Ice hockey intelligent training method, system and device and storage medium

Abstract

The invention relates to an ice hockey intelligent training method, system and device and a storage medium, and belongs to the technical field of physical training, and the method comprises the steps that a mobile robot perceives the surrounding environment in real time, and recognizes the state information of an ice hockey and a trained player according to the perceived surrounding environment in real time; the mobile robot dynamically generates and executes a simulation behavior instruction based on the state information recognized in real time and a predefined training mode; in the training process, the mobile robot monitors and records an interaction event with the trained player, after training is completed, a training report is generated according to the interaction event, and the training report is sent to a preset monitoring terminal so that a user holding the monitoring terminal can know the training report. The ice hockey intelligent training method can simulate a real competition scene, has environment perception and intelligent decision-making capabilities, and can provide personalized data feedback.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of sports training, in particular to an intelligent ice hockey training method, system, device and storage medium. BACKGROUND

[0002] Ice hockey is a high-speed, strong-competitive team project, and the individual skills and team tactical cooperation ability of athletes are crucial. The traditional ice hockey training mode highly depends on the subjective experience of coaches and manual sparring. This mode has many limitations, such as the training intensity and continuity are limited by the physical fitness of the coaches and sparring partners, and it is difficult to provide high-intensity, high-frequency repetitive training To overcome the above problems, some automated equipment such as an ice hockey shooting machine is introduced in the prior art. However, such equipment can usually only perform fixed-point, pre-programmed shooting operations, and its behavior is static and pre-set, so it cannot interact and intelligently compete with athletes in real time; leading to a disconnection between the training scene and actual combat, and limited training effect, so there is room for improvement. SUMMARY

[0003] In order to provide an ice hockey auxiliary training solution that can simulate real game scenarios, has environmental perception and intelligent decision-making ability, and can provide personalized data feedback, the present application provides an intelligent ice hockey training method, system, device and storage medium.

[0004] In a first aspect, the present application provides an intelligent ice hockey training method, which is executed by an intelligent ice hockey training system including a pre-set mobile robot and a monitoring terminal. The intelligent ice hockey training method specifically includes: The mobile robot perceives the surrounding environment in real time, and identifies the state information of the ice hockey and the trained player in real time according to the perceived surrounding environment; wherein the state information at least includes position information; The mobile robot dynamically generates and executes simulation behavior instructions based on the real-time identified state information and the pre-defined training mode; wherein the simulation behavior instructions are used to simulate the action behavior of the ice hockey player to perform real-time dynamic confrontation or cooperation training with the trained player; During the training process, the mobile robot monitors and records the interaction events with the trained player, and after completing the training, generates a training report according to the interaction events, and sends the training report to the pre-set monitoring terminal, so that the user holding the monitoring terminal knows the training report.

[0005] By adopting the above technical solution, and through real-time perception of the surrounding environment and dynamic generation of simulated behavioral commands, the mobile robot is no longer an automated machine executing fixed programs, but a smart training device capable of reacting instantly to real-time changes in the training environment (such as the movement of trained players or the movement of the puck). Furthermore, since the mobile robot's behavior (i.e., simulated behavioral commands) is generated based on real-time perceived state information, it can simulate the intelligent behavior of opponents or teammates in a real match, thus providing trained players with a training environment closer to actual combat and effectively training their on-the-spot reaction and decision-making abilities. Finally, by monitoring interactive events and generating training reports, the training effects, traditionally reliant on subjective observation, are digitized and visualized. This allows monitoring terminal holders (such as coaches) to evaluate the trained players' performance based on concrete training reports, thereby optimizing training effectiveness through improvements to the entire training process.

[0006] Optionally, the step of dynamically generating and executing simulated behavior instructions based on real-time identified state information and predefined training modes includes: The mobile robot predicts the future trajectory of the ice hockey puck based on the real-time identification of the puck's status information. If the current training mode is a mobile practice mode for simulating dynamic cooperation, the mobile robot generates and executes simulated behavior instructions, which are specifically: based on the real-time predicted future trajectory of the puck, it moves to the puck's landing point before the puck to receive the puck, and executes a passing action based on the real-time position information of the trained player. If the current training mode is a defensive mode for simulating confrontation, the mobile robot generates and executes simulated behavior instructions, which specifically include: moving to the defensive area formed between the trained player and the goal based on the real-time position information of the trained player, and moving to the future trajectory of the ice puck based on the real-time predicted future trajectory of the ice puck to perform an interception action.

[0007] By adopting the above technical solution, this solution publicly explains the simulated behavioral instructions under different training modes. For example, in the mobile training mode, by predicting the landing position of the puck and moving to that position before the puck, it simulates the role of a teammate with excellent predictive ability, performing receiving and passing operations. In the defensive mode, it simulates positioning defense by moving between the trained player and the goal, and simulates interception defense by moving along the future trajectory of the puck. It is evident that the "trajectory prediction" and "predictive movement" in this solution are the decisive features that distinguish it from the "pre-programmed" devices mentioned in the prior art. It proves that the mobile robot of this application is not merely "responding" to the environment, but also "understanding" and "predicting" the environment, which is precisely the manifestation of its intelligence level.

[0008] Optionally, the method further includes: The mobile robot provides real-time feedback on its own status and training progress, which is available to the trained player and / or monitoring terminal. The mobile robot's status includes at least its movement state and the state of being hit by the puck. The training progress includes at least the current training mode.

[0009] By adopting the above technical solution, the feedback content enables direct and real-time interaction between the mobile robot and the trained player during the training process. This allows the trained player to understand the intention of the mobile robot (such as whether a teammate is preparing to receive a pass or an opponent is preparing to defend) without the need for external devices during high-speed training, thereby making faster reactions and enhancing the immersion in training.

[0010] Optionally, the method further includes: The monitoring terminal constructs and updates the individual ability model of the trained player, wherein the individual ability model includes at least the technical strengths and weaknesses of the trained player identified based on training data prior to the current time. The method of dynamically generating and executing simulated behavior instructions based on real-time identified state information and predefined training modes includes: Based on the individual ability model corresponding to the trained player, the mobile robot adaptively adjusts the details of the simulated behavioral instructions so that: when the training mode is a mobile sparring mode for simulating dynamic cooperation, the generated and executed simulated behavioral instructions tend to guide and amplify the technical advantages of the trained player; when the training mode is a defensive mode for simulating confrontation, the generated and executed simulated behavioral instructions tend to target and train the technical weaknesses of the trained player. The mobile robot executes the simulated behavior instructions and records the mobile robot's movement trajectory and auxiliary training strategies during the execution of the simulated behavior instructions, generates a corresponding training template, and adds the training template to the final generated training report; wherein, the auxiliary training strategies include the mobile robot's passing strategy when the training mode is a mobile sparring mode for simulating dynamic cooperation, or the mobile robot's defensive strategy when the training mode is a mobile sparring mode for simulating dynamic cooperation.

[0011] By adopting the above technical solutions, the training content is dynamically tailored to the unique skill profiles of different trained players. This solves the deep-seated problem of "single training modes and inability to provide personalized data feedback" in the background technologies, elevating data processing from simple "counting" (such as the number of hits) to the level of "ability profiling." Furthermore, by utilizing the individual ability models of the trained players, it helps generate simulated behavioral instructions for the mobile robot. For example, in the cooperative mode (as a teammate), this solution is no longer a mechanical passing drill, but actively creates conditions for athletes to leverage their strengths (for instance, for players with a "short-range fast break" advantage, the robot will execute closer, faster "wall-run" instructions; for players with a "long-range attack" advantage, the robot will execute instructions to create space and long-range shooting opportunities). This greatly enhances the athletes' confidence and effectiveness in their areas of expertise. In the defensive mode (as an opponent), this solution follows the same defensive logic as the aforementioned solutions but is more intelligent. It's no longer a general defensive strategy, but rather a targeted attack on specific weaknesses of athletes (for example, targeting the weakness of "weak backhand ball control," the robot will focus on pressuring their backhand side. This high-intensity, repetitive attack on weaknesses can most effectively solidify correct movements and overcome technical deficiencies). Finally, by recording and summarizing the movement trajectory and auxiliary training strategies of the mobile robot in executing the adaptively adjusted simulated behavioral instructions, training templates are generated. This provides coaches with data-driven, practice-proven, high-quality tactical options. For example, coaches can use these to teach other real players the movement trajectories and auxiliary training strategies, thereby quickly improving the team's overall tactical awareness and teamwork.

[0012] Optionally, the method further includes: The mobile robot receives a simulated character creation instruction from the monitoring terminal, the simulated character creation instruction containing feature parameters for defining the behavioral style of the simulated target player; The mobile robot generates a virtual personal ability model based on the feature parameters contained in the simulated character creation instruction; wherein, the virtual personal ability model is used to simulate the technical characteristics of the target player; The mobile robot adaptively adjusts the details of the simulated behavioral instructions based on the individual ability model corresponding to the trained player, so that: when the training mode is a mobile sparring mode for simulating dynamic cooperation, the generated and executed simulated behavioral instructions tend to guide and amplify the technical strengths of the trained player; when the training mode is a defensive mode for simulating confrontation, the generated and executed simulated behavioral instructions tend to target and train the technical weaknesses of the trained player, including: The mobile robot determines the strategic objectives of the simulated behavioral instructions based on the individual ability model corresponding to the trained player. The strategic objectives are: when the training mode is a mobile sparring mode for simulating dynamic cooperation, to guide and amplify the technical advantages of the trained player; or when the training mode is a defensive mode for simulating confrontation, to target and train the technical weaknesses of the trained player. The mobile robot generates simulated behavioral instructions for executing the strategic objectives based on the technical characteristics corresponding to the virtual personal capability model.

[0013] By adopting the above technical solution, this application introduces a simulated individual ability model to further restore the mobile robot to a specific player with specific technical characteristics and physical fitness change patterns. This means that athletes no longer face a general algorithm during training, but rather a highly realistic "hypothetical enemy" or "ideal teammate," providing an unprecedentedly realistic environment for pre-match preparation and tactical drills. The final generated simulated behavioral instructions not only accurately implement the tactical intention (i.e., strategic goal) of training the individual abilities of the trained player, but also strictly follow the objective abilities and behavioral styles of the target player simulated by the mobile robot, making the final generated simulated behavioral instructions intelligent behaviors that are both targeted and consistent with the role logic.

[0014] Optionally, the simulated character creation instructions also include feature parameters for limiting the physical characteristics of the simulated target player; the generated virtual personal ability model is also used to simulate the physical change pattern of the target player; When the mobile robot generates simulated behavioral instructions for executing the strategic objectives according to the technical characteristics corresponding to the virtual personal ability model, it also dynamically adjusts at least one of the following in the simulated behavioral instructions based on the physical fitness change pattern: the mobile robot's moving speed, the mobile robot's reaction time, and the mobile robot's hitting force on the hockey puck.

[0015] By adopting the above-mentioned technical solutions, and further simulating the laws of physical fitness changes, the realism of training is extended from the "technical level" to the "physiological and tactical endurance level." It simulates the fatigue effect of players in real matches, enabling the mobile robot to exhibit states such as decreased speed and sluggish reaction. This provides the trained players with advanced training scenarios on how to expand their advantage when the simulated target player's physical fitness declines, and how to maintain tactical execution when their own physical fitness declines. This represents a revolutionary improvement in the realism of training.

[0016] Optionally, the method further includes: The monitoring terminal aggregates the individual ability models of different trained players. Based on the aggregated individual ability model data, it calculates the ability matching degree between different trained players and tactical roles in the preset tactical formation. Based on the ability matching degree, it generates and outputs a player pairing recommendation list, which includes recommended player combinations and corresponding tactical formations. The player combination includes players assigned to each tactical role in the corresponding tactical formation; and the same player combination satisfies the following: in the corresponding tactical formation, the technical weaknesses and technical strengths of the players can complement each other; and / or the technical strengths of the players can be superimposed to enhance the tactical effect.

[0017] By adopting the aforementioned technical solutions, and building upon the construction of individual ability models for personalized training of individual players, a qualitative leap from individual training to team structure has been achieved. Specifically, by quantifying the individual ability models of all trained players, the system more intelligently identifies player combinations that are difficult for coaches to perceive, providing objective decision support for lineup arrangements in actual ice hockey games. Furthermore, the system evaluates player combinations within a pre-set tactical framework. Through the specific computational dimensions of "complementary abilities" and "synergistic abilities," the system can automatically identify how players can generate additional tactical value by leveraging their strengths and mitigating their weaknesses. For example, it automatically pairs a player skilled in long-range shooting but weak in penetration with a player skilled in dribbling and drawing defenders, forming a highly threatening offensive link, thereby creating a team offensive power that individual players cannot possess.

[0018] Secondly, this application provides an intelligent ice hockey training system, which includes a mobile robot and a monitoring terminal; the mobile robot includes an environmental perception module, a behavior decision control module, an interactive monitoring module, and a communication module. The environmental perception module is used to perceive the surrounding environment in real time and identify the status information of the hockey puck and the trained players based on the perceived surrounding environment; wherein, the status information includes at least location information. The behavior decision control module is used to dynamically generate and execute simulated behavior instructions based on real-time identified status information and predefined training modes; wherein, the simulated behavior instructions are used to simulate the actions of ice hockey players to conduct real-time dynamic confrontation or cooperative training with the trained players. The interaction monitoring module is used to monitor and record interaction events with the trained player during the training process; it is also used to generate a training report based on the interaction events after the training is completed, and then send the training report to the monitoring terminal through the communication module. The monitoring terminal is used to receive and output the training report so that the user holding the monitoring terminal can be aware of the training report.

[0019] Thirdly, this application provides an intelligent ice hockey training device, including a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as described in any of the first aspects.

[0020] Fourthly, this application provides a computer-readable storage medium storing a computer program that can be loaded by a processor and execute the method described in any of the first aspects.

[0021] In summary, this application includes at least one of the following beneficial technical effects: 1. In this application, by sensing the surrounding environment in real time and dynamically generating simulated behavioral commands, the mobile robot is no longer an automated machine executing fixed programs, but becomes an intelligent training device capable of reacting instantly to real-time changes in the training field (such as the movement of trained players or the movement of the puck). Furthermore, since the mobile robot's behavior (i.e., simulated behavioral commands) is generated based on real-time perceived state information, it can simulate the intelligent behavior of opponents or teammates in real matches, thereby providing trained players with a training environment closer to actual combat and effectively training athletes' on-the-spot reaction and decision-making abilities. Finally, by monitoring interactive events and generating training reports, the training effects, which traditionally relied on subjective observation, are digitized and visualized. This allows monitoring terminal holders (such as coaches) to evaluate the training performance of trained players based on concrete training reports, thereby optimizing training effects through improvements to the entire training process. 2. Furthermore, in the mobile training mode, by predicting the puck's landing point and moving to that position before the puck, the robot simulates a teammate with excellent anticipation abilities, performing receiving and passing maneuvers. In the defensive mode, it simulates positioning and defense by moving between the trained player and the goal, and simulates interception and defense by moving along the puck's future trajectory. It is evident that the "trajectory prediction" and "anticipatory movement" in this solution are decisive features distinguishing it from the "pre-programmed" devices mentioned in existing technologies. This demonstrates that the mobile robot of this application not only "responds" to the environment but also "understands" and "predicts" the environment, which is a manifestation of its intelligence level. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating an intelligent ice hockey training method disclosed in an embodiment of this application.

[0024] Figure 2 This is a structural block diagram of an intelligent ice hockey training system disclosed in an embodiment of this application.

[0025] Explanation of reference numerals in the attached drawings: 401, Environmental perception module; 402, Behavior decision control module; 403, Interactive monitoring module; 404, Communication module; 405, Drive unit; 501, Monitoring terminal. Detailed Implementation

[0026] The following is in conjunction with the appendix Figures 1-2 This application will be described in further detail.

[0027] This application discloses an intelligent ice hockey training method. The executing entity of the intelligent ice hockey training method is an intelligent ice hockey training system, which specifically includes a mobile robot and a preset monitoring terminal. (Refer to...) Figure 1 The mobile robot and the monitoring terminal work together to execute this intelligent ice hockey training method. The specific execution process is as follows.

[0028] S101, which senses the surrounding environment in real time and identifies the status information of the hockey puck and the trained players based on the sensed environment; the status information includes at least position information. S102, based on real-time identified status information and predefined training modes, dynamically generates and executes simulated behavior instructions; wherein, the simulated behavior instructions are used to simulate the actions of ice hockey players to conduct real-time dynamic confrontation or cooperative training with the trained players. S103 monitors and records interaction events with the trained players during training. After training is completed, it generates a training report based on the interaction events and sends the training report to a preset monitoring terminal so that the user holding the monitoring terminal can be informed of the training report.

[0029] S201, Real-time feedback on the mobile robot's status and training progress, so that the trained player and / or monitoring terminal can be aware of it; wherein, the mobile robot's status includes at least the movement state and the state of being hit by the puck; the training progress includes at least the current training mode.

[0030] S102 includes the following steps: S1021, based on the real-time identified state information of the ice puck, predict the future trajectory of the ice puck in real time; S1022, If the current training mode is a mobile training mode for simulating dynamic cooperation, then generate and execute simulated behavior instructions, and the simulated behavior instructions are as follows: according to the real-time predicted future movement trajectory of the ice puck, move to the landing position of the ice puck in advance to receive the ice puck, and execute a passing action according to the real-time position information of the trained player. S1023, if the current training mode is a defensive mode for simulating confrontation, then generate and execute simulated behavior instructions, and the simulated behavior instructions are as follows: based on the real-time position information of the trained player, move to the defensive area formed between the trained player and the goal, and based on the real-time predicted future movement trajectory of the ice hockey puck, move to the future movement trajectory of the ice hockey puck to perform an interception action.

[0031] In implementation, the mobile robot includes a robot body, a drive unit for moving the robot body, and an environment perception module, behavior decision control module, interaction detection module, and communication module mounted on the robot body. The drive unit includes a two-wheel differential structure and servos, which work together to achieve precise drive and control of the robot body. The hardware configuration of the two-wheel differential structure is existing technology and will not be elaborated upon. Its core lies in supporting and moving the robot body through two independent wheels, each driven by a motor (left drive wheel and right drive wheel), and one or more omnidirectional driven wheels (such as Mecanum wheels or ball wheels). The behavior decision control module issues high-level movement commands to the drive unit. These commands typically include a target linear velocity vector V (including magnitude and direction) and a target angular velocity ω. The low-level controller within the drive unit (such as a microcontroller or motor drive board) calculates the high-level (V, ω) command into the target rotational speeds of the two drive wheels based on the two-wheel differential kinematic model. The calculation formulas are as follows: Left wheel speed ω_L = (V - ω * L / 2) / R, Right wheel speed ω_R = (V + ω * L / 2) / R. Where ω_L and ω_R are the target speeds (radians per second or converted to RPM) of the left and right drive wheels, respectively; V is the target linear velocity of the robot's center of mass; ω is the target angular velocity of the robot around its center of mass; L is the wheelbase between the two drive wheels (fixed mechanical parameter); and R is the radius of the drive wheel (fixed mechanical parameter). The underlying controller uses a PID control algorithm to compare the actual speed feedback from the drive wheel motor encoders with the calculated target speeds, dynamically adjusting the duty cycle of the PWM (Pulse Width Modulation) signals output to the left and right wheel motors, thereby precisely controlling the motors to reach and maintain the target speeds.

[0032] Specific movement actions include forward / backward (when ω = 0 and V ≠ 0, ω_L = ω_R, the two wheels rotate at the same speed and in the same direction (or in opposite directions), translation (this is existing technology and will not be elaborated on. It is usually achieved by using wheel systems such as Mecanum wheels or omnidirectional wheels in conjunction with differential control. The logic is to control the velocity vectors generated by all wheels to combine into a pure lateral velocity), and zero-radius rotation (when V = 0 and ω ≠ 0, ω_L = -ω_R, the two wheels rotate at equal speeds but in opposite directions, and the robot rotates in place with its center of mass as the center).

[0033] The servo motor's drive logic is as follows: 1. Rotation control: Method one is controlling the driven wheels (the servo motor's output shaft is connected to the steering mechanism of one or more directional driven wheels. When the behavior decision module requires the robot to make precise pointing adjustments with a non-zero radius (e.g., accurately aligning the hitting surface with a teammate before passing the ball), it sends a target angle pulse signal to the servo motor. The control circuit inside the servo motor drives the motor to rotate and rotates the output shaft to a specified angle through the gear set, thereby causing the driven wheels to deflect. This generates a deflection torque during the robot's movement, assisting it in quickly and accurately adjusting the robot's orientation); Method two is controlling the rotating platform (the servo motor drives a small rotating platform equipped with a vision or hitting mechanism. This is existing technology and will not be elaborated further. Its logic is to receive angle commands and rotate the platform relative to the robot body, thereby achieving the orientation of specific components without changing the overall robot's movement trajectory). 2. Braking control includes differential reverse braking (this is existing technology and will not be elaborated on further. After the behavior decision module issues an emergency stop command, the drive unit briefly controls the two drive wheels to reverse at high speed, generating braking force opposite to the direction of movement to achieve an emergency stop) and mechanical braking (the servo controls a mechanical brake pad through a linkage or cam mechanism. When braking is required, the servo receives a signal and drives the brake pad to press against the drive wheel or a dedicated brake disc, achieving braking through friction).

[0034] The following preparations will be performed before training begins: The mobile robot connects to a monitoring terminal (such as a mobile phone or tablet) via a wireless ad hoc network through a communication module (such as Wi-Fi, Zigbee, or ESP-NOW). The user holding the monitoring terminal (such as an instructor) selects the training mode for the current training session (hereinafter referred to as the current training mode) from a pre-loaded training mode software library through the user interface of the monitoring terminal (such as a pre-installed APP on a tablet or mobile phone), thus forming a selection command. This selection command is sent to the mobile robot through the communication module. Correspondingly, the mobile robot also has a pre-loaded training mode software library identical to the one pre-loaded in the control terminal. This software library contains several training modes, including but not limited to: a fixed-point passing mode (the robot remains stationary, maintaining its initial position, and only pushes the puck away upon impact, achieving a pass; for example, this pre-loaded method could be: the robot body is not circular but has multiple edges, such as a triangular cross-section, so that when the puck hits one of its edges, the elastic force of the elastic structure (such as an elastic band) on that edge can push the puck away), a mobile practice mode (the mobile robot simulates a teammate, performing positioning, receiving, and passing), and a dynamic defense mode (the mobile robot acts as an opponent, performing defensive interception, i.e., preventing the puck from entering the goal). The mobile robot learns the current training mode by sending a selection command corresponding to the selection instruction.

[0035] Furthermore, before training begins, the mobile robot is pre-placed at a known initial location on the training field (e.g., the center circle), or its environmental perception module identifies training field markers (such as the center point marker) to determine its initial position (X0, Y0) in the field coordinate system. After completing these operations, the mobile robot uses its included interaction detection module to inform the players on the training field that it is ready to begin training. For example, the interaction detection module includes an LED light strip pre-installed on the robot body. By controlling the LED light strip to display different colors or flashing patterns, the robot's status is reflected (e.g., ready - blue, moving - green, hit by a puck - red, low battery - yellow). The interaction detection module may also include a digital display screen installed on the robot body to display the training progress (e.g., training countdown: displaying the remaining time for this round of training; current training mode: defensive mode / passing mode). The current training mode can indirectly inform the players being trained about the role the mobile robot is currently playing (e.g., the role of an opponent in defensive mode, or a teammate in passing mode).

[0036] After the training countdown begins, the training process commences. Training is considered complete when the countdown ends. During training, the environmental perception module monitors the robot's environment in real time to identify the status information of the hockey puck and the trained player, corresponding to step S101 above. Next, the behavior decision control module generates simulated behavior commands based on the identified status information and a predefined training mode, and controls the robot to execute these commands. This simulates the actions of a hockey player to engage in real-time dynamic competition or cooperative training with the trained player, corresponding to step S102 above. Then, the interactive monitoring module monitors the training process and generates a training report. This report is then sent to the monitoring terminal via the communication module. Finally, the monitoring terminal displays the training report on a pre-set app, allowing users with the monitoring terminal to view it, corresponding to step S103.

[0037] Specifically, for S101, this step is performed by the environment perception module, a combined hardware and software module that includes a camera, an IMU (Inertial Measurement Unit), and an image processing chip (such as an embedded GPU or VPU). After training begins, the camera acquires RGB images at a certain frame rate (e.g., 30fps), and the IMU measures the robot's linear acceleration (ax, ay, az) and angular velocity (wx, wy, wz) in three axes at a higher frequency (e.g., 200Hz). The image processing chip is used to ensure that each frame of image is precisely aligned in time with a set of IMU data through hardware timestamps.

[0038] Next, the image processing chip uses a pre-trained and running convolutional neural network model (such as a target detection model based on YOLO or SSD architecture) to identify the target (the target includes ice hockey, players, and goals). This convolutional neural network model is trained on a large dataset containing images of ice hockey, players (assuming players wear uniforms, which are used as a reference for identification), and goals. The model's input is each frame of image captured by the camera, and its output is a specific visual recognition result. This result is specifically the bounding box (i.e., a rectangular bounding box is used to select the target in the image) and class probability for each target (i.e., ice hockey, player, goal) in the image. For example, the model might output [ice hockey, 98% confidence, bounding box coordinates (x1, y1, x2, y2)] and [player, 95% confidence, bounding box coordinates...].

[0039] For the target identified from the image, a preset algorithm (such as the PnP algorithm) is used to calculate the geometric relationship between the pixel position of the target in the image and its real-world size, thus calculating the target's three-dimensional position (x_rel, y_rel, z_rel) and pose relative to the camera. This is the relative pose of the target (i.e., the position information mentioned in S101). It should be noted that the default PnP algorithm knows the camera's internal parameters (such as focal length and optical center, which can be obtained through pre-calibration) and the physical dimensions of the target (such as the standard diameter of an ice hockey puck, the length, width, and height of a goal, and the height of a player).

[0040] The image processing chip defines a state vector [X, Y, θ, Vx, Vy, ω] ^T for the robot body, which is: the robot's global position (X,Y), orientation θ, velocity (Vx, Vy), and turning angular velocity ω; where T represents transpose, changing it from a row vector to a column vector.

[0041] The image processing chip also pre-runs a filtering algorithm (such as extended Kalman filtering). This algorithm is used to predict the global position and attitude of the robot body at the current moment based on the robot body's global position and velocity at the previous moment and the current IMU data (acceleration a, angular velocity ω). The specific prediction steps are as follows: using the optimal state estimate of the previous moment and the acceleration measured by the IMU in the current time period, predict the robot state (including position and velocity) at the current moment. This prediction result is called the prior state estimate.

[0042] Based on the state X_{k-1} of the previous time step (e.g., k-1), the predicted state vector X_k of the current time step (e.g., k) can be obtained using the formula: X_k = X_{k-1} + V_{x, k-1} * Δt + 0.5 * a_x * Δt^2. Here, X_k refers to the predicted position of the robot body in the X-axis direction of the global coordinate system at the current time k. This is a priori estimate because it is calculated only based on past motions and IMU input. X_{k-1} represents the optimal estimated position of the robot body in the X-axis direction of the global coordinate system at the previous time step k-1, which is the most reliable position obtained after the previous round of Kalman filtering correction. V_{x, k-1} represents the optimal estimate of the robot's velocity in the X-axis direction at the previous time k-1, Δt is the time interval between the previous and current times, and a_x is the average acceleration in the X-axis direction measured by the IMU from time k-1 to time k. Correspondingly, X_0, i.e., the position and velocity corresponding to k-1=0, is the initial position (X0, Y0) of the moving robot before training begins, with a velocity of 0. The formula X_k = X_{k-1} +V_{x, k-1} * Δt + 0.5 * a_x * Δt^2 is the update formula for the first element X (i.e., the X-axis position) in the state vector. Similarly, the formulas for Y (i.e., the Y-axis position), Vx (X-axis velocity), and Vy (Y-axis velocity) in the state vector are: Y_k = Y_{k-1} + V_{y, k-1} * Δt + 0.5 * a_y * Δt^2; V_{x, k} = V_{x, k-1} + a_x * Δt; V_{y, k} = V_{y, k-1} + a_y * Δt.

[0043] After obtaining the prior state estimate X_k^-, the image processing chip needs to correct it to obtain the optimal state estimate. The correction steps are as follows: First, the observation value Z_k needs to be obtained. The methods for obtaining Z_k include two aspects: one is to calculate the pose change through vision: assume that at time k-1 and time k, the robot's camera sees the same site landmark (such as a special corner point). Visual algorithms (such as feature point matching) can be used to calculate the movement of the marker in the image. Using this movement information and the camera model, the robot's displacement and rotation from time k-1 to time k can be directly calculated. That is, the pixel coordinates (u1, v1) of the site marker in frame k-1 and the pixel coordinates (u2, v2) in frame k, along with the camera model (known internal parameters, such as focal length and principal point), are used to calculate, through geometric calculations (e.g., essential matrix or homography matrix decomposition), the rotation and translation that the camera (i.e., the robot itself) must have undergone to cause this pixel change. The calculated result is the robot's displacement (Δx, Δy, Δz) and rotation (Δα, Δβ, Δγ) from time k-1 to time k. This displacement can then be used as the observation value Z_k.

[0044] Secondly, it involves observing stationary targets: the robot identifies a stationary target, such as a goal. Using the PnP algorithm, the position (x_rel, y_rel) of the goal relative to the robot can be calculated. If the absolute position (X_goal, Y_goal) of the goal in the global coordinate system is known, then through coordinate transformation, the robot's own global position (X_cam, Y_cam) can be deduced. This calculated position (X_cam, Y_cam) can then be used as the observation value Z_k.

[0045] Finally, the extended Kalman filter algorithm weighted and fused the predicted value X_k^- and the observed value Z_k to correct the prediction error and obtain a smoother and more accurate final estimate, thus updating the prediction result. The specific fusion calculation formula is: X_k = X_k^- + K * (Z_k - H * X_k^-); where H is the observation matrix, a preset fixed matrix used to map the state vector X_k^- to the space where the observation value is located; the X_k obtained after this weighted fusion is the optimal and smoothest state estimate at the current time k, which is used to calculate the robot's body state at the next time k+1, realizing the cyclic real-time calculation of the robot's body state (i.e., position and velocity).

[0046] Finally, combining the robot's own state vector X_k calculated earlier with the relative position of the target identified by visual recognition, the absolute positions (X_puck, Y_puck) and (X_player, Y_player) of the target (hockey puck, trained player, goal) in the global field coordinate system are calculated through coordinate transformation. By differentiating the absolute positions of consecutive frames (e.g., v = (pos_t - pos_{t-1}) / Δt), the velocity (Vx_puck, Vy_puck) and direction of the target are estimated, which are then used as the state information of the trained player and the hockey puck. In other words, the environmental perception module ultimately outputs a quantized dynamic situation map containing timestamps, including the position, velocity vector, and direction vector of the hockey puck and player relative to the robot, as well as the position and posture of the robot after filtering (i.e., the prediction result); thus completing all execution steps of S101.

[0047] S102 is completed by the behavior decision control module and the drive unit. Specifically, the behavior decision control module runs a decision algorithm, which receives the puck position sequence output by S101 and uses a preset motion dynamics model (such as a uniform deceleration model) to predict the puck's future trajectory. Since the puck's sliding on the ice is mainly decelerated by friction, the decision algorithm fits the position data of the most recent frames (such as the most recent 5 frames before the current time), uses the least squares method to perform curve fitting on the position sequence, and estimates the puck's velocity vector (Vx_0, Vy_0) and deceleration magnitude a (in the opposite direction to the velocity direction) at the current time t0. The predicted position (X_pred, Y_pred) of the puck at any future time t is given by the following formulas: X_pred = X_0 + Vx_0 * t + 0.5 * (-a * Vx_0 / V_0) * t^2; Y_pred = Y_0 + Vy_0 * t + 0.5 * (-a * Vy_0 / V_0) * t^2; where V_0 is the current velocity, and -a * Vx_0 / V_0 is the acceleration component in the X direction, which is used as the future trajectory of the puck for prediction. The landing point refers to a special case of the predicted future trajectory of the puck, namely, the position when the puck's velocity decreases to 0, or the position where the puck's trajectory intersects the rink boundary / goal line. The landing point (X_landing, Y_landing) can be obtained by solving for V(t) = 0 or the intersection of the trajectory and the boundary equation.

[0048] Next, the behavior decision control module generates and executes specific simulated behavior instructions based on the known current training mode. Specifically, when the current training mode is the mobile training mode described in S1022, based on the predicted puck landing position obtained in S1021, the simulated behavior instructions generated include: a movement instruction to move to the puck landing position (X_landing, Y_landing), and a pass instruction to pass the puck to the trained player upon contact with the puck. The specific execution operation of this simulated behavior instruction is as follows: using the A-Star algorithm, starting from the robot's current position and ending at the puck landing position, a search is performed. The optimal path is found by evaluating the total cost of each grid, F = G + Q, where G is the actual cost (usually the movement distance) from the starting point to the current grid, and Q is the estimated cost from the current grid to the destination, i.e., the heuristic function (usually estimated using Manhattan distance or Euclidean distance). The algorithm prioritizes exploring the grid with the smallest F value, and finally outputs a path sequence from the starting point to the destination with the shortest total movement distance and no collisions.

[0049] The behavior decision control module uses the path point sequence calculated by the A-Star algorithm as a series of sub-target points that the robot needs to reach sequentially. Then, a proportional-integral-derivative (PID) controller calculates and adjusts the rotational speeds of the left and right drive wheels (specifically, the PID controller calculates the error by comparing the robot's current orientation with the target orientation. Based on this error, it outputs the adjusted PWM duty cycle signals for the left and right wheels in real time. If a left turn is needed, the PWM value of the left wheel is decreased and the PWM value of the right wheel is increased. If a right turn is needed, the PWM value of the left wheel is increased and the PWM value of the right wheel is decreased). Through this precise differential control, the drive unit is controlled to propel the robot smoothly and accurately along the planned path to the puck's landing point.

[0050] Once the robot moves to the puck's landing point, it will execute the pass command. Specifically, the behavior decision control module obtains the real-time position (X_player, Y_player) of the trained player from S101, and then calculates the pass direction vector (a vector pointing from the robot to the trained player) and the striking force. The direction vector is calculated as (X_player - X_robot, Y_player - Y_robot). The decision control module controls the servo motors of the drive unit to rotate the robot's body (usually a specific face of its triangular structure, i.e., the striking face) to align with the direction vector. This is achieved by calculating the angle between the current robot orientation and the pass direction vector. The striking force is proportional to the required pass distance D, where D is the length of the direction vector. The system determines the force value using a preset lookup table or a simple linear formula (e.g., striking force = K * D, where K is a preset fixed proportional coefficient). Then, through differential drive, the robot body rotates to zero radius until its striking surface is aligned with the player. Next, multi-level control signals are generated, which control servos to rotate rapidly at a certain angle, driving a striking mechanism (such as a paddle or spring lever) connected to the servo to accelerate instantaneously and strike a specific part of the puck, thus transmitting the puck to the player with calculated force and direction. The magnitude of the striking force is controlled by adjusting the servo rotation speed or the final output torque PWM signal. It should be noted here that the interaction detection module included in the mobile robot also includes infrared sensors and LED light strips set on each side wall of the robot body. The infrared sensors are used to detect whether the side wall is hit. If so, it is assumed that it is hit by the hockey puck. At this time, the decision control module knows that the hockey puck has hit the robot body, records and generates an impact event (i.e., an interaction event), and controls the LED light strip on the side wall that was hit to display red, so as to indicate that it has been hit (i.e., the state of the robot body being hit by the hockey puck as mentioned above).

[0051] The behavior decision control module is also used to calculate an initial defensive area (e.g., the training area between the goal and the trained player) when the current training mode is the defensive mode described in S1023. Then, on the line connecting the trained player and the center of the goal, a point is selected as the initial defensive target point (X_defend, Y_defend). This point is usually a preset defensive distance D (e.g., 3 meters) from the player, which can be obtained through vector operations: direction vector = (X_goal - X_player, Y_goal - Y_player); normalized direction vector = direction vector / |direction vector| (i.e., divided by the length of the direction vector); defensive target point (X_defend, Y_defend) = (X_player, Y_player) + normalized direction vector * D. Correspondingly, the behavior decision control module generates simulated behavior instructions, including a movement instruction to move to the defensive target point (X_defend, Y_defend), and an interception instruction after moving to the defensive target point (intercepting the puck to prevent it from entering the goal).

[0052] To execute the simulated behavior command, the behavior decision control module needs to generate a movement path, which is the path for the robot to move from its current position to the defense target point (specifically, a collision-free movement path can be generated using the A-star algorithm). The specific generation method is as follows: the training ground is modeled as a grid map, with each grid cell labeled as "free" or "obstacle" (obstacles may include boundaries, other training equipment, etc.). Starting from the robot's current position and ending at the defense target point (X_defend, Y_defend), the cost of each candidate grid cell is evaluated (cost = actual distance from the starting point to the target point + estimated distance from the target point to the target point, i.e., a heuristic function), ultimately finding the shortest path that avoids all obstacles. This path consists of a series of consecutive grid center points (hereinafter referred to as path points).

[0053] Then, the behavior decision control module uses the path point sequence as the sub-target points that the robot body needs to reach sequentially. For the line segment formed by the robot's current point and the next sub-target point (hereinafter referred to as the sub-stroke), a target orientation angle is calculated. Through a proportional-integral-derivative controller, the rotational speeds of the left and right drive wheels are calculated and adjusted. (If the robot needs to turn left to align with the sub-target, the PID controller will output a control signal to reduce the left wheel speed and increase the right wheel speed, generating a differential rotational torque to turn the robot. Once the robot body is aligned, the controller will output a signal to make the left and right wheels move forward at the same speed, with the forward stroke being the sub-stroke length, thereby realizing the robot body moving to the sub-target point.) This process continues. Finally, the behavior decision control module transmits continuously adjusted PWM duty cycle signals for the left and right wheels to the drive unit, thereby precisely controlling the robot's movement trajectory and allowing it to smoothly follow the planned collision-free path. Since the technology for controlling the robot body's steering and movement is existing technology, and corresponding implementation examples have already been given above, it will not be elaborated further here. Once the robot moves to the initial defensive target point, it monitors the future trajectory of the puck in real time. When the future trajectory of the puck intersects with the initial defensive area (i.e., part of the future trajectory of the puck is within the initial defensive area), the point closest to the robot in the intersecting trajectory is calculated as the optimal intercept point (X_intercept, Y_intercept). This is usually achieved by calculating the shortest distance from the point to the straight line (the puck's trajectory).

[0054] Next, the interception command is executed. This involves calculating an interception path from the robot's current position to the optimal interception point using the same method. Simultaneously, based on the real-time predicted trajectory of the puck, the time T at which the puck reaches the optimal interception point is determined. The robot's speed (speed = interception path / t', t' < T) is then determined based on this time T and the interception path, ensuring the robot reaches the optimal interception point before the puck. Then, based on the path point sequence and the determined speed, the PID controller outputs PWM signals to drive the left and right wheels at high speed, propelling the robot to the optimal interception point. After finding the optimal interception point, as discussed earlier, the robot's own structure (multi-faceted, such as the sidewalls of a triangular structure) serves as the interception tool. Therefore, the robot can be used to block the puck, thus completing the interception. Furthermore, by combining the passing instructions mentioned earlier, an arbitrary point outside the initial defensive zone can be determined. Using this point and the robot's current position, a direction vector is generated (this direction vector points to this point). The ball is then passed to this point (that is, the position of the trained player in the passing instruction to the trained player mentioned earlier is replaced by the position of this point, and the passing instruction is re-executed). This passes the ball to this point, thus driving the puck out of the initial area and achieving defense.

[0055] The behavior decision control module is also used to maintain the current position when the current training mode is fixed-point passing mode (e.g., by instantly locking the wheels through differential reversal, or activating the mechanical braking mechanism to keep the robot body stationary). When the infrared sensor detects a puck hitting the robot, it triggers the LED light strip to quickly switch to red for a short period (e.g., 1 second) before returning to blue (i.e., ready state). This provides the athlete with immediate, visual feedback on the hit, while simultaneously recording the impact event in real time. In summary, all steps of S102 are ultimately implemented.

[0056] For S103, in addition to performing the previously described operation of "real-time feedback on the status of the mobile robot and the training process", the interactive detection module is also used to count the collision events recorded throughout the training process and generate a training report with collision events (such as JSON format: {"total training time": "15min", "number of hits": 25}). Finally, it is sent to the monitoring terminal through the communication module so that the monitoring terminal can display the training report.

[0057] Optionally, the intelligent hockey training method may also include the following steps: S301, Construct and update the individual ability model of the trained player, wherein the individual ability model includes at least the technical strengths and weaknesses of the trained player identified based on training data prior to the current time.

[0058] S302 receives a simulated character creation instruction from the monitoring terminal. The simulated character creation instruction contains feature parameters used to limit the behavioral style and physical characteristics of the simulated target player.

[0059] S303, Generate a virtual individual ability model based on the feature parameters contained in the simulated character creation instruction; wherein, the virtual individual ability model is used to simulate the technical characteristics and physical fitness change patterns of the target player.

[0060] Accordingly, S102 includes the following: S1024, based on the individual ability model corresponding to the trained player, adaptively adjusts the details of the simulated behavioral instructions so that: when the training mode is a mobile training mode for simulating dynamic cooperation, the generated and executed simulated behavioral instructions tend to guide and amplify the technical advantages of the trained player; when the training mode is a defensive mode for simulating confrontation, the generated and executed simulated behavioral instructions tend to target and train the technical weaknesses of the trained player. S1025, execute simulated behavior instructions, and record the mobile robot's motion trajectory and auxiliary training strategies during the execution of simulated behavior instructions, generate corresponding training templates, and add the training templates to the final generated training report; wherein, the auxiliary training strategies include the mobile robot's passing strategy when the training mode is a mobile sparring mode for simulating dynamic cooperation, or the mobile robot's defensive strategy when the training mode is a mobile sparring mode for simulating dynamic cooperation.

[0061] S1024 specifically includes the following sub-steps: S10241, Based on the individual ability model corresponding to the trained player, determine the strategic objectives of the simulated behavioral instructions, wherein the strategic objectives are: when the training mode is a mobile training mode for simulating dynamic cooperation, guide and amplify the technical advantages of the trained player; or when the training mode is a defensive mode for simulating confrontation, target and train the technical weaknesses of the trained player. S10242, Generate simulated behavioral instructions for executing strategic objectives according to the technical characteristics corresponding to the virtual personal ability model. When generating simulated behavioral instructions for executing strategic objectives according to the technical characteristics corresponding to the virtual personal ability model, at least one of the following parameters required in the simulated behavioral instructions—the mobile robot's movement speed, the mobile robot's reaction time, and the force with which the mobile robot strikes the puck—is dynamically adjusted based on the laws governing physical fitness changes.

[0062] In implementation, step S301 is executed by the monitoring terminal, which uses the communication module to acquire environmental perception data (real-time position (X_player, Y_player) and motion vectors (velocity magnitude and direction) of the trained player, real-time position (X_puck, Y_puck) and motion vectors of the puck, and field coordinate system (known goal and boundary positions)) from the mobile robot, as well as interactive monitoring data (impact events of the puck hitting the robot and their timestamps) acquired in step S103. The aforementioned data generated by the same trained player in each training session is collected and stored, and the data generated in each training session is segmented into independent events. The segmentation operation is as follows: The training area is pre-divided into several simple virtual zones (e.g., left flank, center flank, right flank, shooting zone. The boundaries of the zones are defined by a preset coordinate range). When a collision event occurs, the position information of the trained player and the puck is retrieved within a specified time (e.g., within 5 seconds) before the event. Based on the position of the trained player (X_player, Y_player), the virtual zone in which the interaction occurred is determined (e.g., "right flank"). Based on the motion vector of the puck within a specified time (e.g., within 5 seconds) before it hits the robot, and combined with the relative positions of the player and the robot, the player's technical action intention is inferred: if the puck moves towards the robot and the trajectory is straight, it is inferred to be a "passing action"; if the puck's trajectory is not straight and it moves towards the robot, it is inferred to be a "shooting action".

[0063] The monitoring terminal calculates the success rate for each virtual area and each inferred technical action (such as passing and shooting): Passing success rate of a virtual area = Number of passes hit in that virtual area / The number of passing actions occurring within the virtual area; the number of pass hits within the virtual area refers to the number of times the puck successfully hits the robot body (i.e., an impact occurred) when a passing action is inferred within the virtual area. The calculation process for the corresponding pass hit count is as follows: when an impact event occurs, the time of occurrence T_hit is recorded immediately. Then, the instantaneous data snapshot at time T_hit-Δt' (e.g., 0.2 seconds ago) is retrieved. This snapshot includes: the position information of the trained player, the position information of the puck, and the velocity vector of the puck. Based on the coordinates of the trained player, it is determined which virtual area the player is in. Then, the direction of the puck velocity vector is calculated. If the azimuth angle of this direction pointing to the robot body is within a threshold range (i.e., the puck is passed to the robot body, not shot towards the goal), then this impact event is marked as a successful pass impact event occurring within the aforementioned determined virtual area. The number of successful pass events in each virtual area during the entire training process is recorded as the corresponding virtual area pass hit count. Shot success rate = Number of shots hit in the shooting zone / Number of shooting actions occurring in the shooting zone; Here, the number of shots hit in the shooting zone refers to the number of events within the shooting zone that are inferred to be shooting intentions and ultimately result in the puck successfully hitting the robot body (at which point the robot body is located on the goal line). Its calculation logic is the same as the calculation process for the number of passes hits mentioned above. The only difference is the judgment condition: When an impact event occurs and the player's position is within the shooting zone, the direction of the puck's velocity vector is calculated. If the azimuth angle of this direction pointing to the goal is within a threshold range, then the impact event is marked as an impact event that resulted in a successful shot. The number of impact events marked as successful shots is counted as the number of shots hit in the shooting zone.

[0064] The monitoring terminal then calculates the arithmetic mean μ_pass and standard deviation σ_pass of each player's passing success rate in each virtual zone, and determines their technical strengths and weaknesses. The determination is based on the following criteria: Iterating through all zones, if R_pass(Zone_i) > (μ_pass + k * σ_pass) (where k is a preset coefficient, such as 0.5), then the passing in Zone_i is marked as a technical strength; for example, "Technical Strength: Right-wing passing". Iterating through all zones, if R_pass(Zone_i) < (μ_pass - k * σ_pass), then the passing in Zone_i is marked as a technical weakness; for example, "Technical Weakness: Left-wing passing". Finally, a personal ability model with technical strengths and weaknesses is output, for example: Player A_Model: {"Technical Strength: Right-wing passing", "Technical Weakness: Left-wing passing"}. This completes all the steps described in S301.

[0065] For S302 and S303: The coach defines characteristic parameters on the monitoring terminal's APP interface, such as: behavioral style (defensive strategy = "pressing" or "boxing"; passing speed = "fast" or "normal"), and physical characteristics (endurance level = "high" or "medium" or "low"). The monitoring terminal generates simulated role-shaping instructions with the aforementioned characteristic parameters and transmits these instructions to the mobile robot via the communication module. This allows the behavior decision control module to acquire the simulated role-shaping instructions and execute S303. Specifically, the behavior decision control module maps the characteristic parameters to specific data, such as: If the defensive strategy is "pressure", then the mapping is to the internal variable: defensive distance D_defend = 1.5 meters; if the defensive strategy is "block", then the mapping is to the defensive distance D_defend = 3 meters. If the pass speed is "fast", then the PWM power value of the servo at the time of impact is 80%; if the pass speed is "normal", then the PWM power value of the servo at the time of impact is 50%. The specific values ​​of the endurance level are mapped to a linear speed decay function. For example, if the endurance level is "Medium", the corresponding speed decay function is: Current maximum speed V_max(t) = Initial speed * (1 - 0.2 * (t / Total training time)). This means that the longer the training time, the smaller the upper limit of the robot's movement speed. The speed decay functions corresponding to different endurance levels can be pre-set and stored in the behavior decision control module. Finally, the results obtained from mapping behavioral style and physical characteristics are integrated to form a virtual individual ability model, which includes defensive distance, servo PWM power value during ball striking, and speed decay function.

[0066] Next, regarding S1024, which corresponds to the generation of simulated behavioral instructions described in S1022 and S1023 above, S1024 further utilizes the individual ability model and virtual individual ability model to collaboratively generate simulated behavioral instructions. Specifically, the behavior decision control module first queries the individual ability model of the currently trained player, and then determines the strategic objective based on the current training mode and the individual ability model of the trained player. The specific determination logic is as follows: If the current training mode is the mobile training mode, the robot's strategic objective is determined based on the technical advantages in the individual ability model. The strategic objective is defined as: "Prioritize cooperating with the player in the area corresponding to the technical advantage." For example, if the individual ability model shows that the technical advantage is right-side passing, the specific strategic objective is: pass the puck to the right side of the trained player (that is, in the passing instruction, based on the current position of the trained player, select a point at a specified distance from the current position of the trained player to the right, use this point as the position pointed to by the direction vector, determine the direction vector, and generate and execute the corresponding passing instruction), guiding the player to receive the puck on the right side.

[0067] If the current training mode is defensive, the robot's strategic objective is determined based on the technical weaknesses identified in the individual player's ability model. This strategic objective is defined as: "Prioritizing interception of the trained player's attacks in the areas corresponding to their technical weaknesses." For example, if the individual player's ability model indicates a left-wing pass weakness, the specific strategic objective is: designate the area corresponding to the trained player's technical weakness within the initial defensive zone as the primary defensive zone. When the puck's future trajectory intersects with this primary defensive zone, select the point closest to the robot as the optimal interception point from the intersecting trajectories. Alternatively, when the puck's future trajectory intersects with the initial defensive zone (i.e., part of the puck's future trajectory lies within the initial defensive zone), prioritize selecting the trajectory corresponding to the trained player's technical weakness (hereinafter referred to as the strategic trajectory) from the intersecting trajectories. Select the point closest to the robot as the optimal interception point from this strategic trajectory. If no strategic trajectory exists, select the point closest to the robot as the optimal interception point from the intersecting trajectories.

[0068] Then, by leveraging strategic objectives and virtual individual capability models, simulated behavioral instructions are generated collaboratively. Specifically, a basic instruction framework (such as defending the left flank) is generated based on strategic objectives, and then specific instructions are generated based on the feature parameters contained in the virtual individual capability model, such as defending the left flank at a defensive distance of 1.5 meters. In addition, the behavior decision control module maintains a real-time clock t to record the elapsed time of the current training process. It reads the speed decay function (current maximum speed V_max(t) = initial speed * (1 - 0.2 * (t / total training time)) from the virtual personal ability model. Then, when generating any instruction parameters involving speed, reaction time, and force, it multiplies the baseline value of the instruction parameter by (1 - 0.2 * (t / total training time), such as maximum speed = baseline value * (1 - 0.2 * (t / total training time); hitting force = baseline hitting force * (1 - 0.2 * (t / total training time); reaction delay = baseline reaction delay time * (1 - 0.2 * (t / total training time)).

[0069] This application also discloses an intelligent ice hockey training system. (See also...) Figure 2 The ice hockey intelligent training system includes a mobile robot and a monitoring terminal 501; the mobile robot includes an environmental perception module 401, a behavior decision control module 402, an interactive monitoring module 403, and a communication module 404. The environmental perception module 401 is used to perceive the surrounding environment in real time and identify the status information of the hockey puck and the trained player based on the perceived surrounding environment; wherein, the status information includes at least location information. The behavior decision control module 402 is used to dynamically generate and execute simulated behavior instructions based on real-time identified status information and predefined training modes; wherein, the simulated behavior instructions are used to simulate the actions of ice hockey players to conduct real-time dynamic confrontation or cooperative training with the trained players. The interaction monitoring module 403 is used to monitor and record interaction events with the trained player during the training process; it is also used to generate a training report based on the interaction events after the training is completed, and then send the training report to the monitoring terminal 501 through the communication module 404. The monitoring terminal 501 is used to receive and output the training report so that the user holding the monitoring terminal 501 can be aware of the training report.

[0070] Optionally, the behavior decision control module 402 is further configured to predict the future movement trajectory of the ice puck in real time based on the real-time identified ice puck status information; if the current training mode is a mobile practice mode for simulating dynamic cooperation, then a simulated behavior instruction is generated and executed, and the simulated behavior instruction specifically is: according to the real-time predicted ice puck movement trajectory, move to the ice puck landing point position before the ice puck to receive the ice puck, and perform a passing action according to the real-time position information of the trained player; if the current training mode is a defensive mode for simulating confrontation, then a simulated behavior instruction is generated and executed, and the simulated behavior instruction specifically is: according to the real-time position information of the trained player, move to the defensive area formed between the trained player and the goal, and move to the ice puck movement trajectory according to the real-time predicted ice puck movement trajectory to perform an interception action.

[0071] Optionally, the interactive monitoring module 403 is also used to provide real-time feedback on the mobile robot's physical state and training progress, so that the trained player and / or monitoring terminal can be aware of it; wherein, the mobile robot's physical state includes at least the moving state and the state of being hit by the puck; and the training process includes at least the current training mode.

[0072] Optionally, the monitoring terminal 501 is used to construct and update the individual ability model of the trained player, wherein the individual ability model includes at least the technical strengths and weaknesses of the trained player identified based on training data prior to the current time. The behavior decision control module 402 is further configured to adaptively adjust the details of the simulated behavior instructions based on the individual ability model corresponding to the trained player, so that: when the training mode is a mobile sparring mode for simulating dynamic cooperation, the generated and executed simulated behavior instructions tend to guide and amplify the technical advantages of the trained player; when the training mode is a defensive mode for simulating confrontation, the generated and executed simulated behavior instructions tend to target and train the technical weaknesses of the trained player; it is also configured to execute the simulated behavior instructions, record the movement trajectory of the mobile robot and the auxiliary training strategy during the execution of the simulated behavior instructions, generate a corresponding training template, and add the training template to the final generated training report; wherein, the auxiliary training strategy includes the passing strategy of the mobile robot when the training mode is a mobile sparring mode for simulating dynamic cooperation, or the defensive strategy of the mobile robot when the training mode is a mobile sparring mode for simulating dynamic cooperation.

[0073] Optionally, the behavior decision control module 402 is further configured to receive a simulated role-shaping instruction issued by the monitoring terminal, the simulated role-shaping instruction including feature parameters for limiting the behavioral style of the simulated target player; and generate a virtual personal ability model based on the feature parameters included in the simulated role-shaping instruction; wherein the virtual personal ability model is used to simulate the technical characteristics of the target player. The behavior decision control module 402 is further configured to determine the strategic objectives of the simulated behavior instructions based on the individual ability model corresponding to the trained player, wherein the strategic objectives are: when the training mode is a mobile sparring mode for simulating dynamic cooperation, to guide and amplify the technical advantages of the trained player; or when the training mode is a defensive mode for simulating confrontation, to target and train the technical weaknesses of the trained player; and to generate simulated behavior instructions for executing the strategic objectives according to the technical characteristics corresponding to the virtual individual ability model.

[0074] Optionally, the behavior decision control module 402 is further configured to, when generating simulated behavior instructions for executing the strategic objective according to the technical characteristics corresponding to the virtual personal ability model, dynamically adjust at least one of the following parameters required in the simulated behavior instructions: the mobile robot's moving speed, the mobile robot's reaction time, and the mobile robot's hitting force against the hockey puck, based on the physical fitness change pattern: This application also discloses an intelligent ice hockey training device, which includes a memory and a processor. The memory stores a computer program that can be loaded by the processor and executed as described above for the intelligent ice hockey training method.

[0075] This application also discloses a computer-readable storage medium that stores a computer program that can be loaded by a processor and executed as described above for the intelligent hockey training method. The computer-readable storage medium includes, for example, various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0076] It should be noted that in this paper, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0077] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit the scope of protection of the application. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on these embodiments, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

Claims

1. A smart training method for ice hockey, characterized in that, The intelligent hockey training method is executed by an intelligent hockey training system, which includes a pre-set mobile robot and a monitoring terminal. The intelligent hockey training method specifically includes: The mobile robot perceives its surrounding environment in real time and identifies the status information of the hockey puck and the trained player based on the perceived environment; wherein the status information includes at least location information. The mobile robot dynamically generates and executes simulated behavior instructions based on real-time recognized status information and predefined training modes; wherein, the simulated behavior instructions are used to simulate the actions of ice hockey players to conduct real-time dynamic confrontation or cooperative training with the trained players. During training, the mobile robot monitors and records interaction events with the trained player. After training is completed, it generates a training report based on the interaction events and sends the training report to a preset monitoring terminal so that the user holding the monitoring terminal can be aware of the training report.

2. The intelligent ice hockey training method according to claim 1, characterized in that, The method of dynamically generating and executing simulated behavior instructions based on real-time identified state information and predefined training modes includes: The mobile robot predicts the future trajectory of the ice hockey puck based on the real-time identification of the puck's status information. If the current training mode is a mobile practice mode for simulating dynamic cooperation, the mobile robot generates and executes simulated behavior instructions, which are specifically: based on the real-time predicted future trajectory of the puck, it moves to the puck's landing point before the puck to receive the puck, and executes a passing action based on the real-time position information of the trained player. If the current training mode is a defensive mode for simulating confrontation, the mobile robot generates and executes simulated behavior instructions, which specifically include: moving to the defensive area formed between the trained player and the goal based on the real-time position information of the trained player, and moving to the future trajectory of the ice puck based on the real-time predicted future trajectory of the ice puck to perform an interception action.

3. The intelligent ice hockey training method according to claim 1, characterized in that, The method further includes: The mobile robot provides real-time feedback on its own status and training progress, which is available to the trained player and / or monitoring terminal. The mobile robot's status includes at least its movement state and the state of being hit by the puck. The training progress includes at least the current training mode.

4. The intelligent ice hockey training method according to claim 2, characterized in that, The method further includes: The monitoring terminal constructs and updates the individual ability model of the trained player, wherein the individual ability model includes at least the technical strengths and weaknesses of the trained player identified based on training data prior to the current time. The method of dynamically generating and executing simulated behavior instructions based on real-time identified state information and predefined training modes includes: Based on the individual ability model corresponding to the trained player, the mobile robot adaptively adjusts the details of the simulated behavioral instructions so that: when the training mode is a mobile sparring mode for simulating dynamic cooperation, the generated and executed simulated behavioral instructions tend to guide and amplify the technical advantages of the trained player; when the training mode is a defensive mode for simulating confrontation, the generated and executed simulated behavioral instructions tend to target and train the technical weaknesses of the trained player. The mobile robot executes the simulated behavior instructions and records the mobile robot's movement trajectory and auxiliary training strategies during the execution of the simulated behavior instructions, generates a corresponding training template, and adds the training template to the final generated training report; wherein, the auxiliary training strategies include the mobile robot's passing strategy when the training mode is a mobile sparring mode for simulating dynamic cooperation, or the mobile robot's defensive strategy when the training mode is a mobile sparring mode for simulating dynamic cooperation.

5. The intelligent ice hockey training method according to claim 4, characterized in that, The method further includes: The mobile robot receives a simulated character creation instruction from the monitoring terminal, the simulated character creation instruction containing feature parameters for defining the behavioral style of the simulated target player; The mobile robot generates a virtual personal ability model based on the feature parameters contained in the simulated character creation instruction; wherein, the virtual personal ability model is used to simulate the technical characteristics of the target player; The mobile robot adaptively adjusts the details of the simulated behavioral instructions based on the individual ability model corresponding to the trained player, so that: when the training mode is a mobile sparring mode for simulating dynamic cooperation, the generated and executed simulated behavioral instructions tend to guide and amplify the technical strengths of the trained player; when the training mode is a defensive mode for simulating confrontation, the generated and executed simulated behavioral instructions tend to target and train the technical weaknesses of the trained player, including: The mobile robot determines the strategic objectives of the simulated behavioral instructions based on the individual ability model corresponding to the trained player. The strategic objectives are: when the training mode is a mobile sparring mode for simulating dynamic cooperation, to guide and amplify the technical advantages of the trained player; or when the training mode is a defensive mode for simulating confrontation, to target and train the technical weaknesses of the trained player. The mobile robot generates simulated behavioral instructions for executing the strategic objectives based on the technical characteristics corresponding to the virtual personal capability model.

6. The intelligent ice hockey training method according to claim 5, characterized in that, The simulated character creation instructions also include feature parameters for limiting the physical characteristics of the simulated target player; the generated virtual personal ability model is also used to simulate the physical change pattern of the target player; When the mobile robot generates simulated behavioral instructions for executing the strategic objectives according to the technical characteristics corresponding to the virtual personal ability model, it also dynamically adjusts at least one of the following in the simulated behavioral instructions based on the physical fitness change pattern: the mobile robot's moving speed, the mobile robot's reaction time, and the mobile robot's hitting force on the hockey puck.

7. The intelligent ice hockey training method according to claim 4, characterized in that, The method further includes: The monitoring terminal aggregates the individual ability models of different trained players. Based on the aggregated individual ability model data, it calculates the ability matching degree between different trained players and tactical roles in the preset tactical formation. Based on the ability matching degree, it generates and outputs a player pairing recommendation list, which includes recommended player combinations and corresponding tactical formations. The player combination includes players assigned to each tactical role in the corresponding tactical formation; and the same player combination satisfies the following: in the corresponding tactical formation, the technical weaknesses and technical strengths of the players can complement each other; and / or the technical strengths of the players can be superimposed to enhance the tactical effect.

8. An intelligent ice hockey training system, characterized in that, The ice hockey intelligent training system includes a mobile robot and a monitoring terminal (501); the mobile robot includes an environmental perception module (401), a behavior decision control module (402), an interactive monitoring module (403), and a communication module (404). The environmental perception module (401) is used to perceive the surrounding environment in real time and identify the status information of the hockey puck and the trained player based on the perceived surrounding environment; wherein the status information includes at least location information. The behavior decision control module (402) is used to dynamically generate and execute simulated behavior instructions based on real-time identified status information and predefined training modes; wherein, the simulated behavior instructions are used to simulate the actions of ice hockey players to conduct real-time dynamic confrontation or cooperative training with the trained players. The interaction monitoring module (403) is used to monitor and record interaction events with the trained player during the training process; it is also used to generate a training report based on the interaction events after the training is completed, and then send the training report to the monitoring terminal (501) through the communication module (404). The monitoring terminal (501) is used to receive and output the training report so that the user holding the monitoring terminal (501) can know the training report.

9. An intelligent ice hockey training device, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer program is stored that can be loaded by a processor and executed as described in any one of claims 1 to 7.

Images