Ball state estimation method, multi-source data fusion hitting point estimation method and system

Abstract

This invention relates to the field of motion parameter measurement and calculation technology, and particularly to a method and system for calculating the state of a ball, a method for calculating the impact point through multi-source data fusion, and the following: acquiring inertial data through an inertial measurement unit inside the moving ball; detecting impact events based on the inertial data, extracting the linear acceleration vector corresponding to the geometric center of the moving ball, and the angular acceleration vector corresponding to the rotation of the moving ball around its own geometric center; calculating the offset vector of the impact point relative to the geometric center of the moving ball based on a rigid body dynamics model and the translational-rotational coupling relationship between the linear acceleration vector and the angular acceleration vector; and determining the rotation type and rotation axis direction of the moving ball after the impact event is triggered based on the spatial relative orientation relationship between the offset vector and the linear acceleration vector. This invention can complete the core parameter calculation through a built-in inertial measurement unit, has low cost, strong adaptability, and can be flexibly applied in various general sports scenarios, both indoors and outdoors.

Description

Technical Field

[0001] This invention relates to the field of motion parameter measurement and calculation technology, and in particular to a method for calculating the state of a ball, a method and system for calculating the hitting point by multi-source data fusion. Background Technology

[0002] In the fields of ball sports training, competition analysis, and intelligent sports equipment development, the position of the hitting point, the type of ball rotation, and the direction of the rotation axis are the core measurement parameters for evaluating the quality of the hit and optimizing the sports posture. Accurately measuring and obtaining these parameters is of great significance for improving training results and realizing sports data analysis.

[0003] Currently, existing ball sports parameter acquisition technologies are mainly divided into the following two categories, but both have insurmountable technical defects, which lead to many limitations in their practical applications and make it impossible to meet the actual needs of low-cost, high-precision, and externally independent sports parameter measurement.

[0004] The first type of solution is the racket-end sensing and detection solution. This solution uses an IMU sensor built into the racket handle to collect inertial data and calculate the position of the hitting point and the sweet spot distribution. The core drawback of this solution is that users need to purchase a dedicated smart racket, which cannot be used with existing regular rackets. Dedicated smart rackets are expensive, significantly increasing the cost and adaptation barrier for users. At the same time, the sensor array needs to be precisely installed in the handle, making it impossible to use immediately after playing. Furthermore, the data collected by the racket-end sensor is easily affected by racket face vibration and hitting angle deviation, resulting in low accuracy in calculating the hitting point. More importantly, this solution can only provide the position information of the hitting point and cannot directly obtain the ball's own rotation parameters. It cannot achieve coordinated and accurate measurement of the hitting point and rotation state, making it difficult to meet the needs of comprehensive acquisition of core motion parameters in actual training.

[0005] The second type of solution is the external visual acquisition solution. This solution deploys external devices such as high-speed cameras and visual tracking equipment to capture the ball-hitting process in real time, and uses image recognition and motion tracking algorithms to extract the ball's trajectory, the position of the hitting point, and its spin state. The key drawbacks of this solution are that the required high-speed cameras and other external equipment are expensive, easily affected by occlusion, and cannot be flexibly applied in various common indoor and outdoor sports scenarios. Furthermore, this solution requires post-processing analysis of a large amount of image data, cannot achieve real-time output of hitting parameters, and cannot meet the needs of real-time training guidance. In addition, it is completely dependent on external equipment, cannot achieve independent measurement, and has extremely poor flexibility.

[0006] In summary, existing ball sports parameter estimation techniques all have their own limitations. How to establish an independent measurement mode for ball sports parameters that does not rely on external equipment has become a core technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0007] This invention provides a method for calculating the state of a ball, a method and system for calculating the hitting point by fusing multi-source data, which can effectively solve the problems in the background art.

[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows: Methods for estimating the state of a ball include: The inertial data of the moving sphere during its motion is collected. The collection is achieved by an inertial measurement unit built into the moving sphere. The inertial data includes triaxial acceleration data and triaxial angular velocity data. The ball-hitting event is detected based on the inertial data, and the linear acceleration vector corresponding to the geometric center of the moving ball and the angular acceleration vector corresponding to the rotation of the moving ball around its own geometric center are extracted at the instant the ball-hitting event is triggered. Based on the rigid body dynamics model, and according to the translational-rotational coupling relationship between the linear acceleration vector and the angular acceleration vector, the offset vector of the striking point relative to the geometric center of the moving sphere is calculated; the offset vector includes the offset distance and the direction of action in the sphere coordinate system, which is a three-dimensional Cartesian coordinate system fixed to the geometric center of the moving sphere; Based on the spatial relative orientation relationship between the offset vector and the linear acceleration vector, the rotation type and rotation axis direction of the moving ball after the ball-hitting event is determined.

[0009] Furthermore, the detection of ball-hitting events includes: Based on the triaxial acceleration data, the triaxial acceleration modulus is continuously calculated. When the triaxial acceleration modulus exceeds a preset impact threshold, it is determined to be a suspected ball-hitting event. Based on the angular acceleration vector within a preset time window before and after the ball is struck, the validity of the suspected ball-striking event is verified to determine the valid ball-striking event.

[0010] Furthermore, the rigid body dynamics model includes: Based on the law of rigid body translation, the resultant force vector of the hitting ball is obtained from the linear acceleration vector; Based on the rigid body rotation law, the resultant external torque vector is obtained from the angular acceleration vector and the moment of inertia of the sphere. And based on the vector cross product of torque and force The offset vector is obtained by solving, where Let be the resultant external torque vector. Let be the resultant force vector of the hitting ball. The offset vector is denoted as .

[0011] Furthermore, the offset vector is solved by inverse method based on the vector cross product relationship. The calculation formula is: ; Where I is the moment of inertia of the sphere. It is the angular acceleration vector. Let m be the linear acceleration vector, and m be the mass of the sphere.

[0012] Furthermore, determining the rotation type of the moving sphere includes: Based on the cross product of the offset vector and the linear acceleration vector, it is determined whether the hitting event generates spin; When the cross product result is zero, it is determined that the striking force passes through the center of the ball and the ball has no rotation. When the cross product result is not zero, it is determined that the sphere has rotated, and the direction of the rotation axis is determined based on the vector cross product result; When the rotation axis is perpendicular to the initial flight direction and parallel to the ground, it is determined to be either an upward or downward rotation; When the rotation axis is perpendicular to the ground, it is determined to be a side rotation; When the rotation axis has both horizontal and vertical components, it is determined to be a mixed rotation.

[0013] Further, determining the direction of the rotation axis includes: Based on the cross product of the offset vector and the linear acceleration vector, the spatial orientation of the rotation axis in the spherical coordinate system is determined. Based on the angles between the rotation axis and the reference axes of the spherical coordinate system, the tilt angle and azimuth angle of the rotation axis are output.

[0014] Furthermore, it also includes: The rotational speed of the moving sphere is calculated based on the integral result of the angular acceleration vector within a preset time window. The rotation type, rotation axis direction, and rotation rate are combined to form a complete rotation state parameter output.

[0015] Multi-source data fusion methods for calculating the point of impact include: The inertial data of the moving sphere during its motion is collected. The collection is achieved by an inertial measurement unit built into the moving sphere. The inertial data includes triaxial acceleration data and triaxial angular velocity data. The ball-hitting event is detected based on the inertial data, and the linear acceleration vector corresponding to the geometric center of the moving ball and the angular acceleration vector corresponding to the rotation of the moving ball around its own geometric center are extracted at the instant the ball-hitting event is triggered. Based on the rigid body dynamics model, and according to the translational-rotational coupling relationship between the linear acceleration vector and the angular acceleration vector, the offset vector of the striking point relative to the geometric center of the moving sphere is calculated; the offset vector includes the offset distance and the direction of action in the sphere coordinate system, which is a three-dimensional Cartesian coordinate system fixed to the geometric center of the moving sphere; Receive racket face posture data, which includes the racket face normal vector and the racket face coordinate system; The offset vector is mapped from the spherical coordinate system to the racket face coordinate system through coordinate transformation; The coordinates of the hitting point on the racket face are calculated and output based on the mapping results.

[0016] Furthermore, the coordinate transformation includes: Synchronize the timestamp of the ball-hitting event with the timestamp of the racket face posture data to establish the correspondence between the ball coordinate system and the racket face coordinate system at the moment of hitting the ball; The projection calculation of the offset vector is corrected based on the changes in racket face posture within a preset time window before and after the shot.

[0017] A multi-source data fusion system for calculating the point of impact includes: An inertial data acquisition unit, built into the moving sphere, is used to collect inertial data during the movement of the sphere. The inertial data includes triaxial acceleration data and triaxial angular velocity data. The ball-hitting detection and vector extraction unit detects ball-hitting events based on the inertial data and extracts the linear acceleration vector corresponding to the geometric center of the moving ball and the angular acceleration vector corresponding to the rotation of the moving ball around its own geometric center at the instant the ball-hitting event is triggered. The offset vector calculation unit, based on the rigid body dynamics model, calculates the offset vector of the hitting point relative to the geometric center of the moving ball according to the translational-rotational coupling relationship between the linear acceleration vector and the angular acceleration vector; the offset vector includes the offset distance and the direction of action in the spherical coordinate system, which is a three-dimensional rectangular coordinate system fixed to the geometric center of the moving ball; A racket face posture receiving unit receives racket face posture data, which includes the racket face normal vector and the racket face coordinate system. The coordinate transformation unit maps the offset vector from the spherical coordinate system to the racket face coordinate system through coordinate transformation. The hitting point output unit calculates and outputs the position coordinates of the hitting point on the racket face based on the mapping result.

[0018] The technical solution of this invention can achieve the following technical effects: The inertial data of this invention comes solely from the inertial measurement unit built into the moving ball, without relying on racket-end sensing devices, external visual acquisition devices, or other external measurement devices. This allows users to directly use existing conventional balls and calculate core parameters solely through the built-in inertial measurement unit. It is low-cost, highly adaptable, and can be flexibly applied in various general indoor and outdoor sports scenarios.

[0019] During implementation, based on the rigid body dynamics model, the translational-rotational coupling relationship between the linear acceleration vector and the angular acceleration vector is used to calculate the impact point offset vector, thus clarifying the reference benchmark for the offset vector. At the same time, the spatial relative orientation relationship between the offset vector and the linear acceleration vector is used to determine the rotation type and rotation axis direction. This method is suitable for high-impact scenarios at the moment of impact and can improve the measurement and calculation accuracy of the impact point offset vector, rotation type, and rotation axis direction, ensuring stable and reliable parameter output.

[0020] The method of this invention organically combines the calculation of the impact point offset vector with the determination of the rotation state. By using the same set of inertial data and the same set of dynamic models, the calculation of the two core parameters is completed simultaneously, realizing a complete technical closed loop of data acquisition, event detection, parameter calculation and state determination, and providing a reliable solution for the comprehensive and accurate acquisition of ball sports parameters. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 A flowchart of the method for calculating the state of a ball; Figure 2 A flowchart of a method for calculating the impact point using multi-source data fusion; Figure 3 This is a framework diagram of a multi-source data fusion system for calculating the point of impact. Detailed Implementation

[0023] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0024] Example 1 This embodiment uses a tennis ball as an application scenario. The inertial measurement unit built into the moving ball employs a six-axis IMU sensor. The sensor used in this embodiment integrates a three-axis accelerometer and a three-axis gyroscope, enabling real-time acquisition of three-axis acceleration and angular velocity data during the ball's motion. During implementation, the sampling frequency can be adaptively adjusted according to actual needs to ensure accurate capture of instantaneous data at the moment of impact, meeting the data acquisition requirements under high-impact scenarios.

[0025] like Figure 1 As shown, the method for estimating the state of a ball includes: S1: Inertial data of the moving sphere during its motion is collected via a six-axis IMU sensor embedded inside the sphere. The collected inertial data specifically includes three-axis acceleration data (Ax, Ay, Az) and three-axis angular velocity data (ωx, ωy, ωz). The collected data can be transmitted in real time to the corresponding data processing unit for subsequent data preprocessing. In some embodiments of this invention, the preprocessing includes noise reduction and filtering to eliminate errors caused by sensor noise and external interference, ensuring data accuracy. S2: Detect ball-hitting events based on inertial data, and extract the linear acceleration vector corresponding to the geometric center of the moving ball and the angular acceleration vector corresponding to the rotation of the moving ball around its own geometric center at the instant the ball-hitting event is triggered; During implementation, the occurrence of ball-hitting events can be detected by real-time analysis of the preprocessed inertial data. The linear acceleration vector a is composed of triaxial acceleration data (Ax, Ay, Az), and the angular acceleration vector β is obtained by differentiating the triaxial angular velocity data (ωx, ωy, ωz) with respect to time. S3: Based on the rigid body dynamics model, the offset vector of the hitting point relative to the geometric center of the moving ball is calculated according to the translational-rotational coupling relationship between the linear acceleration vector and the angular acceleration vector. The offset vector includes the offset distance and the direction of action in the spherical coordinate system, which is a three-dimensional rectangular coordinate system fixed to the geometric center of the moving ball. As a specific implementation of step S3, the origin can be taken as the geometric center of the sphere, the x-axis is along the initial flight direction of the sphere, the y-axis is perpendicular to the x-axis and lies in the horizontal plane, and the z-axis is perpendicular to the horizontal plane and upwards, forming a right-handed coordinate system; S4: Based on the spatial relative orientation relationship between the offset vector and the linear acceleration vector, determine the rotation type and rotation axis direction of the moving ball after the ball is hit.

[0026] The inertial data of this invention comes solely from the inertial measurement unit built into the moving ball, without relying on racket-end sensing devices, external visual acquisition devices, or other external measurement devices. This allows users to directly use existing conventional balls and calculate core parameters solely through the built-in inertial measurement unit. It is low-cost, highly adaptable, and can be flexibly applied in various general indoor and outdoor sports scenarios.

[0027] During implementation, based on the rigid body dynamics model, the translational-rotational coupling relationship between the linear acceleration vector and the angular acceleration vector is used to calculate the impact point offset vector, thus clarifying the reference benchmark for the offset vector. At the same time, the spatial relative orientation relationship between the offset vector and the linear acceleration vector is used to determine the rotation type and rotation axis direction. This method is suitable for high-impact scenarios at the moment of impact and can improve the measurement and calculation accuracy of the impact point offset vector, rotation type, and rotation axis direction, ensuring stable and reliable parameter output.

[0028] The method of this invention organically combines the calculation of the impact point offset vector with the determination of the rotation state. By using the same set of inertial data and the same set of dynamic models, the calculation of the two core parameters is completed simultaneously, realizing a complete technical closed loop of data acquisition, event detection, parameter calculation and state determination, and providing a reliable solution for the comprehensive and accurate acquisition of ball sports parameters.

[0029] As a preferred embodiment of the above embodiments, detecting a ball-hitting event includes: A1: The triaxial acceleration modulus is continuously calculated based on triaxial acceleration data. When the triaxial acceleration modulus exceeds the preset impact threshold, it is determined to be a suspected ball-hitting event. During step A1, the triaxial acceleration modulus a is calculated using the following formula; When the triaxial acceleration modulus 'a' exceeds the preset impact threshold, a suspected ball-hitting event is determined, and the process enters the validity verification stage. The preset impact threshold is an empirical threshold pre-calibrated based on the ball's material, mass, and motion scenario, used to distinguish between normal motion disturbances and the instantaneous strong impact generated by a real ball-hitting event. A2: Based on the angular acceleration vector within a preset time window before and after the ball is hit, the validity of suspected ball-hitting events is verified to determine valid ball-hitting events.

[0030] As one implementation of step A2 above, the suspected ball-hitting event is verified as a valid ball-hitting event based on the ratio of the peak linear acceleration to the integral angular acceleration within the same time window.

[0031] In this embodiment, based on the triaxial angular velocity data ωx, ωy, and ωz collected at the same time, the triaxial angular accelerations βx, βy, and βz are obtained by differentiating with respect to time, and then the angular acceleration vector is obtained. The diagonal acceleration vector within a preset time window [t1, t2] before and after the ball-hitting event. Perform time integration to obtain the integral value of angular acceleration; at the same time, extract the peak value a1 of the linear acceleration vector within the same time window; calculate the ratio of the peak value of linear acceleration to the integral value of angular acceleration. When the ratio is within the preset effective ratio range, the suspected ball hitting event is determined to be a valid ball hitting event; otherwise, it is determined to be an interference trigger and is rejected.

[0032] This method uses the ratio of the peak linear acceleration to the integral angular acceleration for validity verification. It can fully combine the inherent correlation between the impact intensity and rotation intensity generated by the hitting action. It can identify true and false hits simply by the proportional relationship of the motion signals. It has low computational overhead, strong real-time performance, and a wider range of applications. It can maintain a stable recognition effect in different types and sizes of ball sports scenarios, and has strong versatility and robustness.

[0033] As another implementation of step A2 above, the change in angular momentum obtained by integrating the angular acceleration vector with the moment of inertia of the ball is used to verify that the suspected hitting event is a valid hitting event.

[0034] This method uses triaxial angular velocity data ωx, ωy, and ωz collected at the same time to obtain triaxial angular accelerations βx, βy, and βz by taking the derivative with respect to time, and then obtains the angular acceleration vector. Combining the ball's own moment of inertia I, the angular acceleration vector within the preset time window [t1, t2] before and after the ball-hitting event is determined. By integrating over time, we obtain the change in angular momentum ΔL before and after the moment of impact: When the change in angular momentum ΔL exceeds the preset angular momentum threshold, the suspected ball-hitting event is determined to be a valid ball-hitting event; otherwise, it is determined to be an interference trigger and is rejected.

[0035] In this approach, the change in angular momentum is obtained by combining the angular acceleration vector with the integral of the ball's rotational inertia for threshold judgment. This fully utilizes the ball's own physical properties to construct a dynamic detection model, which is closer to the physical mechanism of a real shot. It can accurately distinguish between a real shot and non-shot interference such as vibration, drop, and collision. It has stronger anti-interference ability and higher detection accuracy, providing a more reliable triggering benchmark for subsequent shot point offset vector calculation and rotation state recognition, significantly improving the accuracy and stability of the overall solution.

[0036] As a preferred embodiment of the above, the rigid body dynamics model includes: The resultant force vector of the ball is obtained from the linear acceleration vector based on the law of rigid body translation. Based on the rigid body rotation law, the resultant external torque vector is obtained from the angular acceleration vector and the moment of inertia of the sphere. And based on the vector cross product of torque and force The offset vector is obtained by solving, where The resultant external torque vector, The resultant force vector of the hitting ball, This is the offset vector.

[0037] In this embodiment, a rigid body dynamics model is used to accurately calculate the impact point offset vector. The construction and calculation process of the rigid body dynamics model is as follows: Specifically, according to the law of rigid body translation, the vector of the resultant force acting on the ball at the instant of impact is... The linear acceleration vector relative to the geometric center of the sphere The following relationship must be satisfied: ;-(1) Where m is the mass of the moving sphere, which is a known constant; This is the linear acceleration vector extracted at the instant the ball is struck. Using the formula above, the linear acceleration vector can be... The resultant force vector of the impact force applied to the ball at the moment of impact is directly calculated. .

[0038] Furthermore, according to the law of rigid body rotation, the net external torque vector acting on the ball at the moment of impact is... Angular acceleration vector of the sphere The moment of inertia I satisfies the following relationship: =I ;-(2) Where I is the moment of inertia of the sphere about its geometric center, which is a known constant determined in advance based on the material and radius of the sphere; This is the angular acceleration vector extracted at the instant a valid hitting event is triggered. Using the formula above, the angular acceleration vector can be... The net external torque vector at the moment of impact is obtained by solving for the moment of inertia I of the sphere. .

[0039] In rigid body dynamics, the net external moment vector, the net impact force vector, and the impact point offset vector satisfy the following vector cross product constraint relationship: = × ;-(3) in: The resultant external torque vector, The resultant force vector of the hitting ball, The offset vector of the striking point relative to the geometric center of the ball is to be calculated.

[0040] Based on the aforementioned vector cross product equation, and combining the three-dimensional components of each vector in the spherical coordinate system, the offset vector is obtained through inverse vector algebra operations. The offset vector contains the offset distance of the striking point relative to the geometric center of the ball and the direction of action in the ball coordinate system, providing precise positional parameters for determining the subsequent spin type and the direction of the spin axis.

[0041] This embodiment constructs a complete rigid body dynamics model using the laws of translation, rotation, and torque cross product, establishing the calculation of the impact point offset vector on strict physical constraints, significantly improving the accuracy and stability of parameter calculation in high-impact scenarios.

[0042] As a preferred embodiment of the above, the offset vector is solved by inverse solution based on the vector cross product relationship. The calculation formula is: ; Where I is the moment of inertia of the sphere. It is the angular acceleration vector. Let m be the linear acceleration vector, and m be the mass of the sphere.

[0043] Specifically, by combining formulas (1)-(3), we can obtain: I· = - (4) At the moment of impact, the effective point of action is located near the center of the ball, and the offset vector... With linear acceleration vector The approximate perpendicular constraint is satisfied. Based on vector algebra operations and using the unified right-handed coordinate system convention of this embodiment, the formula for calculating the offset vector is: 。 - (5) The offset vector in the above formula The direction is determined by the right-hand rule, representing the spatial orientation of the striking point relative to the geometric center of the ball.

[0044] As a preferred embodiment of the above embodiments, determining the rotation type of the moving sphere includes: Based on the cross product of the offset vector and the linear acceleration vector, it is determined whether the hitting event produces spin; When the cross product is zero, it is determined that the force of the shot passes through the center of the ball and the ball has no rotation. When the cross product result is not zero, it is determined that the sphere has rotated, and the direction of the rotation axis is determined based on the vector cross product result; When the axis of rotation is perpendicular to the initial flight direction and parallel to the ground, it is determined to be either an upspin or a downspin; When the axis of rotation is perpendicular to the ground, it is determined to be a side rotation; When the rotation axis has both horizontal and vertical components, it is determined to be a mixed rotation.

[0045] Rotation types are classified by the spatial angle relationship between the offset vector and the linear acceleration vector. The classification of rotation states can be achieved without additional sensors. The judgment logic is simple, the amount of computation is small, and the real-time performance is strong.

[0046] As a preferred embodiment of the above, determining the direction of the rotation axis includes: Based on the cross product of the offset vector and the linear acceleration vector, the spatial orientation of the rotation axis in the spherical coordinate system is determined. Based on the angles between the rotation axis and the reference axes of the spherical coordinate system, output the tilt angle and azimuth angle of the rotation axis.

[0047] Specifically, by performing a cross product operation on the offset vector and the linear acceleration vector, the torque direction vector is obtained, which is consistent with the direction of the sphere's rotation axis. After normalizing the rotation axis direction vector obtained by the cross product, it is mapped to the sphere's own coordinate system. The angles between the rotation axis direction and each reference axis of the sphere's coordinate system are calculated. Based on the magnitude of the angles, the tilt angle and azimuth angle of the rotation axis are further determined, thus fully characterizing the spatial attitude of the sphere's rotation axis.

[0048] The rotation axis direction can be determined in the above way without the need for external vision or positioning equipment. It can be achieved solely by inertial measurement data and dynamic calculation results. The calculation is stable and reliable and is suitable for real-time rotation axis calculation under high-speed motion conditions.

[0049] Methods for estimating the state of a ball also include: The rotational speed of the moving sphere is calculated based on the integral of the angular acceleration vector within a preset time window. The rotation type, rotation axis direction, and rotation rate are combined to form a complete rotation state parameter output.

[0050] During implementation, taking the moment of impact as the benchmark, and considering that the strong impact at the moment of impact can easily cause distortion of the original angular velocity sampling data, the angular acceleration vector is integrated over time within a set short-time integration window to obtain the change in the ball's rotational angular velocity per unit time. Based on this change in angular velocity, the ball's rotation rate is further calculated. The angular acceleration vector used in this preferred scheme has often undergone noise reduction and interference removal. By using angular acceleration integration, the instantaneous distortion error of angular velocity can be avoided. Relying on reliable angular acceleration data, the rotation rate at the moment of impact is accurately calculated, which is suitable for the data measurement needs of high-impact impact scenarios. During implementation, in the stable, impact-free flight phase, the rotation angle can be directly solved by integrating the angular velocity collected by the gyroscope.

[0051] The aforementioned rotation type, rotation axis direction, and rotation rate are combined to form complete rotation state parameters that include rotation category, rotation axis spatial orientation, and rotation speed, and then output to the outside world.

[0052] Through the above steps, the complete rotation state can be detected and output using only the inertial measurement unit built into the sphere, without the need for external auxiliary equipment. The detection process is not affected by environmental factors such as the venue, lighting, or obstructions. It can stably, continuously, and accurately output the sphere's rotation state in real-world motion scenarios, meeting the needs of real-time analysis and data feedback for ball sports.

[0053] Example 2 like Figure 2 As shown, the multi-source data fusion method for calculating the impact point includes: The system collects inertial data during the motion of the sphere. The data is collected through an inertial measurement unit built into the sphere. The inertial data includes triaxial acceleration data and triaxial angular velocity data. The ball-hitting event is detected based on inertial data, and the linear acceleration vector corresponding to the geometric center of the moving ball and the angular acceleration vector corresponding to the rotation of the moving ball around its own geometric center are extracted at the moment the ball-hitting event is triggered. Based on the rigid body dynamics model, and according to the translational-rotational coupling relationship between the linear acceleration vector and the angular acceleration vector, the offset vector of the striking point relative to the geometric center of the moving ball is calculated. The offset vector includes the offset distance and the direction of action in the spherical coordinate system, which is a three-dimensional rectangular coordinate system fixed to the geometric center of the moving ball. Receive racket face attitude data, which includes the racket face normal vector and the racket face coordinate system; The offset vector is mapped from the spherical coordinate system to the racket face coordinate system through coordinate transformation; The coordinates of the hitting point on the racket face are calculated and output based on the mapping results. At the same time, the relative position of the hitting point with respect to the standard sweet spot of the racket can be obtained.

[0054] Using the above method, without relying on dedicated smart rackets or high-speed camera equipment, the hitting point can be mapped from the ball coordinate system to the racket face coordinate system by data fusion of smart ball and racket face posture data. This directly obtains the true position of the hitting point on the racket, achieving low-cost, real-time, and highly robust hitting positioning. It is applicable to various indoor and outdoor sports scenarios and facilitates subsequent judgment of the degree of deviation of the hitting point from the racket's sweet spot based on the relative position of the hitting point to the racket's standard sweet spot, thus completing the hitting action quality analysis.

[0055] In this embodiment, racket face posture data is obtained through at least one of the following methods: a wearable device worn on the wrist of the racket holder to calculate the racket face angle based on the wrist posture; an inertial sensor module installed on the racket handle, which, as a preferred method, is independent of the racket frame and detachable; and a visual sensor of a mobile terminal to track racket face movement through image recognition.

[0056] During implementation, the sphere, wearable device, inertial sensor module, or vision sensor operate independently and do not establish direct communication. Each device interacts with the data processing terminal separately via wireless communication protocols. In some embodiments of the invention, the data processing terminal uses its own system clock as a unified time reference to map the locally acquired timestamps uploaded by the sphere, wearable device, inertial sensor module, or vision sensor to the same absolute time axis to complete time regularization.

[0057] As a preferred embodiment of the above, the coordinate transformation includes: Synchronize the timestamp of the hitting event with the timestamp of the racket face posture data to establish the correspondence between the ball coordinate system and the racket face coordinate system at the moment of hitting; The projection calculation of the offset vector is corrected based on the changes in racket face posture within a preset time window before and after the shot.

[0058] In this embodiment, the coordinate transformation specifically includes two steps: time synchronization and projection correction. First, the timestamp of the ball-hitting event and the timestamp of the racket face posture data are synchronized and aligned to eliminate time offset errors between different sensors and devices, establishing a unique and accurate spatial correspondence between the ball coordinate system and the racket face coordinate system at the moment of impact, ensuring the consistency of the coordinate mapping reference. Simultaneously, based on the real-time changes in racket face posture within a preset time window before and after impact, the projection calculation of the offset vector onto the racket face coordinate system is dynamically corrected to compensate for calculation deviations caused by racket face wobbling and sudden posture changes at the moment of impact, improving the accuracy and stability of the position coordinates of the hitting point on the racket's sweet spot.

[0059] In some embodiments of the present invention, a global geodetic coordinate system can be selected when performing coordinate transformation. Specifically, this includes: firstly, using the inertial measurement unit built into the sphere in conjunction with an attitude calculation algorithm, calculating the absolute attitude of the moving sphere relative to the global geodetic coordinate system, such as obtaining the heading angle, pitch angle, and roll angle, to determine the global spatial orientation of the sphere coordinate system; combining the synchronously acquired racket face attitude data, similarly determining the global spatial orientation of the racket face coordinate system; and constructing the spatial relationships between the sphere coordinate system and the racket face coordinate system relative to the geodetic coordinate system, respectively, and deriving the transformation relationship from the sphere coordinate system to the racket face coordinate system.

[0060] In other embodiments of the present invention, when performing coordinate transformation, the ball coordinate system at the moment of impact can be selected as a unified transformation reference. This method does not require calculating the absolute position of the ball on the ground, and the reference reference can be established solely based on the ball's own motion state at the moment of impact. In this method, the ball, wearable device, inertial sensor module, or vision sensor are all independent acquisition units. Due to the inherent crystal oscillator clock drift, there is still a time alignment requirement between the ball posture data and the racket face posture data at the moment of impact. This requirement can also be uniformly processed by using mature and universal time registration technology at the data processing end.

[0061] By combining time synchronization and attitude correction in coordinate transformation, the system error caused by asynchronous multi-source data and instantaneous attitude changes can be effectively reduced, significantly improving the accuracy of the hitting point positioning and ensuring reliable positioning results even in high-speed, dynamic real-world hitting scenarios.

[0062] During implementation, the following two modification methods are provided as examples: The first method involves selecting multiple frames of racket face posture data within a preset time window before and after the shot, centered on the moment of impact. The racket face normal vector and the racket face coordinate system of the multiple frames are then subjected to a moving average filter to obtain a smoothed racket face reference posture. Using the smoothed racket face reference posture, the offset vector is projected to eliminate the instantaneous noise caused by racket shaking and sudden changes in posture at the moment of impact, making the projection result closer to the actual shot posture.

[0063] This method is a sliding average correction of the attitude. Centered on the moment of impact, it selects multiple frames of racket face attitude data within a preset time window before and after the impact and performs sliding average filtering to obtain a smooth and stable racket face reference attitude. Based on this smooth attitude, it completes the offset vector projection to suppress instantaneous attitude noise.

[0064] The second method involves calculating the angular velocity and angular acceleration of the racket face within a preset time window before and after the shot, obtaining the rate of change of the racket face's posture at the moment of impact; dynamically compensating the racket face normal vector based on the rate of change of posture, correcting the posture deviation at the moment of impact, and then projecting the offset vector onto the compensated racket face coordinate system to improve the accuracy of the hitting point position.

[0065] This method is instantaneous attitude change rate compensation. By calculating the angular velocity and angular acceleration of the racket face at the moment of impact, the racket face normal vector is dynamically compensated to eliminate the projection error caused by the sudden change in attitude at the moment of impact, and further improve the positioning accuracy of the hitting point.

[0066] In this embodiment, the multi-source data fusion method for calculating the hitting point further includes: determining the hitting eccentricity index based on the modulus of the offset vector; and outputting the evaluation results of center hitting, slightly eccentric hitting, or heavily eccentric hitting based on the comparison result of the eccentricity index and a preset threshold. This allows for the quantitative evaluation and grading feedback of hitting quality while simultaneously locating the hitting point. This method eliminates the need for additional detection devices, utilizing only the calculated offset vector for evaluation, resulting in high computational efficiency and low resource consumption; it also provides a direct reflection of the deviation between the hitting point and the racket's sweet spot.

[0067] Example 3 like Figure 3 As shown, the multi-source data fusion impact point calculation system includes: An inertial data acquisition unit, built into the moving sphere, is used to collect inertial data during the movement of the sphere. The inertial data includes triaxial acceleration data and triaxial angular velocity data. The ball-hitting detection and vector extraction unit detects ball-hitting events based on inertial data and extracts the linear acceleration vector corresponding to the geometric center of the moving ball and the angular acceleration vector corresponding to the rotation of the moving ball around its own geometric center at the instant the ball-hitting event is triggered. The offset vector calculation unit, based on the rigid body dynamics model, calculates the offset vector of the hitting point relative to the geometric center of the moving ball according to the translational-rotational coupling relationship between the linear acceleration vector and the angular acceleration vector. The offset vector includes the offset distance and the direction of action in the spherical coordinate system, which is a three-dimensional rectangular coordinate system fixed to the geometric center of the moving ball. The racket face attitude receiving unit receives racket face attitude data, which includes the racket face normal vector and the racket face coordinate system. The coordinate transformation unit maps the offset vector from the spherical coordinate system to the racket face coordinate system through coordinate transformation. The hitting point output unit calculates and outputs the position coordinates of the hitting point on the racket face based on the mapping results.

[0068] The technical effects achieved in this embodiment are the same as those in Embodiment 2 above, and will not be repeated here.

[0069] Those skilled in the art should understand that this invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to this invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A method for estimating the state of a sphere, characterized in that... ,include: The inertial data of the moving sphere during its motion is collected. The collection is achieved by an inertial measurement unit built into the moving sphere. The inertial data includes triaxial acceleration data and triaxial angular velocity data. The ball-hitting event is detected based on the inertial data, and the linear acceleration vector corresponding to the geometric center of the moving ball and the angular acceleration vector corresponding to the rotation of the moving ball around its own geometric center are extracted at the instant the ball-hitting event is triggered. Based on the rigid body dynamics model, and according to the translational-rotational coupling relationship between the linear acceleration vector and the angular acceleration vector, the offset vector of the striking point relative to the geometric center of the moving sphere is calculated; the offset vector includes the offset distance and the direction of action in the sphere coordinate system, which is a three-dimensional Cartesian coordinate system fixed to the geometric center of the moving sphere; Based on the spatial relative orientation relationship between the offset vector and the linear acceleration vector, the rotation type and rotation axis direction of the moving ball after the ball-hitting event is determined.

2. The method for estimating the state of a ball according to claim 1, characterized in that, The detected ball-hitting events include: Based on the triaxial acceleration data, the triaxial acceleration modulus is continuously calculated. When the triaxial acceleration modulus exceeds a preset impact threshold, it is determined to be a suspected ball-hitting event. Based on the angular acceleration vector within a preset time window before and after the ball is struck, the validity of the suspected ball-striking event is verified to determine the valid ball-striking event.

3. The method for estimating the state of a ball according to claim 1, characterized in that... The rigid body dynamics model includes: Based on the law of rigid body translation, the resultant force vector of the hitting ball is obtained from the linear acceleration vector; Based on the rigid body rotation law, the resultant external torque vector is obtained from the angular acceleration vector and the moment of inertia of the sphere. And based on the vector cross product of torque and force The offset vector is obtained by solving, where Let be the resultant external torque vector. Let be the resultant force vector of the hitting ball. The offset vector is denoted as .

4. The method for estimating the state of a ball according to claim 3, characterized in that, Based on the vector cross product relationship, the offset vector is inversely solved. The calculation formula is: ； Where I is the moment of inertia of the sphere. It is the angular acceleration vector. Let m be the linear acceleration vector, and m be the mass of the sphere.

5. The method for estimating the state of a ball according to claim 1, characterized in that, The determination of the rotation type of the moving sphere includes: Based on the cross product of the offset vector and the linear acceleration vector, it is determined whether the hitting event generates spin; When the cross product result is zero, it is determined that the striking force passes through the center of the ball and the ball has no rotation. When the cross product result is not zero, it is determined that the sphere has rotated, and the direction of the rotation axis is determined based on the vector cross product result; When the rotation axis is perpendicular to the initial flight direction and parallel to the ground, it is determined to be either an upward or downward rotation; When the rotation axis is perpendicular to the ground, it is determined to be a side rotation; When the rotation axis has both horizontal and vertical components, it is determined to be a mixed rotation.

6. The method for estimating the state of a ball according to claim 5, characterized in that, Determining the direction of the rotation axis includes: Based on the cross product of the offset vector and the linear acceleration vector, the spatial orientation of the rotation axis in the spherical coordinate system is determined. Based on the angles between the rotation axis and the reference axes of the spherical coordinate system, the tilt angle and azimuth angle of the rotation axis are output.

7. The method for estimating the state of a ball according to claim 1, characterized in that, Also includes: The rotational speed of the moving sphere is calculated based on the integral result of the angular acceleration vector within a preset time window. The rotation type, rotation axis direction, and rotation rate are combined to form a complete rotation state parameter output.

8. A method for calculating the hitting point through multi-source data fusion, characterized in that, include: The inertial data of the moving sphere during its motion is collected. The collection is achieved by an inertial measurement unit built into the moving sphere. The inertial data includes triaxial acceleration data and triaxial angular velocity data. The ball-hitting event is detected based on the inertial data, and the linear acceleration vector corresponding to the geometric center of the moving ball and the angular acceleration vector corresponding to the rotation of the moving ball around its own geometric center are extracted at the instant the ball-hitting event is triggered. Based on the rigid body dynamics model, and according to the translational-rotational coupling relationship between the linear acceleration vector and the angular acceleration vector, the offset vector of the striking point relative to the geometric center of the moving sphere is calculated; the offset vector includes the offset distance and the direction of action in the sphere coordinate system, which is a three-dimensional Cartesian coordinate system fixed to the geometric center of the moving sphere; Receive racket face posture data, which includes the racket face normal vector and the racket face coordinate system; The offset vector is mapped from the spherical coordinate system to the racket face coordinate system through coordinate transformation; The coordinates of the hitting point on the racket face are calculated and output based on the mapping results.

9. The multi-source data fusion method for calculating the impact point according to claim 8, characterized in that, The coordinate transformation includes: Synchronize the timestamp of the ball-hitting event with the timestamp of the racket face posture data to establish the correspondence between the ball coordinate system and the racket face coordinate system at the moment of hitting the ball; The projection calculation of the offset vector is corrected based on the changes in racket face posture within a preset time window before and after the shot.

10. A multi-source data fusion system for calculating the point of impact, characterized in that, include: An inertial data acquisition unit, built into the moving sphere, is used to collect inertial data during the movement of the sphere. The inertial data includes triaxial acceleration data and triaxial angular velocity data. The ball-hitting detection and vector extraction unit detects ball-hitting events based on the inertial data and extracts the linear acceleration vector corresponding to the geometric center of the moving ball and the angular acceleration vector corresponding to the rotation of the moving ball around its own geometric center at the instant the ball-hitting event is triggered. The offset vector calculation unit, based on the rigid body dynamics model, calculates the offset vector of the hitting point relative to the geometric center of the moving ball according to the translational-rotational coupling relationship between the linear acceleration vector and the angular acceleration vector; the offset vector includes the offset distance and the direction of action in the spherical coordinate system, which is a three-dimensional rectangular coordinate system fixed to the geometric center of the moving ball; A racket face posture receiving unit receives racket face posture data, which includes the racket face normal vector and the racket face coordinate system. The coordinate transformation unit maps the offset vector from the spherical coordinate system to the racket face coordinate system through coordinate transformation. The hitting point output unit calculates and outputs the position coordinates of the hitting point on the racket face based on the mapping result.

Images