting method described above.

[0043] Beneficial Effects: The customizable table tennis robot hitting method, device, and storage medium of this invention enable the table tennis robot to execute specific hitting techniques while retaining a certain degree of randomness. Reproducible hitting techniques allow humans to train countermeasures against different playing styles, while the randomness enhances the enjoyment of human training and practice. Furthermore, this invention allows users to participate in the formulation of hitting strategies, increasing user engagement in human-machine competition and enabling personalized customization of hitting strategies. Attached Figure Description

[0044] Figure 1 This is a structural diagram of a table tennis robot;

[0045] Figure 2 A flowchart illustrating a customizable table tennis robot's hitting method;

[0046] Figure 3 This is a schematic diagram of the components of a customizable table tennis robot's striking device. Detailed Implementation

[0047] The invention will now be further described with reference to the accompanying drawings.

[0048] Example 1

[0049] The customizable table tennis robot's hitting method in this embodiment is implemented by the table tennis robot's control system, such as... Figure 1 As shown, the ping-pong robot includes a dual-axis translational module 1 and a robotic arm 2. The robotic arm 2 includes a base 21, a large arm 22, a forearm 23, and an end effector 24, which are sequentially rotatably connected to form an open-chain structure. The racket 25 is mounted on the end effector 24. The method includes the following steps S101-S104:

[0050] Step S101: Obtain user-defined target hitting parameters, or select target hitting parameters corresponding to a specific hitting technique. The target hitting parameters include target racket posture and target racket speed.

[0051] In this step, the aforementioned hitting techniques include push shots, attack shots, chop shots, and topspin shots. Users can issue operation commands to the control system through the pre-installed human-computer interaction units (such as buttons, touchscreens, etc.) in the table tennis robot, or through terminals such as PP that can establish a network connection with the table tennis robot. User-defined target hitting parameters can be obtained through preset interfaces, which can be hardware interfaces and / or software interfaces. Hardware interfaces include USB interfaces, and software interfaces include network interfaces. Users can provide target hitting parameters that differ from traditional hitting techniques to the table tennis robot's control system as needed for specific training. In addition, the aforementioned target racket posture and target racket speed can be fixed values, or at least one of them can be a numerical range.

[0052] Back Figure 2 Step S102: Identify and track the position of the ping-pong ball to obtain its predicted trajectory.

[0053] In this step, the identification and position tracking of the ping-pong ball are based on images acquired by the camera and image algorithms. The camera can acquire three-dimensional position data of the ping-pong ball at multiple moments. The trajectory of the ping-pong ball is fitted based on all the three-dimensional position data, and the predicted trajectory is obtained based on the extension trend of the current trajectory.

[0054] Back Figure 2 Step S103, select the expected receiving position; specifically: select the expected receiving position according to the predicted trajectory and the target hitting parameters;

[0055] In this step, different hitting parameters generally result in different receiving positions for the same incoming ball. For example, the receiving positions for hitting techniques such as push, attack, chop, and topspin will definitely differ. Therefore, when selecting the expected receiving position on the predicted trajectory, it is necessary to refer to the target hitting strategy.

[0056] Back Figure 2 Step S104: Control the robot to perform a hitting action at the expected receiving position;

[0057] Specifically, this step involves controlling the ping-pong robot to complete the shot based on the target shot parameters and the expected receiving position.

[0058] Preferably, the step of selecting the expected receiving position based on the predicted trajectory and the target hitting parameters specifically includes the following steps S201-S204:

[0059] Step S201: Perform a rationality analysis on the target hitting posture and target racket speed;

[0060] The purpose of this step is to analyze whether the ping-pong robot can achieve the target posture and target racket speed.

[0061] Step S202: When the target hitting posture and / or target racket speed are reasonable, the target hitting parameters are directly used as execution parameters.

[0062] Step S203: When the target hitting posture and / or target racket speed are unreasonable, adjust the unreasonable parameters to reasonable values ​​and use the adjusted values ​​as execution parameters;

[0063] In this step, the adjusted value can be selected according to the principle of proximity. For example, if the target racket speed exceeds the maximum racket speed that the ping-pong robot can achieve, then the maximum racket speed will be used as the adjusted racket speed.

[0064] Step S204: Select the expected receiving position based on the predicted trajectory and the execution parameters.

[0065] Preferably, the step of controlling the table tennis robot to complete the shot based on the target shot parameters and the expected receiving position specifically includes the following steps S301-S303:

[0066] Step S301: Based on the racket posture and racket speed in the execution parameters, calculate the joint parameters of each joint in the table tennis robot based on robot kinematics. The joint parameters include the target position and the target velocity.

[0067] Step S302: Based on the joint parameters, perform spatial trajectory planning for each joint;

[0068] Step S303: Based on the joint parameters and spatial trajectory, drive each joint to operate to the expected ball receiving position to complete the shot.

[0069] Preferably, in step S303, during the process of controlling the operation of the table tennis robot based on the target hitting parameters and the expected receiving position, random errors are introduced into the parameters of specific joints to ensure that the actual receiving position of the racket when receiving the ball differs from the expected receiving position. In practice, these differences can be controlled within a preset range to prevent excessive deviations that could cause the ball to go out of bounds or fail to clear the net. The introduction of these random errors allows the robot to achieve a certain degree of randomness in its hitting technique. This randomness ensures that the trajectory of each return shot by the robot varies to some extent, thereby further enhancing the enjoyment of playing and improving the training effect for humans. Specifically, factors can be randomly selected from an array with many factors and multiplied with the parameters of a specific joint to obtain the actual control parameters. These adjusted control parameters are then used to control the operation of the relevant joints in subsequent steps.

[0070] Preferably, such as Figure 1 As shown, the random error is applied to the rotating joint between the upper arm 22 and the base 21, with the axis of the rotating joint extending vertically. Choosing this joint as the point of application of the random error allows it to cause a deviation in the angular velocity (w in the figure), which in turn causes a deviation in the angular velocity of the racket 25. This angular velocity error of the racket 25 can create a linear velocity error proportional to the error distance, thus causing a certain disturbance in the linear velocity of the racket 25 when the ping-pong robot receives a ball. This allows the ping-pong robot to possess a certain degree of randomness when performing specific hitting techniques, preventing trainees from consistently receiving balls along fixed trajectories and improving training effectiveness.

[0071] Preferably, if a suitable expected receiving position cannot be selected in the steps, the hitting parameters corresponding to the backup hitting technique are changed, and the expected receiving position is selected again. The aforementioned backup hitting technique can be preset by the user or selected automatically by the control system. Thus, when the table tennis robot cannot receive the ball according to the target hitting parameters selected in step S101, it can use other backup hitting techniques to receive the ball, ensuring the continuity of training. When the situation returns to where the target hitting parameters can be used to receive the ball, the control system selects the expected receiving position again based on the target hitting parameters to receive the ball.

[0072] Example 2

[0073] This embodiment discloses a customizable table tennis robot hitting device 400 (hereinafter referred to as: device 400). Device 400 may include or be divided into one or more program modules. One or more program modules are stored in a storage medium and executed by one or more processors to complete the present invention and realize the above-mentioned customizable table tennis robot hitting method. The program module referred to in this embodiment refers to a series of computer program instruction segments capable of performing a specific function, which is more suitable than the program itself for describing the execution process of the customizable table tennis robot hitting method in the storage medium. The following description will specifically introduce the functions of each program module in this embodiment:

[0074] The definition module 401 is used to obtain user-defined target hitting parameters or select target hitting parameters corresponding to a specific hitting technique. The target hitting parameters include target racket posture and target racket speed.

[0075] The trajectory prediction module 402 is used to identify and track the position of the ping-pong ball, thereby obtaining the predicted trajectory of the ping-pong ball.

[0076] The position selection module 403 is used to select the expected receiving position; specifically, it selects the expected receiving position based on the predicted trajectory and the target hitting parameters.

[0077] The execution module 404 is used to control the robot to perform a hitting action at the expected receiving position; specifically, it controls the operation of the table tennis robot to complete the hitting action based on the target hitting parameters and the expected receiving position.

[0078] Other aspects of implementing the customizable table tennis robot hitting method based on device 400 have been described in detail in previous embodiments. Please refer to the corresponding content in the previous embodiments. They will not be repeated here.

[0079] Example 3

[0080] This embodiment also provides a computer-readable storage medium, such as flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, server, App application store, etc., which stores a computer program. When the program is executed by a processor, it implements the corresponding function. In this embodiment, the computer-readable storage medium is used in storage device 400, and when executed by a processor, it implements the customizable table tennis robot hitting method of the present invention.

[0081] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A customizable table tennis robot hitting method, the method comprising: (a) Identify and track the position of a ping-pong ball to obtain its predicted trajectory; (b) Select the expected receiving position; (c) Control the robot to perform a hitting action at the intended receiving position; The method is characterized in that, prior to step (a), it further includes: (d) Obtain user-defined target hitting parameters, or select target hitting parameters corresponding to a specific hitting technique, wherein the target hitting parameters include target racket posture and target racket speed; Step (b) specifically includes: The expected receiving position is selected based on the predicted trajectory and the target hitting parameters; Step (c) specifically involves: controlling the table tennis robot to operate based on the target hitting parameters and the expected receiving position, thereby completing the hitting; The table tennis robot is controlled to operate based on the target hitting parameters and the expected receiving position. During the hitting process, random errors are introduced into the parameters of specific joints so that the actual receiving position of the racket when the table tennis robot receives the ball differs from the expected receiving position.

2. The customizable table tennis robot hitting method according to claim 1, characterized in that, The step of selecting the expected receiving position based on the predicted trajectory and the target hitting parameters specifically includes: A rationality analysis is performed on the target racket posture and target racket speed; When the target racket posture and / or target racket speed are reasonable, the target hitting parameters are directly used as the execution parameters. When the target racket posture and / or target racket speed are unreasonable, the unreasonable parameters will be adjusted to reasonable values, and the adjusted values ​​will be used as execution parameters. The expected receiving position is selected based on the predicted trajectory and the execution parameters.

3. The customizable table tennis robot hitting method according to claim 2, characterized in that, The step of controlling the ping-pong robot to complete the shot based on the target hitting parameters and the expected receiving position specifically includes: Based on the racket posture and racket speed in the execution parameters, the joint parameters of each joint in the table tennis robot are calculated based on robot kinematics, and the joint parameters include target position and target velocity. Based on the joint parameters, spatial trajectory planning is performed for each joint; Based on the joint parameters and spatial trajectory, each joint is driven to move to the expected receiving position to complete the shot.

4. The customizable table tennis robot hitting method according to claim 1, characterized in that, The ping-pong robot includes a dual-axis translation module (1) and a manipulator (2). The manipulator (2) includes a base (21), an upper arm (22), a lower arm (23), and an end arm (24). A racket (25) is mounted on the end arm (24). The random error is applied to the rotational joint between the upper arm (22) and the base (21).

5. The customizable table tennis robot hitting method according to claim 1, characterized in that, If a suitable expected receiving position cannot be selected in step (b), change the hitting parameters corresponding to the backup hitting technique and select the expected receiving position again.

6. A customizable table tennis robot striking device, comprising: The trajectory prediction module (402) is used to identify and track the position of the ping-pong ball, thereby obtaining the predicted trajectory of the ping-pong ball; Position selection module (403) is used to select the expected receiving position; Execution module (404) is used to control the robot to perform a hitting action at the expected receiving position; Its characteristic is that it further includes: The definition module (401) is used to obtain user-defined target hitting parameters or select target hitting parameters corresponding to a specific hitting technique. The target hitting parameters include target racket posture and target racket speed. The location selection module (403) is used to implement: The expected receiving position is selected based on the predicted trajectory and the target hitting parameters; The execution module (404) is used to implement: Based on the target hitting parameters and the expected receiving position, the table tennis robot is controlled to operate and complete the hitting; The table tennis robot is controlled to operate based on the target hitting parameters and the expected receiving position. During the hitting process, random errors are introduced into the parameters of specific joints so that the actual receiving position of the racket when the table tennis robot receives the ball differs from the expected receiving position.

7. A storage medium, characterized in that, It contains an executable program that, when executed by a processor, can implement the customizable table tennis robot hitting method as described in any one of claims 1-5.

Images