System and method for tracking sports balls

Abstract

A system for determining potential changes in the trajectory of golf shots comprises: a sensor array that senses swing data associated with a golfer's swing and trajectory data associated with the trajectory of each of a plurality of shots hit by the golfer; and a computing arrangement that includes a data repository and a processor, the processor being configured to: store the swing data and the trajectory data in the data repository; identify shots hit by the golfer that represent misses; and determine output data based on the swing data and the trajectory data for all of the shots hit by the golfer that are not identified as misses, the output data indicating an optimal shot achievable by the golfer.

Description

[0001] Priority claim

[0002] This application claims priority to U.S. Provisional Patent Application No. 62 / 963,868, filed on January 21, 2020. The specification of the above application is incorporated herein by reference. Technical Field Background Art

[0003] In today's technologically advanced world, athletes using sports balls increasingly use measurement devices that provide information about the trajectory of the sports ball and the launching motion associated with the launch of the sports ball (e.g., the swing of a club or bat, a kicking or throwing motion, etc.).

[0004] To progress efficiently, these measurements and various parameters must be combined to create a system that can most effectively improve performance.

[0005] For example, in the sport of golf, the interactions between the various factors in the swing and equipment selection are too complex to be monitored and analyzed by any human, such as a traditional coach. When the interaction between equipment and the biomechanical response to any adjustment is taken into account, it becomes even more complex and far beyond human capabilities to determine the practice path that will produce improvement, let alone quantify the potential benefits of any given path of action. Summary of the invention

[0006] Embodiments of the present invention relate to a system for determining potential changes in the trajectory of golf shots, the system comprising: a sensor array that senses swing data associated with a golfer's swing and trajectory data associated with the trajectory of each of a plurality of shots hit by the golfer; and a computing arrangement that includes a data repository and a processor, the processor being configured to: store the swing data and the trajectory data in the data repository; identify shots hit by the golfer that represent misses; and determine output data based on the swing data and the trajectory data for all shots hit by the golfer that are not identified as misses, the output data indicating an optimal shot achievable by the golfer.

[0007] Embodiments of the present invention provide a system for tracking sports balls and analyzing player performance to generate an efficient improvement plan, thereby eliminating or reducing the need for professional sports coaches or instructors. Embodiments of the present invention include a system that includes one or more sensors that measure ball trajectories and data such as swings, kicks, and throws that launch a ball into the achieved sports ball trajectory. The system can operate over time to learn the impact of measured parameters on the launch motion (e.g., swings, kicks, throws, etc.) to calculate the impact of changes in these parameters on the resulting sports ball trajectory. Based on this learning, the system identifies the launch motion parameters to be changed and the desired degree of change to improve the performance of the athlete.

[0008] By tracking the launched sports balls, embodiments of the present invention benchmark the performance of the athlete with one or more performance scores that allow the system to calculate the impact of different possible launch motions and / or equipment changes on the athlete's performance, thereby being able to develop a plan for the athlete that identifies the changes that have the greatest positive impact on performance. Because the physics and the interactions between multiple parameters are very complex, coaches or instructors cannot perform these calculations. For human coaches, more limited focus is required because it is almost impossible for anyone to calculate the complex results of a parameter change on multiple additional parameters that may be affected by any change.

[0009] To understand the nature of the system's learning about how changes in launch motion affect the trajectory of a sports ball, consider the following golf example.

[0010] When a right-handed golfer's club path is detected to be to the left of the face angle of the clubface (e.g., a so-called "open face facing the path"), the ball will typically spin about a spin axis that is tilted to the right (e.g., as can be determined and measured by a ball tracking system), which will typically produce a ball trajectory that curves to the right.

[0011] If the detected club speed increases, while everything else remains constant, the detected ball speed and spin rate will generally increase. This will also generally produce a higher and longer detected ball trajectory.

[0012] If the detected point of impact of the ball on the club face is spaced apart from the center of gravity of the club head projected onto the club face, the detected ball speed will likely decrease and the detected trajectory curvature may also be affected.

[0013] If the detected club head attack angle increases, the detected ball trajectory after launch will generally be steeper (closer to vertical), resulting in a higher detected ball trajectory.

[0014] However, due to the complex biomechanics of the golf swing, a change in any one of these variables is likely to cause changes in the other variables, which may vary significantly from person to person. The present disclosure may be constructed based on an empirical model of each of a plurality of golfers so that the effects of a change to one variable may be predicted with confidence. For example, data corresponding to various parameters (e.g., biomechanical, equipment-related, etc.) from many previous shots of the golfer himself and / or from many other golfers may be used to train a neural network to develop a model for predicting the outcome of any single change in the golfer's performance. The complexity of the launch motion of other sports presents similar problems and is addressed in a similar manner by embodiments of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A system is shown that may be used by a golfer to measure golfer performance, determine golfer characteristics, and determine one or more effective ways to improve golfer performance in accordance with an embodiment.

[0016] Figure 2 A flow chart describing a method for determining golfer characteristics and determining effective ways to improve the golfer's performance is shown, according to an embodiment.

[0017] Figure 3 The logical sequence of events that determines the trajectory of a hit golf ball is shown.

[0018] Figure 4 A golfer improvement cycle enabled by an embodiment of the present invention is shown.

[0019] Figure 5 An example of a display of golfer benchmarks and golf characteristics is shown.

[0020] Figure 6 An example is shown of displaying golfer improvement topics in a prioritized order that can lead directly to a focused practice plan.

[0021] Figure 7 An example of implementation of an embodiment of the present invention is shown.

[0022] Figure 8 The use of two benchmark test scores is shown to classify golfers with similar problems into groups.

[0023] Fig. 9 Shown are average club delivery and ball launch and landing conditions for male golfers as measured by a golf shot tracking system. DETAILED DESCRIPTION

[0024] The exemplary embodiments may be further understood with reference to the following description and the associated drawings, wherein like elements are provided with like reference numerals.

[0025] In the following, a golfer hitting a golf ball is used as an example of an embodiment of the present invention. However, those skilled in the art will appreciate that the same principles can be applied to various other sports in which players launch a sports object such as a sports ball by, for example, hitting the ball directly (e.g., volleyball or soccer), hitting the ball with a club or bat (e.g., cricket, baseball, and golf), or throwing the sports ball with the intention of achieving a specific trajectory or final position of the sports ball.

[0026] The terms used in this application will define a coordinate system to describe the dynamic interaction between the club and the ball at impact. However, it will be understood by those skilled in the art that any other coordinate system or system may be used to analyze the three-dimensional characteristics of the impact between the golf ball and the golf club (or between any sports ball impact device and the sports ball) without departing from the scope of the embodiment, as long as the derived data enables people to understand the quality of the resulting shot and predict its improvement method. When analyzing the impact characteristics of the golf club and the golf ball, variables such as the head path, the angle of impact, the dynamic loft, the face angle, and the dynamic lie angle are useful and can be understood with respect to a coordinate system whose origin is the position of the ball before the impact. For example, the first axis of the coordinate system can be a horizontal line from the ball to a given target (e.g., the target direction), while the second axis of the coordinate system is a vertical line passing through the ball, and the third axis is a horizontal line perpendicular to the target direction, thereby forming a Cartesian coordinate system.

[0027] Those skilled in the art will appreciate that the club head path at the heel of a golf club is not the same as the club head path at the toe of the golf club, given the three-dimensional orientation of the golf club at the point of impact, and therefore the path of the actual location of the impact between the ball and the club face may be different from either of these paths. However, for the purposes of these embodiments, the point selected as a reference point when determining, for example, the strike angle and the club head path (as defined below) can be any reference point on the club face selected to provide an indication of the dynamic relationship between the ball and the club face at impact. Thus, regardless of whether the center of the club face is selected as the reference point for determining the direction of the golf club, the geometric center of the club head or the center of mass of the club head will be selected as the reference point, and the movement of the selected reference point relative to the ball at impact is determined. The strike angle is defined as the vertical direction of the club head measured relative to the horizon before impact, and the club head path is defined as the horizontal direction of the club head measured relative to the target direction before impact. Similarly, those skilled in the art will appreciate that for a club such as a driver having a significant convex radius, the face angle at the heel will be different than the face angle at the toe, given the three-dimensional orientation of the golf club face at the point of impact.

[0028] However, for purposes of these embodiments, the point selected as a reference point when determining, for example, dynamic loft and face angle (as defined below) can be any reference point on the club face selected to provide an indication of the dynamic relationship between the ball and the club face at impact. Thus, regardless of whether the center of the club face is selected as a reference point for determining the direction of the golf club, the point of impact of the ball on the club face will be selected as a reference point to determine the orientation of the selected reference point relative to the ball at impact. Dynamic loft is defined as the vertical angle of the club face normal (e.g., 90 degrees to the club face) at a reference point on the club face measured relative to the horizon at impact, i.e., the amount of loft on the club face at impact. Similarly, face angle is defined as the club face normal (e.g., 90 degrees to the club face) at a reference point measured in a horizontal plane relative to the target direction at impact.

[0029] Figure 1 A first system 100 is shown for determining golfer characteristics and for identifying the most effective way to improve a golfer's abilities. A golfer 130 swings a golf club 120 and launches a golf ball 110 on a ball trajectory 109. The system 100 contains one or more sensors 106 that record swing and shot parameters, as will be described in more detail below.

[0030] The swing and shot parameters include ball data sufficient to predict the trajectory of a golf ball and the approximate final position of the ball (i.e., the final resting position of the ball after it lands, bounces, and stops after rolling). This can be accomplished, for example, by measuring the complete ball trajectory using a camera and / or radar based sensor system 100 (or a system combining radar and camera tracking), or the system 100 can measure the launch conditions of the ball, such as ball speed, launch angle, launch direction, spin rate, and spin axis, with sufficient accuracy to enable calculation of the golf ball trajectory using predetermined information about golf ball aerodynamics and environmental conditions such as temperature, pressure, wind speed and direction, and ground conditions, as will be understood by those skilled in the art.

[0031] In addition to determining the trajectory of the golf ball, the system 100 also measures at least one swing parameter associated with the resulting ball trajectory, which is related to the golfer's swing motion and / or club delivery. In a preferred environment, the club delivery parameters include club head speed, club head path, face angle, angle of attack, dynamic loft of the club face, and impact location on the club face. However, additional swing parameters of the golfer 130, such as the position and angular orientation of the club head, hands, shoulders, hips, pelvis, and spine during the golfer's swing, are relevant parameters and may also be measured. These swing motion parameters may also be used to determine linear and angular velocities as part of a kinematic sequence that is critical to producing a given club delivery and making these actions reliably repeatable.

[0032] Figure 3 A sequence of events is shown for a sports ball to be launched along a trajectory and for the ball to continue to travel to a final resting point as it bounces and rolls after landing. The sequence begins with the movement of the golfer's arms, legs, shoulders, etc., by which the golfer delivers the club head to the ball in a certain orientation, direction, and speed. The movement of the club head in response to the movement of the golfer is also affected by the construction of the golf club (e.g., the mass of the club head, the length and stiffness of the shaft). Thus, the club face of the club head strikes the ball, causing the ball to be launched under initial ball launch conditions and then fly in the air in an aerodynamic manner until it strikes the ground under certain landing conditions (e.g., angle, speed, and spin rate), and depending on the turf conditions (e.g., hardness, angle of the impact area to the horizontal plane, etc.), the ball will bounce and / or roll to its final resting position.

[0033] In system 100, sensors 106 may include, for example, cameras 106(ab) and / or radar 106(c), but as will be appreciated by those skilled in the art, may include any sensor or combination of different types of sensors capable of measuring ball trajectory and any shot parameter and / or swing motion parameter delivered by the club, including, for example, lidar, ultrasonic, acoustic or magnetic sensors, accelerometers, etc. Some parameters may also be determined based on data from sensors attached to golfer 130, golf club 120, golf ball 110, or any combination of such sensors. However, where possible, sensors 106 are contactless to the golfer and do not require modification of the golf club or ball for use by the golfer.

[0034] The system 100 includes a processor 105 that records and analyzes swing and shot parameters sensed by sensors 106. The processor 105 determines golfer characteristics (e.g., historical data related to the golfer's swing and shot characteristics) based on the recorded parameters and stored data related to the golfer's performance (if any data has been generated), generates or updates a benchmark test for the golfer, and suggests improvements to the golfer aimed at improving the benchmark test. The benchmark test results may be presented to the golfer, for example, via a display 107, but may also be delivered via an audio device, an electronic or printed report, or any other suitable output means.

[0035] Benchmarking of a golfer is typically for a specific club type (e.g., a 6-iron), but may also be a general benchmark for all club types hit by the golfer or a group of clubs such as woods, hybrids, irons, long irons, short irons, or wedges. Typically, for a golfer, the benchmark will not differ significantly from club type to club type (e.g., a 6-iron and a 7-iron), so combining the club types in a group prior to benchmarking may be desirable as this will reduce data storage and processing requirements and allow for more efficient system operation.

[0036] As will be appreciated by those skilled in the art, the cameras 106(ab) may operate in the visual frequency domain and / or in the near infrared or infrared domain. The cameras 106(ab) may be color or monochrome and may include a light system operating therewith to ensure proper lighting during recording. The cameras will typically run 100 to 10,000 frames per second to record club delivery, swing motion parameters, and shot parameters. For cameras used to measure a golfer's hand, arm, shoulder, hip movement, etc. and determine swing motion parameters, a slower frame rate, such as 30fps, may be used because these items move slower than the club head.

[0037] As will be appreciated by those skilled in the art, high frame rates may be particularly useful when observing the interaction between a club head and a ball, because such impacts contain dynamic elements that occur within a very short time frame. In order to detect the club head position and orientation at the time of impact with the ball from the image, a minimum number of frames (e.g., 2-3 frames) before impact are required. These detections need to be adequately distributed relative to the time of impact in order to accurately observe the delivery of the club just before contact with the ball. Frames after impact may also be used, but due to collisions with the ball and possibly with the ground, the club head may deflect to another path that is not easily associated with the golfer's swing after impact. Typically, assuming that the golfer's maximum club head speed is, for example, 150mph, 4-10 frames before impacting the club are desired, which in turn limits the field of view of the camera and the frame rate. If the image is also used to determine the speed, direction, and spin of the launched ball, a similar number of 4-10 frames are typically required to obtain accurate ball launch data. In order to determine the exact time of impact with the ball, a frame rate of 2000fps or higher is desired. The contact time between the club and the ball is typically 0.5 ms, which means that at 2000 fps, there is guaranteed to be one frame with the ball and the club head in contact therebetween.

[0038] Radar 106(c) may be, for example, a continuous wave Doppler radar that transmits microwaves at an X-band frequency (e.g., 10 GHz) at a power of up to 500 milliwatts EIRP (effective isotropic radiated power), thereby complying with FCC and CE regulations for short-range international radiators. However, in other jurisdictions, other power levels and frequencies may be used in accordance with local regulations. In an exemplary embodiment, microwaves are transmitted at a higher frequency, for example, between 5-125 GHz. For more accurate measurements at lower object speeds, frequencies of 20 GHz or higher may be used. Any type of continuous wave (CW) Doppler radar may be used, including phase or frequency modulated CW radars, multi-frequency CW radars, or single-frequency CW radars.

[0039] It will be appreciated that other tracking devices, such as LiDAR, may be used in conjunction with radiation in the visible or invisible frequency regions. Current pulse radar systems are limited in their ability to track objects in the vicinity of the radar device. However, the distance that an object must be from these pulse radar systems has decreased over time and is expected to continue to decrease. Therefore, these types of radars may soon be effective for these operations, and it is contemplated that these types of radars will be used in the systems of the present disclosure described below. Throughout the application, tracking of objects is described based on the use of Doppler spectra. As will be appreciated, these Doppler spectra refer to Doppler spectra from any type of radar or LiDAR used.

[0040] The system 100 also includes a processing unit, which may include one or more processors 105 that communicate with the sensor 106, such as through a wired or wireless connection data storage device, etc., as understood by those skilled in the art. The sensor may include a camera 106 (ab) and / or a radar 106 (c) (or multiple radar devices). In an embodiment, the processor 105 includes a computer associated with the sensor 106. The processor may also be attached with a machine-readable medium for storing code and historical data. The processor may be a handheld device, a personal computer, an embedded computer, or may also be located in the cloud or any other type of remote computing device.

[0041] Figure 2 A process 200 is shown according to an exemplary embodiment. In step 205, golfer 130 performs one or more golf strokes. For each golf stroke, swing and stroke parameters are detected by system 100 and recorded in step 210. The recorded data is then analyzed in step 220 to determine the swing and stroke characteristics of golfer 130 in step 220. Processor 105 may identify certain strokes as good strokes during this analysis and identify these good strokes as representing the potential of golfer 130, while identifying less successfully executed strokes as representing errors or poor swings of golfer 130.

[0042] Typically, golfers are evaluated based on their ability in terms of how far and how efficiently they can hit the ball on a well-hit shot, and how reliably they can repeat this type of well-hit shot. Therefore, when determining the golfer's characteristics in step 220, all shots will be used for part of the analysis (for example, when determining the level of repeatability of a particular type of shot), while certain shots (for example, shots identified as poorly hit) may be ignored for other analyses (for example, evaluating the efficiency of transferring power to the ball). Preferably, the golfer characteristics determined in step 220 are shared with the golfer. This may take the form of: "You hit the ball a landing distance of 158m with a left to right curve on a medium height trajectory. 50% of your shots land to the right of the target line." This may also be as follows: "You hit the ball 158m with a left to right curve on a medium height trajectory. 50% of your shots land to the right of the target line." This may also be as follows: Figure 5 The shown is shown graphically.

[0043] While the golfer's characteristics are preferably determined from measurements made by system 100, the golfer's characteristics may also be input by the user, such as "with my 6 iron, my club head speed is typically 80 mph, my shot shape is typically a high fade, and my distance is 150 m."

[0044] The system 100 may also incorporate input from the golfer when performing the shot analysis. In step 218, the golfer may input an intent, for example, the golfer may input an intent to hit a specific distance, a specific shot shape (e.g., draw, fade, or straight), a specific height (e.g., low, medium, or high), or a specific spin rate (e.g., low, medium, or high), or any combination of such intents. If no golfer intent 218 is available, the system 100 may, for example, assume as intent a repeatable good shot that lands on the target line at the same distance as a previous good shot using the same club.

[0045] Golfer 130 may also have limitations that affect the golfer's abilities (e.g., reduced flexibility, reduced strength, etc.) These may be input to system 100 in step 219. These limitations may also be associated with information about the golfer's height, arm and leg length, and overall body size.

[0046] In step 215, the golfer 130 may enter details about the equipment he / she is using, such as club type (e.g., driver, 3-wood, 2-hybrid, 7-iron, 54-degree wedge, etc.), shaft flex (e.g., regular, soft, stiff, extra stiff), and / or ball information (e.g., premium ball, 2-pieces ball, range ball, etc.). The information may also include more detailed information, such as club head mass, static loft, center of gravity location, moment of inertia, coefficient of restitution, lie angle; shaft length, weight, flex properties, grip type; and / or ball weight, coefficient of restitution, friction properties, etc.

[0047] The system 100 may also automatically determine equipment information in step 215 based on data from the sensors 106. For example, the club type or at least the club category (e.g., driver, wood, hybrid, iron, wedge, or putter) may be determined by identifying the equipment from images obtained by the sensors 106(ab) based on classification, for example, using a machine learning-based network or other type of classification technique.

[0048] Based on the golfer characteristics determined in step 220, any equipment details determined in step 215, and any golfer intent information entered in step 218, as well as any golfer limitations entered in step 219, the processor 105 determines optimal yet golfer-achievable shot parameters and club delivery and swing motion parameters associated with such optimal shots in step 230. In determining what the golfer is achievable, the system 100 may incorporate prior knowledge about which aspects of the golfer's swing are more susceptible to change than others.

[0049] For example, changing the face angle, the impact location on the club face, the golfer's angle of attack, or the club head path may be relatively easy, while changing the club head speed may be more difficult. This ranking of the ease of implementing certain changes will affect the recommendations generated by the system 100, as will be described in more detail below. As will be appreciated by those skilled in the art, the system 100 may adjust the ease of correction of any aspect of the golfer's swing based on the amount of room for improvement in those characteristics (i.e., as the golfer implements the recommendations and the difference between the current swing characteristics and the achievable optimal situation decreases, changes to other characteristics may be ranked as more promising).

[0050] In step 235, the system 100 accesses a database of stored information to determine an optimal shot for the golfer. For example, the system 100 may retrieve from the database in step 235 information related to how to efficiently hit a ball at a given club head speed with a shot shape that matches as closely as possible the characteristics determined for the golfer in step 220. This database may, for example, contain data corresponding to thousands of shots from professional golfers, and / or may contain an artificial model golfer adapted to match the current golfer as closely as possible by adjusting parameters to match the golfer characteristics determined in step 220. In step 230, the golfer's optimal shot may be determined by using a neural network to identify which of a number of optimal trajectories stored in a database 235 (e.g., containing data about multiple shots by multiple golfers using different types of equipment) most closely matches the golfer's characteristics determined in step 220, or based on known historical data learned about the optimal trajectory that is most easily achieved given the golfer's characteristics determined in step 220 (e.g., based on an analysis of performance changes of golfers identified as having similar relevant characteristics, where the performance changes follow specific practice suggestions for similar starting points, etc.).

[0051] The trajectory identified as optimal may depend on many factors that it is impossible for a human to effectively consider. For example, for a driver shot, the optimal ball trajectory is generally considered to be the ball trajectory that maximizes the distance from the launch location of the ball to the final resting location while minimizing the spread of the ball's final location away from the target line. However, some golfers may wish to set the value of the carry distance (e.g., the distance from the launch location to the point where the ball first contacts the ground) to be higher than the distance to the ball's final resting location. This preference may be input by the user.

[0052] The optimal trajectory of the final resting position will vary depending on the turf conditions. For very hard turf conditions, the longest final resting position is generally associated with a trajectory that is lower in flight and thus has a shallower landing angle than a higher ball trajectory that maximizes the distance from the take-off point. The optimal trajectory of a driver will also depend on environmental conditions. At a higher altitude with lower air density, the optimal trajectory will generally have a higher launch angle than the optimal trajectory at a lower altitude. In addition, wind conditions will have an impact on the optimal trajectory. When playing at links courses, the wind is generally very strong. This makes a lower ball trajectory with a shorter time in the air, a lower apex height, and generally lower wind speeds more preferred. However, when playing at a park course where trees limit the wind speed that affects the ball, a higher ball trajectory may be more preferred. The purpose of club types other than drivers and therefore the optimal trajectory are not necessarily to optimize the distance of the shot.

[0053] For example, the type of shot using these other club types is primarily used to hit the ball a certain distance when the golfer swings high (fullswing) to reach a certain target area and control the end point of the ball. Depending on the shot distance (e.g., the distance from the current position of the ball to the desired final resting position of the ball) and the turf and environmental conditions, the golfer's optimal trajectory will be different. For example, if the green is very soft, it is preferred to have a high ball trajectory so that the amount of bounce and roll will be minimized. On the other hand, if the green is very hard, it is usually preferred to drop the ball outside the green and let the ball bounce and roll onto the green. Defining what is the optimal trajectory that defines the optimal shot is a very complex problem, usually involving many complex calculations, and considering many variables related to the golfer, equipment, environment and turf conditions, which is impossible for a person to do.

[0054] When determining the benchmark of a golfer's performance in step 240, two different aspects are generally considered: 1) the optimal shot determined in 230 (which is Figure 5-6 The golfer's actual shot shape compared to the ball (referred to as the trajectory score in ), for example, may omit poorly struck shots such as those described above; and 2) the repeatability with which the golfer is able to execute the same shot (referred to as the trajectory score in ). Figure 5-6The optimal golf shot determined in step 230 is likely to be different from the optimal golf shot corresponding to the golfer characteristics determined in step 220. For example, if the golfer continuously hits the driver on the heel of the club face, not only will the ball speed be lower than the optimal speed, but the actual shape of the golfer's shot as reflected in the golfer characteristics may be a slice shot due to the so-called horizontal gear effect caused by the heel impact position, while the optimal shot shape may be a straight shot. The benchmark test may also include a score that reflects the golfer's ability to consistently land the ball on or in the target area.

[0055] Figure 5 An example of feedback provided to a golfer on display 107 is shown. In this case, a benchmark of the golfer's performance is reflected in two parameters - trajectory and accuracy. In addition, the optimal trajectory and landing point distance / total number are shown to be close to the actual trajectory and landing point distance / total number. Finally, a dispersion graph is shown, which in this example indicates that more shots are to the right of the target than to the left.

[0056] Based on the golfer characteristics determined in step 220 and the golfer's optimal shot determined in step 230, the swing model 245 is calibrated to the golfer's swing motion and club delivery characteristics. The processor 105 then analyzes potential improvements 250 by adjusting all parameters and / or possible swing changes in the swing model to determine which parameters will have the greatest positive impact on the benchmark of the golfer's performance. The potential improvements to the benchmark performance are then compared for each parameter or swing change - while perhaps also considering changes that are more easily achieved by the golfer. The results of the suggested improvements may then be presented to the golfer in step 250, such as Figure 6 The suggested changes may include (a) only those changes with the greatest positive impact; (b) all changes with a positive impact on the baseline; or (c) one or more selected changes with a positive impact on the baseline.

[0057] exist Figure 6 In , both a trajectory score and an accuracy score are given. Therefore, when determining the overall potential improvement, the two scores are combined. This combination can be done in a variety of ways. A simple way is to give the two scores similar weights. This means that the overall performance score can be the sum or average of the two scores. Other scoring mechanisms with more or different sub-scores can also or alternatively be done. However, in order to be able to sort the improvement suggestions, it is preferred to combine the sub-scores into an overall score.

[0058] Alternatively, each of the multiple improvement categories may have its own scoring mechanism. For example, one category may be efficiency / track using only the track score for ranking, while another category may be accuracy using only the accuracy score for ranking improvement. Golfers may then decide which aspect of their game to improve based on these categories.

[0059] The expected relative effort required by the golfer to implement the suggested changes in parameters or swings may be taken into account when determining the recommendations in step 250. Additionally, some changes may be omitted from the analysis, either automatically by the system 100 or by the golfer 130. For example, the golfer may decide to omit equipment changes, or, vice versa, the golfer may request to only view recommendations regarding equipment changes. The latter is typically done by a professional fitting instructor.

[0060] An alternative method of determining the suggested improvements in step 250 may be where a swing model is determined by an artificial intelligence system in step 245 based on historical data of multiple golfers who have had similar swing motions or club delivery errors in the past and the degree and rate at which they have improved over time and the changes they have made to improve. The effectiveness of such a system will increase over time as the system records and uses more and more data.

[0061] The improvements suggested to the golfer may result from the individual focused training program determined in step 260 for changing specific parameters or swing movements. The training program determined in step 260 may be supplemented by suggestions for different equipment, such as another type of club or clubs with different lofts or otherwise different specifications, in step 265, as will be appreciated by one skilled in the art. The suggestions may also include suggestions for individual exercises known to have a positive impact on swing errors exhibited by the golfer in step 268.

[0062] Another aspect of an embodiment of the present invention uses player characteristics and suggestions to suggest changes to a golfer's club set, a so-called club set. The rules of golf allow a player to play up to 14 clubs. However, there are more than 14 types of clubs available. Therefore, each golfer must determine the type of club to carry (e.g., driver, wood, hybrid, iron, wedge, putter), and select club specifications (e.g., loft, length, weight, etc.) for each selected club. Typically, the decision to switch from a wood to a hybrid and from a hybrid to an iron at which distance is based on feeling. Based on the golfer characteristics determined in step 220, the information retrieved from the database in step 235, and the swing model determined in step 245, the system 100 can suggest changes and / or club selections to the golfer's clubs for various distances to the golfer to maximize the benchmark performance determined in step 240 among all the golfer's clubs. In this regard, the system 100 can use information about the expected number of strokes hit by each club during a typical round of golf.

[0063] Golfers often consider the ability to hit a certain distance with some spacing, for example, a golfer may prefer to have adjacent clubs separated by 10m in landing distance for high swings. However, it is usually not possible to do this with all 14 clubs from driver distance down to the shortest clubs such as lob wedges. Typically, golfers prefer 10m spacing between clubs up to the longest iron and possibly the longest hybrid club, and then choose larger distance spacing for the remaining longer clubs. The system 100 can use similar logic to determine the club set.

[0064] This club set recommendation can be generated even if the golfer has not hit all of his / her clubs. In fact, the club set can be generated after the system 100 has analyzed only a few shots - even if those shots were all made with a single club type. This is possible because the way a golfer hits, for example, a 6-iron will be very similar to the way a golfer hits a 4-iron or a 9-iron or other clubs such as drivers and wedges. This is because the body movements to perform almost any type of golf shot, other than putts and chip shots, are essentially the same. Fig. 9 , averages of club delivery, ball launch and flight data for male touring professional golfers are shown. It is observed that club delivery and ball launch data for adjacent clubs are highly correlated, indicating that it is possible to predict club delivery and ball flight details for one club based on data for another club. It has been found that this is also true to a large extent for amateur golfers. However, typically appropriate club set analysis involves shots with a driver and one or more irons.

[0065] Yet another aspect of the present disclosure is to use a database of golfer swing motions and club delivery characteristics to determine whether a particular device will have a positive impact on a golfer's game. This can be used to identify clubs that will be useful for golfers with certain common swing characteristics. For example, given the swing characteristics of these golfers, a product might be identified as having a positive impact on the performance of those golfers.

[0066] exist Figure 7 In FIG. 1 , an implementation example of an exemplary embodiment is shown. Figure 7 In the diagram, all circles represent measurements, parameters, or properties, while the rectangular boxes are the algorithms, processes, or networks that produce outputs from a given set of inputs.

[0067] Figure 7 Include Figure 3 An example of a complete model of the logical sequence of a golfer's swing to the ball trajectory shown. Some embodiments reuse only Figure 7 Portions of the model shown.

[0068] Figure 7 An embodiment 300 is shown that includes a swing model (SM) 330 that can calculate a club delivery (C) 311 based on a swing motion 310 and club swing properties (pCs) 320 . Figure 7 Also included is a collision model (CM) 331 that calculates ball launch conditions (L) 312 based on club delivery 311, club collision properties (pCc) 321, and ball collision properties (pBc) 322. A ball flight model (BFM) 332 then calculates ball trajectory (T) 313 based on ball launch conditions 312, ball aerodynamic properties (pBa) 323, and environmental conditions (cE) 324 such as wind speed and direction, air density, temperature, and air pressure. A bounce and roll model (BRM) 333 then calculates the ball's final resting position (F) 314 based on ball trajectory 313, specifically landing speed, landing angle, and landing spin rate, as well as turf conditions (cT) 325.

[0069] In the following, various embodiments will be described. It is clear from the above description of the different models that a number of variations of the embodiments can be achieved by using the described models or directly measuring the parameters. For example, it is apparent that if the ball launch conditions 312 are known, then the ball trajectory 313 and the final resting position 314 do not need to be directly determined or measured, as the ball trajectory 313 and the final resting position 314 can be calculated with reasonable accuracy using the ball flight model 332 and the bounce and roll model 333. In some embodiments, the use of the models 332, 333 is preferred over directly measuring the parameters, while in other embodiments, the measured parameters are preferred.

[0070] The first step in this embodiment is a swing characterization (SC) 340 for generating a golfer characteristic (GC) 315. This is typically done by recording the swing motion 310 and / or club delivery 311 data for several shots. Additionally, ball launch conditions 312, ball trajectory data 313, and the final resting position 314 of the ball may also be used for the same shot. The process may include intelligent filtering of good shots and not-so-good or bad shots.

[0071] In one example, golfer characteristics 315 are described as follows: "6 iron with 82 mph club head speed, -3.2 degree attack angle, -3 degree face to path, +1.4 degree club head path."

[0072] The swing characterization 340 may also include a statistical description of the golfer's variability, such as described in terms of a standard deviation or a 90% confidence interval or other statistical description. In some embodiments, it may also be desirable to characterize shots that are not well hit. Generally, elite golfers are more interested in eliminating bad shots from their game than in slightly improving their good shots. Therefore, these golfers may be particularly interested in characterizing the swing and club delivery of those shots that are not well hit.

[0073] In order to determine the golfer's optimal shot 316, shot optimization 350 is performed. Shot optimization 350 uses golfer characteristics 315 as input. Golfer characteristics 315 can be determined based on swing characterization 340, or based on user input characteristics (UIC) 341 or other predetermined methods. Naturally, golfer characteristics 315 can be a combination of swing characterization 340 and user input characteristics 341.

[0074] When determining the optimal shot 316, the system 100 may use a user intent (UI) 345. The user intent 345 may be expressed as a user intent for a desired shot type (e.g., a preferred shot shape, such as a high shot, a low shot, a hook, a slice, etc.), or other characteristics such as a specific launch point distance or landing range, spin rate, etc.

[0075] Shot optimization 350 may also use ball launch conditions 312, ball trajectory 313, and / or final resting position 314. As described above, these may be calculated from collision model 331, ball flight model 332, and bounce and roll model 333 depending on available measurements.

[0076] The optimal shot 316 can be represented in a variety of ways. The optimal shot can be expressed using parameters describing the following: optimal swing motion Sopt; club delivery Copt; ball launch condition Lopt; ball trajectory Topt; and / or final resting position Fopt. In a preferred embodiment, the optimal shot 316 is described by the optimal club delivery Copt and the optimal ball launch condition Lopt, for example, consisting of the following: one or more optimal values ​​of club delivery parameters (e.g., club head speed, club head path, hitting angle, face angle, dynamic loft, dynamic bottom angle, impact position, etc.), and one or more optimal values ​​of ball launch condition parameters (e.g., ball speed, launch angle, launch direction, spin rate, spin axis, etc.).

[0077] In one embodiment, shot optimization 350 calculates various ball trajectories and final resting positions based on various club deliveries that are close to golfer characteristics 315 and determined to be achievable by the golfer using collision model 331, ball flight model 332, and bounce and roll model 333. The optimal shot that is closest to golfer characteristics 315 is found from a database of many different ball trajectories and optimal trajectories. Fig. 9 An example of optimal club delivery and trajectory is shown in FIG, which represents the average level of male touring professional golfers. As part of shot optimization 350, the database may be composed of thousands of golf courses from similar Fig. 9 The club delivery and ball launch conditions of the professional golfers shown are composed, so the database covers various clubs, various club head speeds and various different shot shapes. The database can also be replaced in whole or in part with a collision model 331 configured to consider only efficient collisions.

[0078] Benchmarking 360 is accomplished by comparing the golfer's actual ball launch conditions 312, ball trajectory 313, and / or final resting position 314 to corresponding data for an optimal shot 316. Benchmarking 360 may also use the golfer's swing motion 310 and / or club delivery 311 and compare to corresponding optimal data. Benchmarking 360 is largely a comparison of swing motion (S) 310 to Sopt, club delivery (C) 311 to Copt, ball launch conditions (L) 312 to Lopt, ball trajectory (T) 313 to Topt, and / or final resting position (F) 314 to Fopt.

[0079] The result of benchmark test 360 is benchmark 317. Benchmark 317 is preferably expressed in one or more scores Q, which makes it possible to quantify benchmark 317. In a preferred embodiment, two scores are calculated: trajectory score Q 轨迹 , which represents the degree to which a golfer is able to achieve an optimal shot trajectory, typically using only the golfer's well-struck shots; and an accuracy score Q 精确度, which represents the degree to which a golfer is able to get the ball to stop at a given target, typically using all of the golfer's shots.

[0080] In the extreme case where the golfer can repeat the optimal shot every time, the base score will be the largest, like Figure 5 and 6 The same is true for 100 points depending on the scale.

[0081] The golfer's benchmark score 317 can also be used to categorize golfers. Figure 8 In the example of FIG. 1 , golfers are classified according to the trajectory and accuracy scores achieved. This classification can be done for all golfers' clubs, for a number of golfers, or can be broken down into golfers with similar club head speeds. This type of golfer classification can be helpful in understanding what type of service or product will benefit which type of golfer.

[0082] To determine the recommended improvements 370 for the golfer, all of the various parameters that determine the ball trajectory and final resting position are tested to determine the extent to which each parameter or any combination of parameters affects the benchmark score 317. This can be done, for example, by adjusting each parameter from the golfer's actual value to a value corresponding to an optimal shot 316 and calculating the change in the benchmark score 317. This can be done on all of the golfer's shots or a subset of the golfer's shots. The parameters to be tested can include, for example, the golfer's swing motion S or club delivery C, club swing properties pCs, club impact properties pCc, and / or ball impact properties pBc.

[0083] Mathematically, the calculation used to determine the recommendation can be accomplished by calculating dQ / dP, where dQ is the change in benchmark score Q 317 when changing the parameter (or set of parameters) P from the actual value to the golfer's personal optimal value Popt corresponding to the optimal shot 316 (dP=Popt-P).

[0084] All parameter changes that have a positive impact (dQ>0) on the baseline score 317 may be suggested. However, in a preferred embodiment, the suggestions are ranked according to how much each parameter change is expected to improve the baseline score 317. In one embodiment, only changes that provide the greatest improvement in the baseline score 317 are suggested - "whichever change will have the greatest impact - the best bang for the buck".

[0085] The suggestions may be filtered to show only changes within a certain category, such as swing motion or club delivery changes, without suggesting changes in other areas, such as equipment changes. The suggestions may also be limited to showing only equipment parameter changes, such as changes in club shaft specifications, club head model specifications, ball shape specifications, etc. The latter case is typical of the fitting stage, where the equipment is fitted to the golfer while the golfer is actually swinging a golf club.

[0086] The suggestion might be something like Figure 6 The suggestions shown in the example, where "spin tilt" has been identified as the most influential parameter, which if optimized would increase the benchmark score 317 the most. Figure 6 In the example, suggestions are made that are less impactful but still have a significant positive impact (e.g., changing the swing direction, impact location, etc.). Suggestions for improving a specific parameter such as "spin inclination" can be associated with an optimal value and an acceptable range (e.g., spin inclination needs to be 25.5 degrees within a range of +-1 degree), which reflects the natural variability of golfer consistency and measurement tolerance. The acceptable range can also be adjusted to reflect the skill level of the golfer.

[0087] It should be noted that a suggestion to improve a particular parameter may involve changing the mean of the parameter achieved by the golfer, but may also involve reducing the variability of the parameter, even if the average actual value achieved by the golfer is equal to the optimal value. Most commonly, the suggestion is to improve both the mean and the variability.

[0088] An example of a suggestion might be to change the golfer's angle of impact with a driver from an average of 0 degrees to an average of +5 degrees. The system might further calculate the expected improvement resulting from such a change. For example, the system might calculate that if a particular defect is eliminated or minimized, the golfer could achieve an 11-yard increase in landing distance and a 5-point improvement in benchmark score 317 (e.g., based on a scale of 0 to 100). Such motivation for the golfer to follow the suggestion cannot be provided by an instructor or coach, but rather requires a system with a processor that is capable of performing all the necessary calculations to, for example, process many shots by the golfer and potentially hundreds of other golfers.

[0089] Additional embodiments provide users with unique outputs that are not available from human coaches. For example, data may be superimposed on a video of a golfer's swing that shows the positioning changes of different body parts made to implement a swing change suggestion. For example, the video may be altered (e.g., by superimposing a graphic or separate video image) to show the desired repositioning of the foot relative to the ball to achieve a desired change in the angle of impact or to show a new path for the shoulder or elbow to follow to achieve another desired change. The changes so suggested by the system 100 may also be recorded, and the data may be tagged with this information so that subsequent performance changes may be associated with the swing change suggestions. Those skilled in the art will appreciate that the system 100 may also measure and record data corresponding to the golfer's subsequent golf strokes to measure the extent to which the golfer actually implements these suggested swing changes. These data may also be associated with the swing change suggestions and are related to the golfer's performance changes measured over time.

[0090] As will be appreciated by those skilled in the art, this enables the creation of a database of multiple shots from various golfers having various biometrics and swing characteristics so that a neural network can be trained to more accurately identify the changes most likely to result in improvement for a particular golfer and to quantify the potential improvement associated with implementation of a given suggestion. Figure 6 The golfer shown the data can implement the suggested changes, and the golfer's subsequent performance changes are measured and recorded (while also recording the golfer's actual physical swing changes in response to the suggestions).

[0091] The system 100 may also track the timing and results of the associated focused exercises suggested to the golfer. This data is then stored in a database, and the data for this golfer and all other golfers (or any selected subset of golfers having, for example, one or more selected characteristics in common) who received the same or similar advice may then be used to train a neural network that compares predicted results with the results actually achieved. This will allow the neural network to identify golfer characteristics that indicate greater or lesser potential for positive change as a result of any advice, and to improve the ranking and quantification of suggested changes and their associated potential impact on a particular golfer's game. Additionally, by reviewing the focused exercise data, the neural network may identify golfer characteristics that present a particular exercise regimen that is more or less effective for any given golfer.

[0092] Those skilled in the art will also appreciate that embodiments need not go all the way to identifying specific recommendations for valuable focused practice. System 100 identifies only one or more of the most impactful changes, such as strokes, based on the above analysis, and the golfer or coach can choose a way to make such changes to improve the effectiveness of practice. If such a system can then provide data about optimal stroke trajectories to the golfer, this will be even more valuable because the golfer will then understand the actual change in ball flight that the suggested improvements will bring. Finally, the system can calculate benchmarks as described above based on optimal strokes to provide more data that humans cannot calculate.

[0093] It will be appreciated by those skilled in the art that the system of the embodiment combines and analyzes various parameters and generates data that was previously unavailable due to the subtle interaction of many factors to allow sports athletes to understand the quantitative results that can be obtained through different courses of action in a novel way that cannot be used by human advice and training providers. In addition, it will be appreciated by those skilled in the art that much of this data generated by radar and image-based sports ball tracking systems, as well as historical data that can show changes in performance of individual golfers, can be used to train neural networks to comb through the relationships between a large number of variables in a way that far exceeds the capabilities of the human brain to generate entirely new categories of data. Various changes can be made to the embodiments without departing from the scope of the invention, which is intended to be limited only by the appended claims.

Claims

1. A system for determining a potential change in a trajectory of a golf shot, comprising: a sensor array that senses swing data including one or more swing parameters associated with each swing motion or club delivery in a plurality of swings by a golfer and trajectory data associated with the trajectory of each of a plurality of shots resulting from the plurality of swings; A computing arrangement comprising a data repository and a processor, the processor being configured to perform the following steps: storing the swing data and the trajectory data in the data repository; determining a golfer's swing characteristics and a ball-striking characteristics based on the swing data and the trajectory data, the swing characteristics including a swing motion or club delivery parameter representative of the golfer, and the ball-striking characteristics including a trajectory or accuracy parameter representative of the golfer; determining an optimal shot achievable by the golfer based on at least the swing characteristic and the shot characteristic, the optimal shot being related to a swing motion or club delivery parameter achievable by the golfer such that trajectory or accuracy is improved relative to the trajectory or accuracy parameter representative of the golfer; as well as Output data is generated that is indicative of the optimal shot achievable by the golfer, or the swing motion or club delivery parameters associated with the optimal shot.

2. A system according to claim 1, wherein the processor is configured to identify shots hit by the golfer that represent misses, and wherein the swing data includes club head speeds of at least a portion of the shots hit by the golfer, and wherein the optimal shots are determined based on the club head speeds of the shots hit by the golfer that are not identified as misses.

3. The system of claim 2, wherein the swing data includes at least one of a launch angle, a club head path, a dynamic loft angle, and a face angle for at least a portion of the shots hit by the golfer.

4. The system of claim 3 , wherein the processor additionally determines the golfer's accuracy score based on the trajectory data indicating a likelihood that the golfer's shot will land within a target area, and wherein the processor further determines the golfer's trajectory score based on a degree of correspondence between the trajectory data indicating the trajectory of the shot hit by the golfer and a trajectory associated with the optimal shot.

5. The system of claim 4, wherein the processor additionally identifies a change to the golfer's club delivery that will have a positive impact on one of the golfer's accuracy score and the golfer's trajectory score.

6. The system of claim 1 wherein the processor is configured to identify shots hit by the golfer that represent misses, and wherein the output data is based on all of the shots hit by the golfer that are not identified as misses.

7. The system of claim 1, wherein the computing arrangement is configured to: storing the swing data and the trajectory data in the data repository; identifying a shot hit by the first golfer indicating a miss; determining output data based on the swing data and the trajectory data, the output data indicative of an optimal shot achievable by the first golfer; creating a training set including comparison data from the data repository containing swing and trajectory data for a plurality of golf shots for each of a plurality of additional golfers, the comparison data including comparison golfer data corresponding to golfers of the additional golfers having swing and trajectory data preliminarily identified as corresponding to the swing and trajectory data for the first golfer; training a neural network using the comparison data and data corresponding to changes in the golf strokes of the first golfer before and after one of a swing change recommendation and an equipment change recommendation; creating a second training set including golfers identified in the comparison data who demonstrated a similar improvement as the first golfer from before to after receiving the same of the swing change suggestions and equipment change suggestions received by the first golfer; and The neural network is trained using the second training set.

8. The system of claim 7, wherein when the one of a swing change suggestion and an equipment change suggestion is a swing change suggestion, the output data includes a video of the first golfer superimposed with data showing the swing change suggestion.

9. A method for training a golfer, comprising: sensing, by the sensor array, swing data including one or more swing parameters associated with each swing motion or club delivery in a plurality of swings of the golfer, and trajectory data associated with the trajectory of each of a plurality of shots resulting from the plurality of swings; storing the swing data and the trajectory data in a data storage repository; identifying, by a processor, one or more of the plurality of shots as misses, wherein the misses are identified as certain shots having a lower success rate than other shots in the trajectory data; as well as The processor determines an optimal swing or optimal club delivery and an optimal trajectory that the golfer can achieve based on one or more swing parameters associated with each swing or club delivery in the multiple swings and the trajectory data, while excluding those shots identified as misses from the determination results.

10. The method of claim 9, wherein the swing data includes club head speeds for at least a portion of the shots hit by the golfer, wherein the optimal club delivery is determined based on the club head speeds for the shots hit by the golfer that are not identified as misses.

11. The method of claim 9, wherein the swing data includes, for at least a portion of the shots hit by the golfer, a launch angle, a club head path, a dynamic loft angle, and a face angle.

12. The method of claim 9, further comprising: determining an accuracy score for the golfer based on the trajectory data indicating a likelihood that the golfer's shot will land within a target area; as well as A trajectory score for the golfer is determined based on the degree of correspondence between the trajectory data indicating the trajectory of the shot hit by the golfer and the optimal trajectory.

13. The method of claim 12, further comprising: A change to the golfer's club delivery is identified that will have a positive impact on one of the golfer's accuracy score and the trajectory score.

14. The method of claim 9, further comprising: An optimal trajectory achievable by the golfer is determined by the processor based on the swing data and the trajectory data for all of the shots hit by the golfer that are not identified as misses.

15. A system for determining a potential change in a trajectory of a golf shot, comprising: a sensor array that senses swing data including one or more swing parameters associated with each swing motion or club delivery in a plurality of swings by a first golfer and trajectory data associated with the trajectory of each of a plurality of shots resulting from the plurality of swings; A computing arrangement comprising a data repository and a processor, the processor being configured to perform the following steps: storing the swing data and the trajectory data in the data repository; determining output data based on the swing data and the trajectory data, the output data indicative of an optimal shot achievable by the first golfer; creating a training set including comparison data from the data repository containing swing and trajectory data for a plurality of golf shots for each of a plurality of additional golfers, the comparison data including comparison golfer data corresponding to golfers of the additional golfers having swing and trajectory data preliminarily identified as corresponding to the swing and trajectory data for the first golfer; as well as A model is trained using the comparison data and data from the data repository for each of the plurality of additional golfers, the data including swing and trajectory data representative of the golfer's progress over time.

16. The system of claim 15, wherein the data in the data repository includes data representing swing and equipment change recommendations temporally associated with the data from the plurality of additional golfers, and wherein this data is included in the training set.

17. The system of claim 15, wherein the processor is configured to identify shots hit by the first golfer that represent misses.

18. The system of claim 15, wherein the model is a neural network.

Images