FR3142360A1 - Method for detecting golf movement from an electronic device such as a watch - Google Patents

Abstract

The present invention relates to a golf movement detection method configured to operate on an electronic device, in particular worn on the wrist, comprising at least one three-axis accelerometer and/or a three-axis gyroscope, and cooperating with a calculation unit, a movement of golf being defined by a golf movement carried out by a user handling a golf club causing said club to describe a typical trajectory leading to an impact between the face of the golf club and a golf ball, the method comprising the implementation of a golf movement detection function and an impact detection function, the movement detection function being configured to, at any time, detect the performance of a golf movement associated with one of a set of golf movements of predefined golf shot movements, and the impact detection function being configured to detect an impact between the golf club and the golf ball, from measurements carried out by at least sensor embedded in the electronic device. Abstract Figure: Figure 7

Description

Method for detecting golf movement from an electronic device such as a watch

The present invention relates to the field of electronic devices, worn on the wrist or on another part of the arm, such as bracelets or electronic watches, also called connected watches or even “smartwatch” in English. More specifically, the present invention relates to an electronic device worn on the wrist or on another part of the arm, in particular a watch, equipped with hardware and software means for detecting golf shots.

Golf players practicing their sport on a course generally need to track their score, that is to say the number of shots it took them to put the ball in the hole. The accumulation of scores per hole allows you to obtain the player’s score for the course. By comparing his score to the par of the course, the player can evaluate his performance. In competition, this allows competitors to be ranked.

Particularly for amateur players, or those outside of competition, it is tedious for players to note each shot played to know their scores.

There is therefore a need for a method capable of calculating a golf player's score in real time by detecting each of his shots automatically.

Document US 10,709,945 B2 describes a concept aimed at detecting golf shots using sensors embedded in a bracelet or a watch case. However, this document is limited to describing expected results, without disclosing a concrete method making it possible to achieve this detection of golf movement. In addition, this document outlines a concept requiring an impact or vibration sensor embedded on the object (presumably the club or ball) affected by the impact.

The invention thus aims to eliminate at least some of these drawbacks. In particular, the invention makes it possible to dispense with a sensor carried by the club or by the ball.

PRESENTATION OF THE INVENTION

More specifically, the invention relates to a golf movement detection method configured to operate on an electronic device worn on the wrist or on another part of the arm, such as a bracelet or a watch, comprising one or more angular measuring means or linear, particularly in orientation, speed or acceleration. In particular, to implement the invention, the electronic device in question may comprise at least one three-axis accelerometer and/or a three-axis gyroscope and/or at least one three-axis magnetometer, cooperating with a calculation unit and a space memory.

A golf movement being defined by a golf movement made by a user handling a golf club, the club having a face, and causing said club to describe a typical trajectory resulting in an impact between the face of the golf club and a ball. golf, the method comprises implementing a golf motion detection function and an impact detection function, the motion detection function being configured to, at any time, detect the completion of a golf movement associated with one of a set of predefined golf shots, including a drive, a swing and a putt, and the impact detection function is configured to synchronously detect an impact between the club golf ball and the golf ball, from measurements carried out by at least sensor embedded in the electronic device.

According to one embodiment, the golf movement detection function and the impact detection function being implemented solely from measurements carried out by at least one sensor embedded in the electronic device, such as a bracelet or a shows, excluding sensors on board the club or the ball, among at least one three-axis accelerometer and at least one three-axis gyroscope.

Preferably, the measurements taken come from six on-board sensors: accelerometers and gyroscopes following three direct orthogonal axes, respectively.

According to one embodiment, the measurements are carried out continuously and are recorded at a first sampling step at least equal to 10 Hz, in particular approximately equal to 50 Hz plus or minus 10%, and at a second sampling step at less equal to 100 Hz for measurements carried out over at least one reduced time window.

According to one embodiment, the implementation of the golf movement detection function comprises, in particular on the condition that at least one signal corresponding to a measurement has an amplitude exceeding a predefined amplitude threshold,:

the application of a low-pass filter to the signals corresponding to the measurements, in particular a first-order low-pass filter, in particular to remove high-frequency noise and residues contained in the signals corresponding to the measurements;

the normalization of each of the signals corresponding to the measurements carried out, in particular in acceleration, from measurements carried out by the at least one accelerometer, and/or in rotation speed, from measurements carried out by the at least one gyroscope;

comparing the normalized signals to respective simple thresholds;

the identification, for each normalized signal, of at least one peak, when the amplitude of the normalized signal exceeds a simple threshold;

the recording, for each normalized signal, of a date of occurrence of the peak of greatest amplitude.

According to one embodiment, the method further comprises:

if at least one peak has been detected, the characterization, for each normalized signal, of 1 to 3 peaks having the greatest amplitudes in height and/or width and/or prominence;

comparing, for each standardized signal, its height, its width and its prominence to a respective threshold of minimum width and/or minimum height and/or minimum prominence so as to confirm or refute the existence of a corresponding peak;

the regularization of each normalized signal comprising at least one confirmed peak, by means of a temporal homothety consisting of:

contract or stretch each normalized signal, temporally, so that each normalized signal has, over a time window having a predefined regularized duration, for example equal to 2.5 seconds, a regularized main peak at a predefined main peak date , for example at 1.5 seconds in the time window, and where appropriate at least one secondary peak having a secondary peak date at a predefined date in the time window, for example 0.67 seconds before the main peak with a time having undergone temporal homothety.

According to one embodiment, the method further comprises:

the regularization of each normalized signal by means of an amplitude scaling consisting of:

contract or stretch each normalized signal, in amplitude, so that the normalized signal has, over the time window, an amplitude of the main peak equal to a predefined value.

The implementation of the golf motion detection function may include executing a golf motion recognition algorithm comprising, from golf shot datasets:

comparing each regularized normalized signal and a model signal corresponding to an average golf swing movement for a set of golf movement types, including a drive and a putt, said comparison comprising verifying that the regularized normalized signal is within an envelope defined between each model signal shifted by a negative offset and the same model signal shifted by a positive offset;

based on the comparison, identifying a type of golf movement.

According to one embodiment, the method further comprises:

calculating a normalized error for each regularized normalized signal with respect to the model signal of the identified shot type and verifying that the normalized error is less than a threshold and that the sum of the normalized errors on all of the regularized normalized signals is less than a threshold;

depending on the comparison, validation or invalidation of the detection of a type of golf movement.

According to one embodiment, the method further comprises:

calculating an instability of the normalized error for each regularized normalized signal and verifying that the instability of the normalized error is less than a threshold for each regularized normalized signal and that the sum of the instabilities of the normalized errors on the set of regularized normalized signals is less than a threshold;

depending on the comparison, validation or invalidation of the detection of a type of golf movement.

According to one embodiment, the implementation of the impact detection function includes:

carrying out measurements with a sampling step greater than 100 Hz, for example equal to approximately 400 Hz or approximately 800 Hz plus or minus 10%, and storing the measurements in memory over the duration of the time window;

the creation of a temporal sub-window in the time window, centered on a peak having the greatest amplitude in each regularized normalized signal, the width of the sub-window having in particular a duration approximately equal to half the duration of the time window;

the comparison, on said temporal sub-window, between each regularized normalized signal and a sinusoidal signal of the same period and of the same amplitude as said peak on the basis of a modeling of the square error between said regularized normalized signal and said sinusoidal signal ;

the detection of a confirmed impact as soon as the squared error for each regularized normalized signal is less than a threshold.

According to one embodiment, the implementation of the impact detection function may include:

calculating a credibility score of the impact detection and verifying that the credibility score is greater than a threshold.

According to one embodiment, the implementation of the impact detection function includes:

detecting the exceeding of a peak in at least one of the normalized and regularized signals, by exceeding a threshold, then an opposite peak, in the same signal, by exceeding a threshold of opposite sign;

in this case, recording the measurements with a second sampling step greater than 100 Hz, for example equal to approximately 400 Hz plus or minus 10% or approximately 800 Hz plus or minus 10%, and storing the measurements in memory. measurements over the duration of a reduced time window comprising at least a portion of the normalized and regularized signal comprising the peak and the opposite peak;

the comparison, over said reduced time window, between the normalized and regularized signal and a sinusoidal signal of the same period and of the same amplitude as said peak and opposite peak.

According to one embodiment, the comparison, over said reduced time, between the at least one regularized standardized signal and the sinusoidal signal comprises:

modeling the square error between said regularized normalized signal and said sinusoidal signal and detecting a confirmed impact when the square error is less than a threshold; and or

the discrete Fourier series decomposition of the normalized and regularized signal and the detection of a confirmed impact as long as the duration during which the absolute value of the discrete Fourier series decomposition remains, for coefficients of interest of the decomposition, that is to say in particular for coefficients of the discrete Fourier series decomposition corresponding to the lowest non-zero frequency over the reduced time window, greater than or equal to a threshold equal to half the peak of this value absolute, is between a predefined minimum threshold and a predefined maximum threshold. In other words, we then verify that the absolute value of the discrete Fourier series decomposition, taking into account only the coefficient(s) corresponding to the frequency band sought for the detection of an impact (corresponding to the sinusoidal signal), remains greater than or equal to half the peak of this absolute value for a duration between a minimum threshold and a maximum threshold.

The invention also relates to a computer program type product, comprising at least one sequence of instructions stored and readable by a processor and which, once read by this processor, causes the steps of the method as presented previously to be carried out.

The invention further relates to a computer-readable medium comprising the computer program type product as presented previously.

PRESENTATION OF FIGURES

The invention will be better understood on reading the description which follows, given by way of example, and referring to the following figures, given by way of non-limiting examples, in which identical references are given to similar objects .

There is a graph representing signals from IMU sensors and filtered;

There is a graph representing a peak on a signal from an IMU sensor;

There represents the characterization of a peak in a signal from an IMU sensor;

There represents the normalization of signals from IMU sensors;

There shows graphs representing model envelopes for normalized signals;

There represents a zoom on a temporal sub-window applied to normalized and regularized signals, for impact detection;

There shows a comparison between a normalized and regularized signal, on a time sub-window, with an “expected” sinusoidal signal.

It should be noted that the figures explain the invention in detail to implement the invention, said figures can of course be used to better define the invention if necessary.

DETAILED DESCRIPTION OF THE INVENTION

The invention relates to a method for detecting golf movement implemented by means of sensors embedded in an electronic device, such as a bracelet or a watch, worn by the player, in particular on the wrist or possibly on another part of the arm. The onboard sensor means cooperate with calculation means, display means and a memory which can preferably also be embedded in the electronic device, in particular the watch, for better compactness.

Alternatively, at least part of the calculation and display means and at least part of the memory can be hosted by an external device, such as a smartphone, a tablet or a computer for example.

To detect a golf movement, the method according to the invention distinctly implements two functions, namely the detection of a golf movement, corresponding to a golf movement, and the detection of an impact, between the ball and the club, based, in particular exclusively, on data from sensors embedded in the electronic device worn by the player on the wrist or on another part of the arm, in particular the watch.

In the following, the present invention is more particularly described in the context of an implementation via a watch worn by the player. However, this is an example and the invention also applies in the case where the player wears, for example, a bracelet or any other similar electronic device, worn on the wrist or on another part of the arm.

The on-board sensors used include in particular IMU type sensors (for Inertial Measurement Unit), such as accelerometers and/or gyroscopes and/or magnetometers along one to three axes.

The measurements are carried out continuously at a frequency greater than 100 Hz and recorded with a first sampling step preferably greater than 10 Hz, in particular equal to 50 Hz plus or minus 10%, and over at least one reduced time window ( described below), with a second sampling step greater than 100 Hz.

For each function, the invention notably provides for the implementation of successive and progressive detection layers. Initially not very selective, the detection layers become more discriminating and make it possible to determine with certainty whether a real golf movement has been played.

In practice, the golf movement detection function thus includes the continuous recording of data measured by sensors embedded in the electronic device, in particular the watch, in particular an IMU sensor integrating for example three accelerometers and three gyroscopes along three axes respective orthogonal to each other.

According to one embodiment, the accelerometers make it possible to measure accelerations of the watch, and therefore of the user's wrist, along three mutually orthogonal axes. Gyroscopes make it possible to measure the rotation speeds of the watch, and therefore of the user's wrist, along three mutually orthogonal axes.

Thanks to the measurements carried out by these sensors, we obtain time signals formed from a plurality of acceleration or angular speed values as a function of time.

Notably, the temporal signals from these sensors are first filtered to filter out noise and high-frequency residuals. We then obtain filtered signals accx, accy, accz (coming from accelerometers) and gyrx, gyry, gyrz, coming from gyroscopes, as shown in the .

A different filter may be implemented for the purposes of implementing the golf motion detection function and, respectively, for the purposes of implementing the impact detection function.

For the golf movement detection function, for example, filtering of the temporal signals is implemented by a low-pass filter, in particular a first-order low-pass filter.

Below are detailed embodiments of the golf motion detection function.

As shown on the , for each signal, a peak is identified each time the amplitude of the signal exceeds a simple threshold, designated “POSSIBLE SWING” on the , in particular chosen as being “permissive”, that is to say not very discriminating, and a date of occurrence of the peak of greater amplitude is recorded in a memory, embedded in the watch or in a separate device. We thus detect a potential swing-type golf movement when the “POSSIBLE SWING” threshold is crossed. Below this, there is no movement (“NO SWING”) and there is no need to continue the analysis at this stage.

When a golf movement is executed, whether it is a real shot or a test shot, at least one peak is detected, generally two peaks, potentially three peaks or more. In any case, one to three peaks having the greatest amplitudes are kept in memory, with the date of occurrence of the peak of greatest amplitude, for at least one, preferably several, filtered temporal signals. According to the invention, as shown in the , we characterize these one to three peaks in terms of height, width and prominence.

According to one embodiment, each of the height, width and prominence of each peak is then compared to a respective threshold of minimum height, minimum width and minimum prominence so as to confirm or deny the existence of a peak corresponding.

In particular, the respective thresholds can be predefined empirically. For this first part of the method according to the invention, we choose simple thresholds, called “permissive”, that is to say not very discriminating, because it is important not to miss the detection of a golf movement. However, the following steps make it possible to sufficiently discriminate between “false positives”, in particular thanks to the impact detection function.

Once the detection of at least one peak has been confirmed, according to the invention, it is determined that the recorded temporal signals correspond to possible golf movements. Therefore, the time signals corresponding to the measurements carried out by the different IMU sensors are then preferably normalized, as shown in the . The standardization concerns in particular each of the signals delivered by the IMU sensors, corresponding to the measurements carried out, in acceleration, from measurements carried out by at least one accelerometer, and/or, in rotation speed, from measurements carried out by at least one a gyroscope.

Normalization consists in particular, according to one embodiment, of transposing the signal so that the maximum standard reached is equal to 1. The general principle of signal normalization is known.

• contracter ou étirer chaque signal normalisé, sur le plan temporel, de sorte qu’un signal normalisé ait, après régularisation, sur une fenêtre temporelle ayant une durée régularisée prédéfinie, par exemple égale à 2,5 secondes, un pic principal régularisé à une date de pic principal prédéfinie, par exemple à 1,5 secondes dans la fenêtre temporelle, et le cas échéant au moins un pic secondaire ayant une date de pic secondaire à une date prédéfinie dans la fenêtre temporelle, par exemple 0,67 seconde avant le pic principal avec un temps ayant subi l’homothétie temporelle. Cette homothétie donne un coefficient de contraction ou de dilation temporelle qui est le même pour l’ensemble des signaux normalisés.

The invention then preferably comprises the implementation of a regularization of each standardized signal, as shown in the . The regularization of normalized signals includes in particular the implementation of a temporal homothety consisting of:

• contract or stretch each normalized signal, temporally, so that a normalized signal has, after regularization, over a time window having a predefined regularized duration, for example equal to 2.5 seconds, a main peak regularized on a date predefined main peak, for example at 1.5 seconds in the time window, and where appropriate at least one secondary peak having a secondary peak date at a predefined date in the time window, for example 0.67 seconds before the peak main with a time having undergone temporal homothety. This homothety gives a temporal contraction or dilation coefficient which is the same for all the normalized signals.

In other words, the common temporal base of the normalized signals is translated and contracted or dilated so as to bring the main peak and the possible secondary peak to chosen reference abscissa. This makes it easier to carry out further calculations and analyzes by working with normalized and regularized signals.

Said expansion or contraction coefficient of the time base can be mathematically constrained by imposed limits and used separately to contribute to the evaluation of the credibility of the original signal.

Then, according to one embodiment, the implementation of the golf movement detection function comprises the execution of a golf movement recognition algorithm comprising, from a database of golf shots, in other words a database comprising a set of signals (accelerations and/or angular velocities on three axes) corresponding to trajectories of the electronic device during proven golf shots, the establishment of a vector μ and a matrix of covariance P of the set of regularized standardized signals, so that said covariance matrix P mathematically represents an ellipsoid from which to evaluate the probability that a candidate trajectory, in other words a possible golf movement, is actually a golf movement .

To establish the covariance matrix P, we can notably implement the following methods.

According to a first method, we consider a statistical vector variable X, of which each signal in the database is a realization, each temporal value of the signal being considered as a scalar variable distinct from to the product of the number of signals (typically 6 in the case of an IMU sensor allowing the measurement of acceleration and angular speed respectively along three axes) by the number of values forming said signals over the time window considered (typically a hundred time values recorded with the first sampling step). Each trajectory in the database is then considered as a realization of the statistical variable X, for which the mean μ and the covariance matrix P are then established from all the realizations.

According to a second method, we first carry out a subsampling of the temporal signals to reduce the dimensionality n of X, of μ and of the covariance matrix P, this reduction aiming to reduce the volume of calculations without losing representativeness. According to one implementation, a simple convolution (averaging or simple linear filtering) is applied to each temporal signal so that the subsampling is as least destructive as possible.

• on diagonalise la matrice de covariance P obtenue ;
• on seuille la valeur absolue des valeurs propres en remplaçant les valeurs jugées trop faibles (qui seraient anormalement discriminantes) par une valeur plus élevée ;
• on obtient alors la nouvelle matrice de covariance P modifiée.

From this covariance matrix P, we can, according to one implementation, make the subsequent calculations more robust by ensuring a minimum value for the eigenvalues of said covariance matrix P. To do this, according to this implementation:

• we diagonalize the covariance matrix P obtained;
• we threshold the absolute value of the eigenvalues by replacing the values judged to be too low (which would be abnormally discriminating) by a higher value;
• we then obtain the new modified covariance matrix P.

• Selon cette mise en œuvre, on calcule également une fois la matrice racine supérieure de Cholesky, désignée par S, de la matrice de covariance P, de telle sorte que S soit triangulaire supérieure et que S^TS=P. S est donc stockée en mémoire et calculée une fois et est fournie en entrée de l’algorithme décrit ci-après, ainsi que le vecteur μ déjà établi.
• Pour un possible mouvement de golf donné en entrée, on évalue comme précédemment la valeur spécifique de la variable vectorielle X associée.
• On calcule alors X_a=S^T\(X-μ), Xa étant donc le vecteur égal à la division à gauche du vecteur (X-μ) par la matrice S^T(S^Tétant triangulaire inférieure).
• Mathématiquement, X_areprésente la réalisation des tirages aléatoires de chaque variable gaussienne, centrée, réduite, indépendantes entre elles, qui seraient nécessaires pour obtenir le mouvement candidat s’il était issue de la distribution gaussienne multidimensionnelle N(μ, P) décrite par μ et P. Si le mouvement candidat respecte la distribution sous-jacente N(μ, P), alors notamment suit une loi du χ²à n degrés de libertés.
• On peut alors, selon cette mise en œuvre, évaluer le quantile correspondant au mouvement candidat en tant que réalisation de N(μ, P), à partir de la fonction de répartition de la loi du χ²à n degrés de libertés, donnée par ,

From the covariance matrix P established once and stored, we then seek to evaluate the probability for a possible golf movement, in other words a candidate trajectory, of being an effective golf movement. For that :

• According to this implementation, we also calculate once the upper root Cholesky matrix, designated by S, of the covariance matrix P, such that S is upper triangular and that S ^T S=P. S is therefore stored in memory and calculated once and is provided as input to the algorithm described below, as well as the vector μ already established.
• For a possible golf movement given as input, we evaluate as previously the specific value of the associated vector variable
• We then calculate X _a ⁼ S ^T \(X-μ) ^,
• Mathematically _, and P. If the candidate movement respects the underlying distribution N(μ, P), then notably follows a law of χ ² with n degrees of freedom.
• We can then, according to this implementation, evaluate the quantile corresponding to the candidate movement as a realization of N(μ, P), from the distribution function of the law of χ ² with n degrees of freedom, given by ,

where Γ is the gamma function and γ is the incomplete gamma function.

According to this implementation, the candidate movement is then considered as an effective golf movement if the corresponding quantile is less than a fixed threshold quantile.

According to another embodiment, the implementation of the golf movement detection function comprises the execution of a golf movement recognition algorithm comprising, from a database of golf shots, the comparison between each regularized normalized signal and a model signal corresponding to a typical golf movement movement for each type of golf movement including at least one drive, a swing and a putt, said comparison comprising verifying that the regularized normalized signal is found in a probabilistic envelope defined to include the possible variability of a real golf movement. This probabilistic envelope is determined from a database of golf movements and the application of selected appropriate margins.

For example, the golf movement database can be enriched by the collection of new data, particularly that of the user. In particular, according to one embodiment, the database differs between right-handed users and left-handed users and depending on whether the watch is worn on the right wrist or the left wrist.

Based on the comparison, a type of golf movement is identified.

The comparison can then implement a learning algorithm capable of improving the model signals used.

According to one embodiment, to validate the detection of golf movement, a normalized error is calculated for each normalized signal regularized relative to the model signal of the type of shot identified and the verification that the normalized error is less than a threshold and that the sum of the normalized errors over all the regularized normalized signals is less than a threshold.

For example, the normalized error is calculated according to the equation:

where the error is written:

The threshold can be predefined or adjustable.

In addition, it is possible to calculate an instability of the normalized error for each regularized normalized signal and verify that the instability of the normalized error is less than a threshold for each regularized normalized signal and that the sum of the instabilities of the errors normalized on all the regularized normalized signals is less than a threshold. This instability of the normalized error can for example be obtained by calculating the integral or the direct sum of the absolute value of the derivative of the normalized error.

For example, the instability of the normalized error is calculated according to the equation:

The threshold can be predefined or adjustable.

In the preceding equations, we note N the number of measurements present in the time window (for example 0 to 2.5 seconds) after application of temporal homothety.

µ is the average of the envelope for the signal of a normalized and regularized IMU sensor

σ is the standard deviation of the envelope for the normalized and regularized signal X.

According to one mode of implementation of the invention, to validate the detection of a golf movement, the conditions are:

for each normalized and regularized signal, Etot must be less than a given threshold;

optionally, for each normalized and regularized signal, Estab must be less than a given threshold;

optionally, the sum of the Etot for all the normalized and regularized signals must be less than a given threshold;

optionally, the sum of Estab for all the normalized and regularized signals must be less than a given threshold.

According to one embodiment, several of these conditions must be met to validate the detection of a golf movement. In particular, all of these conditions may have to be cumulatively satisfied.

According to one embodiment, several thresholds, in particular two thresholds – a low threshold and a high threshold – can be defined for each parameter. The validation of a golf movement is then acquired for example when we remain below two of the low thresholds for at least one parameter or at least two parameters when the other parameters are above the low thresholds while remaining below the thresholds Tops.

The examples of setting the conditions to validate the detection of a golf movement, given above, should not be interpreted restrictively, being only illustrative.

It should be noted that other signal comparison methods can be implemented. For example, we can define an error function always constructed from the database described previously. Reduced variables are then established, constructed as linear combinations of the individual temporal values of each normalized and regularized signal at each sampling step, the coefficients of these linear combinations being established so as to maximize the difference between the values taken by the reduced variable. between on the one hand a normalized and regularized signal corresponding to a golf movement to be detected and on the other hand a normalized and regularized signal not corresponding to a golf movement. Then, we apply these coefficients of linear combinations to the normalized and regularized signals to validate or invalidate the detection of a golf movement. This error function can also be applied to the impact detection function.

For the impact detection function, measurements recorded with a larger sampling step are analyzed. The sampling step is chosen, in particular greater than 100 Hz, for example equal to approximately 400 Hz plus or minus 10% (in particular 416 Hz) or to approximately 800 Hz plus or minus 10% (in particular 833 Hz). The measurements are stored in memory over the duration of a reduced time window.

Practically, according to the invention, the IMU sensor (preferably accelerometer and/or gyroscope and/or magnetometer) performs measurements with a high sampling step, greater than 100 Hz, in particular of the order of 400 Hz or of the order of 800 Hz. The measurements taken are recorded continuously with the first sampling step, greater than 10 Hz, and the golf movement detection function is implemented on these measurements recorded with the first sampling step. 'sampling.

The impact detection function is preferably implemented over a reduced time window on measurements recorded with the second sampling step, which is greater than 100 Hz, for example of the order of 400 Hz or order of 800 Hz.

To this end, just as for the motion detection function, we continuously apply checks on threshold criteria in signal amplitude at each time step and at high frequency, that is to say at the frequency of carrying out the measurements.

According to one embodiment, the amplitude threshold criterion is in particular applied to a derivative of the signal corresponding to the measurements carried out, which can be obtained by digital filtering, typically by means of a 1st order derivative filter, of the signal coming from the three-axis accelerometer. The time constant of the diverter filter is chosen particularly low (a few milliseconds) so as not to cut off the frequencies of interest, it being understood that an impact is detected over a reduced time window of a duration of between approximately 15 and 30 ms.

According to the invention, as shown in the , if the absolute value of this derivative daccz takes values greater than the predefined threshold, we check whether a change of sign and an exceeding of the threshold in the opposite direction occurs. If this is the case, we consider that an impact could have occurred and a detailed analysis of the signal is implemented over a time window reduced by a few tens of milliseconds, including the portion of the signal comprising the two opposite threshold crossings. as a sign.

It should be noted that this predefined threshold making it possible to identify a potential impact can be chosen dynamically, depending for example on the current amplitude on the acceleration or the rotation speed themselves. In this case, if the acceleration or rotation speed corresponding to the golf movement identified in parallel is high, then the threshold for identifying a potential impact will be higher, and vice versa.

The reduced time window dedicated to the detailed analysis of an impact lasts, for example, between 100 ms and 150 ms.

As shown on the , on this fine analysis window, according to one embodiment of the invention, we seek to verify whether the signal corresponding to the measurements recorded with the second sampling step is similar to a sinusoid having a period double the date gap between the two peaks of opposite sign mentioned above.

According to one embodiment, we first check whether the period thus defined is realistic, by comparing the period corresponding to twice the date difference between the two peaks with a predefined minimum period and with a maximum period. For example, the period should be between 15ms and 30ms.

As shown in , we can thus compare each regularized normalized signal to a sinusoidal signal, designated sinus_approx, whose period corresponds to twice the date difference between the peak and the opposite peak mentioned above.

For example, an error is then calculated, in particular a square error, between the signal corresponding to the measurements and a sinusoidal signal equal to a period. Preferably, the error is therefore calculated only over the time subinterval equal to this period, said sinusoid being signed and temporally placed so that its maximum coincides with the first peak identified.

modeling of the square error between said regularized normalized signal and said sinusoidal signal, and a confirmed impact is detected as soon as the square error for each regularized normalized signal is less than a threshold. For example the threshold for the squared error in this comparison is defined by the equation:

where N is the number of measurements in the normalized and regularized IMU sensor signal daccz (in this case an acceleration along the vertical z axis) present over the period.

Impact detection will then be validated if the squared error rmsd is less than a threshold. The threshold is for example predefined, configurable or adjustable.

If this error is less than a threshold, and, optionally, if its instability (as previously) is less than a threshold, then we confirm the detection of an impact.

It is obvious that, according to the invention, it is possible to exit the impact detection function when a condition is not satisfied.

According to another embodiment, a DFT (Discrete Fourier Transform) is carried out on a sliding time window on the high frequency signal, that is to say on the signal corresponding to the measurements carried out with the high sampling step of carrying out the measurements. Said sliding time window is chosen to be likely to correspond to a nominal sinusoid period as described above. This period is typically between 15 ms and 30 ms, for example equal to approximately 20 ms. The Fourier analysis can be limited to the coefficients of interest of the decomposition, that is to say those of the lowest possible non-zero frequency over such a sliding time window.

If the absolute value of the DFT exceeds a certain predefined threshold which, as before, can be adapted to the amplitude of the signal, in acceleration or rotation speed, then we continue the analysis. Otherwise, we can consider that there was no impact.

If we continue the analysis, we determine the duration during which this threshold is crossed. We then evaluate the maximum reached over this duration.

According to one embodiment, it is also possible to evaluate whether this absolute value of the DFT is consistent with the amplitudes of the acceleration and/or rotation speed signals as recorded with the first sampling step.

If this is the case, we can evaluate the duration during which the absolute value of the DFT remains greater than or equal to a threshold equal to half of the peak of this absolute value. We then determine whether this duration is between a predefined minimum threshold and a predefined maximum threshold.

If this is the case, we can confirm the impact detection. Otherwise, we can consider that there was no impact.

According to the invention, it should be noted that the criteria and verifications listed above can be implemented cumulatively. Alternatively, only some of these criteria and checks are implemented.

Ultimately, we detect, or not, the existence of an impact between the ball and the golf club.

According to a particular embodiment, “permissive” thresholds are defined to allow more potential golf movements and more potential impacts to be analyzed, so as to avoid not detecting some of them. Therefore, to avoid the opposite disadvantage which would be to detect “false positives”, we calculate for each crossing of a threshold a probability associated with the proximity or not observed of these thresholds. Thus, a value included and far from the thresholds is associated with a high probability, while a value included but close to a threshold is associated with a low (or very low) probability. The final decision to detect a golf movement or an impact is then based on the total probability obtained by aggregation (in other words by multiplication) of the different probabilities thus obtained.

Alternatively or in a complementary manner, the validation of the impact detection can include the calculation of error and error instability as described using the equations [Math. 1], [Math. 2] and [Math. 3].

According to one embodiment, the confirmed detection of an impact also includes the verification of consistency between the movement detection and the impact detection. In particular, the movement and the impact must for example be detected substantially simultaneously, typically in a narrow time window, in particular a few tenths of a second, for example 400 ms. Thus, an impact which would be detected for example one second after a movement would be eliminated and would not make it possible to identify a real golf movement.

Furthermore, for example, the amplitude of the temporal signals of the IMU sensors must be consistent between the detected movement and the detected impact.

Claims (12)

Golf movement detection method configured to operate on an electronic device, worn on the wrist or on another part of the arm, such as a bracelet or a watch, comprising at least one or more angular or linear measuring means, in particular in orientation, in speed or acceleration, such as a three-axis accelerometer and/or a three-axis gyroscope and/or a three-axis magnetometer, cooperating with a calculation unit and a memory space; a golf movement being defined by a golf movement made by a user handling a golf club, the club having a face, and causing said club to describe a typical trajectory resulting in an impact between the face of the golf club and a ball golf, the method comprising implementing a golf motion detection function and an impact detection function, the motion detection function being configured to, at any time, detect the completion of a golf motion golf associated with one of a set of predefined golf shots, including a drive, a swing and a putt, and the impact detection function is configured to synchronously detect an impact between the golf club and the golf ball, from measurements (accx, accy, accz, gyrx, gyry, gyrz) carried out by at least sensor embedded in the electronic device. Method according to claim 1, in which the golf movement detection function and the impact detection function are implemented solely from measurements (accx, accy, accz, gyrx, gyry, gyrz) carried out by at least a sensor embedded in the electronic device, excluding sensors embedded by the club or the ball, among at least one three-axis accelerometer and at least one three-axis gyroscope. Method according to one of the preceding claims, in which the measurements (accx, accy, accz, gyrx, gyry, gyrz) are carried out continuously and are recorded at a first sampling step at least equal to 10 Hz, in particular equal to 50 Hz plus or minus 10%, and at a second sampling step at least equal to 100 Hz for measurements carried out over at least one reduced time window. Method according to one of the preceding claims, in which the implementation of the golf movement detection function comprises:

• the application of a low-pass filter to the signals corresponding to the measurements (accx, accy, accz, gyrx, gyry, gyrz) carried out, in particular a first-order low-pass filter;
• the normalization of each of the signals corresponding to the measurements carried out, in particular in acceleration, from measurements carried out by the at least one accelerometer, and/or in rotation speed, from measurements carried out by the at least one gyroscope;
• comparing at least one of the normalized signals to respective simple thresholds;
• the identification, for each normalized signal, of at least one peak, when the amplitude of the normalized signal exceeds a simple threshold;
• the recording, for each normalized signal, of a date of occurrence of the peak of greatest amplitude.

Method according to one of claims 1 to 3, in which the implementation of the golf movement detection function comprises, only if at least one signal corresponding to a measurement carried out has an amplitude greater than or equal to a threshold of predefined amplitude,

• the application of a low-pass filter to the signals corresponding to the measurements (accx, accy, accz, gyrx, gyry, gyrz) carried out, in particular a first-order low-pass filter;
• the normalization of each of the signals corresponding to the measurements carried out, in particular in acceleration, from measurements (accx, accy, accz, gyrx, gyry, gyrz) carried out by the at least one accelerometer, and/or in rotation speed, at from measurements carried out by the at least one gyroscope;
• comparing the normalized signals to respective simple thresholds;
• the identification, for each normalized signal, of at least one peak, when the amplitude of the normalized signal exceeds a simple threshold;
• the recording, for each normalized signal, of a date of occurrence of the peak of greatest amplitude.

Method according to claim 4 or 5, further comprising:

• if at least one peak has been identified, the characterization, for each normalized signal, of 1 to 3 peaks having the greatest amplitudes in height and/or width and/or prominence;
• the comparison, for each standardized signal, of its height and/or its width and/or its prominence to a respective threshold of minimum width, minimum height and/or minimum prominence so as to confirm or refute the existence of 'a corresponding peak;
• the regularization of each normalized signal comprising at least one confirmed peak, by means of a temporal homothety consisting of:
  • contract or stretch each normalized signal, temporally, so that each normalized signal has, over a reduced time window having a predefined regularized duration, for example equal to 2.5 seconds, a main peak regularized at a main peak date predefined, for example at 1.5 seconds in the time window, and where appropriate at least one secondary peak having a secondary peak date at a predefined date in the time window, for example 0.67 seconds before the main peak with a time having undergone temporal homothety.

Method according to claim 6, further comprising:

• the regularization of each normalized signal by means of an amplitude homothety consisting of:
  • contract or stretch each normalized signal, in amplitude, so that the normalized signal has, over the time window, an amplitude of the main peak equal to a predefined value.

A method according to claim 6 or 7, wherein implementing the golf motion detection function comprises executing a golf motion recognition algorithm comprising, from a golf shot database golf:

• the comparison between each regularized normalized signal and a set of model signals corresponding respectively to an average golf movement movement for a set of golf movement types, in particular a drive, a swing and a putt, said comparison comprising the verification that the regularized normalized signal is in a defined envelope between each model signal shifted by a negative offset and the same model signal shifted by a positive offset;
• based on the comparison, identifying a type of golf movement.

Method according to claim 8, further comprising:

• calculating a normalized error for each normalized signal regularized with respect to the model signal of the type of golf movement identified and verifying that the normalized error is less than a threshold and that the sum of the normalized errors over all the signals regularized normalized values is less than a threshold;
• depending on the comparison, validation or invalidation of the detection of a type of golf movement.

Method according to claim 9, further comprising:

• calculating an instability of the normalized error for each regularized normalized signal and verifying that the instability of the normalized error is less than a threshold for each regularized normalized signal and that the sum of the instabilities of the normalized errors on the set of regularized normalized signals is less than a threshold;
• depending on the comparison, validation or invalidation of the detection of a type of golf movement.

Method according to one of claims 6 to 10, in which the implementation of the impact detection function comprises:

• detecting the exceeding of a peak in at least one of the normalized and regularized signals, by exceeding a threshold, then an opposite peak, in the same signal, by exceeding a threshold of opposite sign;
• in this case, recording the measurements with a second sampling step greater than 100 Hz, for example equal to approximately 400 Hz plus or minus 10% or approximately 800 Hz plus or minus 10%, and storing the measurements in memory. measurements over the duration of a reduced time window comprising at least a portion of the normalized and regularized signal comprising the peak and the opposite peak;
• the comparison, over said reduced time window, between the normalized and regularized signal and a sinusoidal signal (sinus_approx) of the same period and of the same amplitude as said peak and opposite peak.

Method according to claim 11, in which the comparison, over said reduced time, between the at least one regularized normalized signal and the sinusoidal signal comprises:

• modeling the square error between said regularized normalized signal and said sinusoidal signal (sinus_approx) and detecting a confirmed impact as soon as the square error is less than a threshold; and or
• the discrete Fourier series decomposition of the normalized and regularized signal and the detection of a confirmed impact as long as the duration during which the absolute value of the discrete Fourier series decomposition remains, for coefficients of interest of the decomposition into discrete Fourier series, greater than or equal to a threshold equal to half the peak of this absolute value is between a predefined minimum threshold and a predefined maximum threshold.