FIELD OF THE INVENTION

The present invention relates to a motivation enhancing exercise system comprising an exercise device and a head mounted output device, as well as to a method for operating the motivation enhancing exercise system. The system and method according to the invention provides a reward to an exercising user, in the form of an output via the head mounted output device, in accordance with an interaction between the user and the exercise device.

BACKGROUND OF THE INVENTION

When exercising, persons often need encouragement and motivation in order to continue performing the exercise, in terms of performing more repeats within a session, as well as in terms of returning to the exercise equipment, e.g. the following day or the following week. This is relevant for healthy persons working to build or maintain their shape, but even more so for elderly or rehabilitating persons, e.g. geriatric patients.

In order to encourage and motivate users of exercising equipment, various approaches for rewarding and/or motivating the exercising user have been attempted. For instance, US 9,629,558 B2 discloses a portable activity monitoring device comprising a housing having a physical size and shape that is adapted to couple to a user's body. A plurality of sensors is disposed in the housing. Processing circuitry, disposed in the housing and electrically coupled to the plurality of sensors, calculates activity points corresponding to the physical activity of the user using the sensor data, wherein the activity points correlate to an amount and intensity of the physical activity of the user.

DESCRIPTION OF THE INVENTION

It is an object of embodiments of the invention to provide a method for operating a motivation enhancing exercise system, in which an activity of an exercising user can be accurately determined and rewarded.

It is a further object of embodiments of the invention to provide a method for operating a motivation enhancing exercise system, in which a customized reward is provided to an exercising user, reflecting an activity of the user. It is an even further object of embodiments of the invention to provide a motivation enhancing exercise system in which an activity of an exercising user can be accurately determined and rewarded.

According to a first aspect the invention provides a method for operating a motivation enhancing exercise system, the exercise system comprising an exercise device configured to be driven by an exercising user; a first sensor unit mounted on a movable part of the exercise device, the first sensor unit being configured to detect movements of the movable part of the exercise device; a head mounted output device configured to output a visual and/or an audible output for an exercising user, the head mounted output device further comprising a second sensor unit configured to detect movements of a head of the exercising user; and a processing unit being communicatively connected to the first sensor unit and to the head mounted output device, the method comprising the steps of:

- measuring movements of the movable part of the exercise device by means of the first sensor unit, in response to an exercising user operating the exercise device,

- measuring movements of a head of the exercising user, by means of the second sensor unit, while the head mounted output device is mounted on the head of the exercising user, and while the exercising user operates the exercise device,

- providing the measurements performed by the first sensor unit and by the second sensor unit to the processing unit,

- the processing unit determining an activity being performed by the exercising user, based on the measurements provided from the first sensor unit and from the second sensor unit,

- the processing unit generating an output signal for the head mounted output device, in accordance with the determined activity, and providing the output signal to the head mounted output device, and

- the head mounted output device providing a visual and/or audible output for the exercising user in accordance with the output signal.

Thus, according to the first aspect, the invention provides a method for operating a motivation enhancing exercise system. In the present context the term 'motivation enhancing' should be interpreted to mean that the method and the system encourages and motivates a user to continue exercising in an appropriate and desired manner. For instance, the method and the system may encourage and motivate a user to follow a specified training or exercise program, e.g. a rehabilitation program or a shape enhancing or maintaining program.

The motivation enhancing exercise system comprises an exercise device, a first sensor unit, a head mounted output device, and a processing unit.

The exercise device is configured to be driven by an exercising user. Thus, when exercising, the user interacts with the exercise device, and thereby performs a related activity. The exercise device could, e.g., be a bicycle, a bicycle trainer, a treadmill, a rowing machine, a pulley system, or any other suitable kind of exercise device.

The first sensor unit is mounted on a movable part of the exercise device, and it is configured to detect movements of the movable part. The movable part is a part of the exercise device which moves in a specific manner when a user interacts with the exercise device during exercising. Thus, when the user performs a specific exercising activity by interacting with the exercise device in a well defined manner, thereby causing the movable part to move in a manner which is also well defined, this movement is detected by the first sensor unit.

In the case that the exercise device is a bicycle or a bicycle trainer, the movable part may, e.g., be a pedal, a pedal arm, a crank, or any other suitable part which moves when a user interacts with the bicycle or bicycle trainer during exercise.

In the case that the exercise device is a treadmill, the movable part may, e.g., be an endless band which support a user's feet during walking or running exercise.

In the case that the exercise device is a rowing machine, the movable part may, e.g., be an arm, a handle, etc., which a user pulls during rowing exercise.

In the case that the exercise device is a pulley system, the movable part may, e.g., be weight or a wire which is lifted by a user performing workout by means of the pulley system.

The first sensor unit may be fixedly connected to the movable part of the exercise device. Alternatively, the first sensor unit may be releasably attached to the movable part of the exercise device, thereby allowing the first sensor unit to be detached from the movable part and moved to a movable part of another exercise device. Thereby first sensor unit may be applied for tracking exercise performed by a specific user who applies multiple kinds of exercise devices. Furthermore, a releasably attached sensor unit may be replaced by another similar sensor unit being associated with another user, in order to allow other users to use the exercise device in a similar manner.

The head mounted output device is configured to output a visual and/or an audible output for an exercising user. In the present context the term 'head mounted output device' should be interpreted to mean that the output device is configured to be mounted on the head of the user during exercise. The visual and/or audible output can thereby be provided directly to the eyes and/or ears of the exercising user. This will be described in further detail below.

The head mounted output device further comprises a second sensor unit configured to detect movements of a head of the exercising user. Thus, apart from detecting movements of the movable part of the exercise device, by means of the first sensor unit, the system is also capable of detecting concurrently occurring movements of the head of the exercising user, by means of the second sensor unit. The second sensor unit may, e.g., be mounted on or form an integrated part of the head mounted output device. Since the head mounted output device is mounted directly on the head of the user during exercise, the second sensor unit naturally moves along with the head of the user, and is therefore readily capable of detecting such movements.

The second sensor unit may, e.g., be integrated in a textile object, which is mounted on the head mounted output device or arranged between the head mounted output device and the head of the user. For instance, the second sensor unit may be a printed sensor.

The processing unit is communicatively connected to the first sensor unit and to the head mounted output device, including the second sensor unit. Accordingly, measurement data collected by the first sensor unit and by the second sensor unit can be communicated to the processing unit. Accordingly, the processing unit is in the possession of correlated data related to movements of the movable part of the exercise device and movements of the head of the user, obtained while the user performs exercise by interacting with the exercise device. Furthermore, the processing unit may communicate with the head mounted output device, in order to control the output of the head mounted output device. This will be described in further detail below.

In the method according to the first aspect of the invention, an exercising user operates the exercise device by interacting therewith in a specified manner, while the user wears the head mounted output device, i.e. the head mounted output device is mounted on the head of the user. As described above, this causes the movable part of exercise device to move. During this, the movements of the movable part of the exercise device, in response to the exercising user operating the exercise device, are measured by means of first sensor unit. Furthermore, movements of the head of the exercising user are measured by means of the second sensor unit, which is mounted on or forms part of the head mounted output device.

The measurements performed by the first sensor unit and by the second sensor unit are provided to the processing unit. Thus, the processing unit is in the possession of concurrent data related to movements of the movable part of the exercise device and movements of the head of the exercising user, obtained while the user exercises by operating the exercise device.

It is not ruled out that part of the processing of the measurement data is performed at one or both of the sensor units. In this case, the data provided to the processing unit may, e.g., be in the form of pre-processed data. Alternatively or additionally, part of the actual processing may be performed at the sensor unit(s), thereby reducing the processing work required at the processing unit. For instance, the sensor unit(s) may even be able to provide an estimated guess on the performed activity. Furthermore, data may only be provided to the processing unit when the sensor unit in question detects a change in the measured movements. This reduces the amount of data which it is required to transfer from the sensor unit to the processing unit. The processing unit can simply assume that the movements detected by the sensor unit when no data is transferred remain constant, and thereby the reduced data transfer will not affect the accuracy of the output from the processing unit.

The processing unit then determines an activity being performed by the exercising user, based on the measurements provided from the first sensor unit and from the second sensor unit. As described above, the activity performed by the user depends on the kind of the exercise device. Flowever, it also depends on how the user uses or interacts with the exercise device. Thus, the activity being determined may not merely be 'cycling', but could also reflect wattage, rotational speed of the crank, duration and/or any other suitable parameter which is relevant with regard to evaluating a training impact, including whether or not the exercising user follows a prescribed training program.

Since the activity is determined based on measurements provided from the first sensor unit and from the second sensor unit, the activity is, thus, determined based on information regarding movements of the movable part of the exercise device, as well as on information regarding movements of the head of the user. Accordingly, correlated information regarding the interaction between the exercising user and the exercise device, in the form of the movements of the movable part of the exercise device, and regarding the movements of the body of the exercising user, in the form of the movements of the head of the user, is applied as a basis for determining the activity being performed by the exercising user. This allows the activity to be determined more accurately, faster and with less data from each sensor unit than would be the case if the activity was determined based on a single sensor input.

The processing unit then generates an output signal for the head mounted output device, in accordance with the determined activity, and provides the output signal to the head mounted output device. Finally, the head mounted output device provides a visual and/or audible output for the exercising user in accordance with the output signal.

Since the output signal is generated in accordance with the determined activity, the activity being performed by the exercising user is reflected in the output signal, and thereby in the visual and/or audible output which is provided to the exercising user, via the head mounted output device. For instance, the output may show the activity in the sense that if the activity is cycling with high wattage, the output may be a visual output indicating that the exercising user rides a bicycle uphill.

Alternatively or additionally, the visual and/or audible output may be a reward to the exercising user, e.g. depending on whether or not, and to which extent, the exercising user is following a prescribed training program. For instance, a geriatric patient operating a cycling trainer may be rewarded by giving the patient the impression that he or she is walking in a forest. Providing such a reward to a user may motivate the user to continue exercising, and the user may not even notice that he or she is operating a cycling trainer rather that walking in a forest.

Since the activity is determined accurately and fast, as described above, a correct reward is also provided to the exercising user fast, thereby enhancing the motivation of the exercising user to perform the exercise.

The first sensor unit may comprise a three-dimensional accelerometer and a gyroscope, and the step of measuring movements of the movable part of the exercise device by means of the first sensor unit may comprise measuring three-dimensional accelerations and gravitational orientation of the movable part of the exercise device.

According to this embodiment, the first sensor unit registers accelerations as well as orientation of the movable part of the exercise device. This allows the movements of the movable part to be determined very accurately.

Alternatively or additionally, the first sensor unit and/or the second sensor unit may comprise a magnetometer, e.g. a three-dimensional magnetometer. In a similar manner, the second sensor unit may measure accelerations and/or orientation of the head mounted output devices, and thereby of the head of the exercising user, e.g. using a three-dimensional accelerometer and a gyroscope.

The first sensor unit and the second sensor unit may be identical or of the same kind. As an alternative, the first sensor unit may be of a different kind than the second sensor unit.

The step of providing the measurements from the first sensor unit to the processing unit may be performed by means of a wireless communication channel. Thereby the first sensor unit need not be communicatively connected to the processing unit by means of a wire, and therefore first sensor unit and/or the processing unit can be arranged at a position which does not readily allow such a wired connection. Furthermore, it is ensured that the interaction between the exercising user and the exercise device is not inhibited by the presence of such a wire.

The wireless communication channel may apply any suitable kind of wireless communication technology, including Bluetooth, WiFi, long range (LoRa) wireless communication, near-field communication (NFC), infrared communication, etc.

As an alternative, the communication channel between the first sensor unit and the processing unit may be hardwired.

The processing unit may form part of the head mounted output device, and the step of providing measurements from the second sensor unit to the processing unit may be performed by direct communication. For instance, the second sensor unit may form part of the processing unit, in which case the measurements performed by the second sensor unit are readily available for the processing unit. According to this embodiment, the first sensor unit communicates with the head mounted output device when providing its measurements to the processing unit. Furthermore, the processing unit may, according to this embodiment, control the head mounted output device directly, e.g. by directly generating the visual and/or audible output for the exercising user.

As an alternative, the processing unit may form part of the first sensor unit, or it may be a separate part. In this case the head mounted output device, including the second sensor unit, may also communicate with the processing unit in a wireless or wired manner, as described above with reference to the first sensor unit.

The step of the processing unit determining an activity may comprise the processing unit determining a type of the exercise device. For instance, the processing unit may determine whether the exercise device is a bicycle, a cycling trainer, a treadmill, and/or any other suitable kind of exercise device, as described above. Once the type of the exercise device has been determined, the number of possible activities is narrowed down to activities which can be performed by means of this type of exercise device.

The type of the exercise device may be determined purely based on measurements performed by means of the first sensor device and the second sensor device. Alternatively or additionally, the determination of the type of the exercise device may be at least partly based on additional input. For instance, the exercising user may enter the type of the exercise device before initiating the exercise, e.g. by selecting one of a number of predefined options.

As an alternative, the first sensor unit may read a machine readable code arranged on the exercise device, and indicating the type of the exercise device, when the first sensor unit is mounted on the exercise device. The first sensor unit may then provide this information to the processing unit along with the measurements performed by the first sensor unit. In this case the first sensor unit may be personal, in the sense that a given user uses the same first sensor unit for various types of exercise devices, but the first sensor is only applied for exercise sessions of that user.

As another alternative, the first sensor unit may be permanently attached to a specific exercise device, and the first sensor unit may provide information regarding the type of the exercise device when the first sensor unit connects to a given, e.g. user specific, processing unit when an exercising user initiates an exercise session.

The step of the processing unit determining an activity may comprise the processing unit comparing the received measurements to expected movement patterns for one or more predefined activities. According to this embodiment, a number of movement patterns for the first sensor unit and for the second sensor unit have previously been obtained, and are therefore available for the processing unit when determining the activity being performed by the exercising user. The movement patterns are each representative for expected movements of the sensor units, during various activities. For instance, the movement patterns may be obtained empirically, e.g. while allowing one or more users to perform a specific activity while measuring the relevant movements by means of the first sensor unit and the second sensor unit. Alternatively or additionally, the movement patterns may be obtained by means of simulations and/or calculations. The movement patterns could, e.g., include frequency of the movements and/or of repetitions performed by the user.

Comparing the measurements obtained from the first and second sensor units to expected movement patterns, as described above, provides an easy, fast and reliable manner of determining the activity being performed by the exercising user. The comparison could, e.g., include simple comparison between constituent parts of the sensor measurements to corresponding constituent parts of the movement patterns, for instance accelerations along various dimensions (x, y and z axes), gravitational orientation, etc. Alternatively or additionally, a more complete comparison between the measurements and the movement patterns may be applied.

The expected movement patterns may be generated by means of machine learning. According to this embodiment, machine learning or artificial intelligence (AI) is applied in order to identify movement patterns in measurements originating from a vast number of previous exercise sessions. In this case the movement patterns may be continuously improved and updated as more data is made available from exercising users using the system.

The data used for training the model may originate from only one user, or it may originate from several users using the same or similar equipment.

Furthermore, the comparison between the sensor measurements and the expected movement patterns may also rely on machine learning or AI techniques.

According to one embodiment, the step of determining an activity being performed by the exercising user may be performed in the following manner.

Initially a machine learning or AI model is trained by obtaining measurement data from sensors units mounted on exercising devices and head mounted output devices, under controlled conditions. For instance, this could include allowing a given user to perform a series of well defined exercising activities while obtaining measurements from a first sensor unit mounted on a movable part of the exercise device and from a second sensor unit mounted on or forming part of a head mounted output device mounted on the head of the exercising user. This may be repeated for a number of further user. The obtained data is provided with information regarding the exercising activity being performed.

A series of exercising activities could, e.g., include varying manner and intensity of the exercise and/or varying position, orientation, etc. of the first sensor unit. For instance, the user may be requested to operate a bicycle trainer at a specified speed and/or with a specified load, while the first sensor unit is positioned with a specific orientation at the crank of the bicycle trainer. Measurements may be repeated at a number of further speeds and/or loads, for instance by increasing the speed and/or the load at specified times. All of this may then be repeated with the first sensor unit arranged on the pedal arm, with a different orientation, etc. Further positions on movable parts of the bicycle trainer may be selected, and similar measurements obtained in the manner described above. Finally, the entire process may be repeated by one or more further users.

Corresponding measurements are also obtained with respect to other bicycle trainers, and with respect to other kinds of exercising devices.

The obtained measurement data, including the applied information regarding the performed exercising activities, are then supplied to a neural network or similar machine learning system, e.g. after removing possible outliers from the data. The neural network then processes the received data and identifies patterns which are characteristic for respective performed exercising activities. This could, e.g., include applying a deep learning algorithm. The patterns could, e.g., relate to combinations of accelerations in various direction, orientation measured by means of gyroscopes, etc.

When a user is subsequently operating an exercising device, while measuring movements by means of a first sensor unit and a second sensor unit, in the manner described above, the measurements performed by the sensor units are provided to the neural network, where patterns in the measured data are compared to the patterns which were identified during the training of the machine learning model, e.g. by tweaking parameters in order to identify a best fit. Based thereon, the neural network may return, for each of the initially defined exercising activities, a likelihood that the currently performed activity is that activity. It may then be determined that the activity with the highest likelihood, i.e. the one providing the best match, is the one being performed. The first sensor unit and/or the second sensor unit may form part of the neural network. It should be noted that the activity being determined is not merely the kind of exercise being performed or the kind of exercise equipment being used, but also includes how the exercising is being performed, such as intensity, speed, etc.

In the case that one of the initially defined activities has a likelihood which significantly higher than the other activities, then the activity being performed by the user can be determined with high confidence. Flowever, if two or more activities appear to be equally or almost equally likely, then there is a risk that the activity being performed by the user is not determined correctly. In such cases, more data may be obtained before the activity is determined, or the user may be requested to identify the activity.

The data obtained during use may be provided to the neural network in order to continuously improve the machine learning model. In particular, information regarding wrong or low confident determination of activities is very useful with regard to improving the machine learning model. Data relating to a vast number of users, a vast number of exercising devices, a vast number of sessions, etc. may advantageously be collected at a central site, in order to ensure that the trained model becomes as accurate as possible.

The head mounted output device may be a virtual reality (VR) or an augmented reality (AR) device, and the step of the head mounted output device providing an output may comprise providing a visual experience output for the exercising user.

In the present context the term 'virtual reality (VR) device' should be interpreted to mean a device comprising a head mounted part which covers the eyes of the user, and which presents a visual information to the user which differs from the real environment in which user is positioned. However, the information presented to the user depends on actions and movements performed by the user. VR devices may be used for allowing a user to simulate various situations.

In the present context the term 'augmented reality (AR) device' should be interpreted to mean a device which is similar to a VR device. However, in AR devices, the real environment in which the user is positioned is presented to the user, but the environment is augmented by virtual features.

VR or AR devices are, thus, very suitable for generating a rewarding visual output for the exercising user, because it is possible to give the user the impression that he or she is in a different location and/or performing another activity than what is actually the case. For instance, a weak patient who is bound to a hospital while recovering may be given the impression that he or she is walking in a forest.

Alternatively or additionally, the head mounted output device may comprise headphones capable of providing an audible output. In this case the reward could, e.g., be playing a selected piece of music or a selected audiobook. Or the audible output may supplement a visual output, e.g. in order to enhance the impression that the user is in a different location.

According to a second aspect the invention provides a motivation enhancing exercise system comprising:

- an exercise device configured to be driven by an exercising user,

- a first sensor unit mounted on a movable part of the exercise device, the first sensor unit being configured to detect movements of the movable part of the exercise device, - a head mounted output device configured to output a visual and/or an audible output for an exercising user, the head mounted output device further comprising a second sensor unit configured to detect movements of a head of the exercising user, and

- a processing unit connected to the first sensor unit and the head mounted output device, wherein the motivation enhancing exercise system is configured to perform the method according to the first aspect of the invention.

Thus, the motivation enhancing exercise system according to the second aspect of the invention can be operated in accordance with the method according to the first aspect of the invention. The remarks set forth above with reference to the first aspect of the invention are therefore equally applicable here.

Thus, the first sensor unit may comprise a three-dimensional accelerometer and a gyroscope, and/or the processing unit may form part of the head mounted output device, as described above.

Furthermore, as described above, the head mounted output device may be a virtual reality (VR) or an augmented reality (AR) device, and/or it may comprise headphones.

BRIEF DESCRIPTION OF THE DRAWINGS

The invention will now be described in further detail with reference to the accompanying drawings in which

Fig. 1 illustrates two different types of head mounted output devices for a motivation enhancing exercise system according to embodiments of the invention,

Figs. 2-4 illustrate sensor units mounted on three different types of exercise devices,

Fig. 5 is a flow chart illustrating a method according to a first embodiment of the invention,

Fig. 6 is a flow chart illustrating a method according to a second embodiment of the invention, Fig. 7 illustrates determining an activity performed by an exercising user in accordance with a method according to an embodiment of the invention, using a neural network, and

Fig. 8 illustrates a single node of the neural network of Fig. 7.

DETAILED DESCRIPTION OF THE DRAWINGS

Fig. 1 illustrates part of a motivation enhancing exercise system 1 according to an embodiment of the invention. More particularly, Fig. 1 illustrates head mounted output devices in the form of a virtual reality (VR) or augmented reality (AG) device 2 for providing a visual output for a user, and a set of headphones 3 for providing an audible output for the user. An audible output may also be provided via earplugs 4 forming part of the VR or AR device 2. A user using the system 1 may choose whether to apply the VR or AR device 2, thereby obtaining visual as well as audible output, or to apply the headphones 3, thereby obtaining audible output only.

The system 1 further comprises a processing unit (not shown) which is arranged to communicate with the VR or AR device 2, with the headphones 3 and with a sensor unit (not shown) which is mounted on a movable part of an exercise device (not shown). The processing unit may form part of the VR or AR device 2 and/or the headphones 3.

During use, a user interacts with an exercise device, thereby causing a movable part of the exercise device to move. This will be described in further detail below with reference to Figs. 2-4. A first sensor unit mounted on the movable part of the exercise device measures these movements and communicates the measurements to the processing unit via a wireless communication channel.

The user further has the VR or AR device 2 or the headphones 3 mounted on the head. A second sensor unit (not shown) mounted on or forming part of the VR or AR device 2 or the headphones 3 measures movements of the head of the user. These measurements are also communicated to the processing unit via a wireless communication channel, via a wired connection, or provided directly to the processing unit in the case that the processing unit forms part of the respective output device 2, 3.

Based on the received measurements, the processing unit determines an activity being performed by the exercising user. Thus, the activity is determined based on the interaction between the exercising user and the exercise device, as well as based on movements of the head of the user. As described above, the determination of the activity is accurate and fast, since it is performed based on measurements performed by the first sensor unit as well as on measurements performed by the second sensor unit.

The processing unit further generates an output signal for the chosen head mounted output device 2, 3, in accordance with the determined activity, and provides the output signal to the output device 2, 3. The output device 2, 3 finally provides a visual and/or audible output for the user in accordance with the output signal. Thus, the exercising user receives a reward, in the form of a specific visual and/or audible output, which reflects the activity which the user is performing.

Fig. 2 illustrates a first sensor unit 7 mounted on a movable part, in the form of a pedal arm 8, of an exercise device, in the form of a bicycle. When an exercising user interacts with the bicycle, the pedal arm 8 performs a rotating movement. This movement is measured by the first sensor unit 7, and the first sensor unit 7 provides these measurements to a processing unit (not shown), via a wireless communication channel. The processing unit then applies the measurements in order to determine the activity being performed by the user, in the manner described above.

Fig. 3 illustrates a first sensor unit 7 mounted on a movable part, in the form of a movable arm 9, of an exercise device, in the form of a rowing machine 10. Similarly to the embodiment described above with reference to Fig. 2, when an exercising user interacts with the rowing machine 10 by performing rowing movements, the movable arm 9 moves back and forth, and the first sensor unit 7 measures this movement and communicates the measurements to a processing unit, in order to allow the processing unit to determine the activity being performed by the user.

Fig. 4 illustrates a first sensor unit 7 mounted on a movable part, in the form of a cable 11, of an exercise device, in the form of a pulley system 12 for weight lifting workout. Also in this embodiment, when an exercising user interacts with the pulley system 12 by lifting the weights 13, the cable 11 moves up and down, and the first sensor unit 7 measures this movement and communicates the measurements to a processing unit, in order to allow the processing unit to determine the activity being performed by the user.

Fig. 5 is a flow chart illustrating a method according to a first embodiment of the invention.

At step 14 measurements from a first sensor unit, mounted on a movable part of an exercise device, are received at a processing unit. At step 15 it is investigated whether or not an exercising user has manually entered the activity being performed by the exercising user. If this is the case, the process is forwarded to step 16, where the processing unit determines the activity being performed as the activity which was manually entered by the exercising user, and a corresponding output signal for an output device is generated.

If step 15 reveals that the exercising user has not manually entered the activity, the process is forwarded to step 17, where the data received from the first sensor unit is processed, and at step 18 a preliminary determination of the activity being performed by the exercising user is obtained, based on the processed data from the first sensor unit.

At step 19 it is investigated whether or not measurement data is available from a second sensor unit forming part of a head mounted output device being worn by the exercising user. If this is the case, the process is forwarded to step 20, where the measurement data from the second sensor unit is processed and combined with the processed data from the first sensor unit, and at step 21 the processing unit determines the activity being performed by the exercising user, based on the combined processed data from the first sensor unit and the second sensor unit.

In the case that step 19 reveals that no measurement data is available from the second sensor unit, the process is forwarded directly to step 21, and the activity being performed by the exercising user is determined solely on the basis of measurement data obtained by the first sensor unit, i.e. the determination is in line with the preliminary determination performed at step 18.

At step 22, an output signal for a head mounted output device is generated in accordance with the activity which was determined at step 21.

In the case that the system has access to a database 23 containing known movement patterns for a number of specified activities, then the process is forwarded from step 21 to step 24, where the processing unit derives a movement pattern from the available data. The derived movement pattern is compared to the stored movement patterns from the database 23, at step 25. At step 26 it is investigated whether or not the derived movement pattern matches one of the stored patterns. If this is the case, the process is forwarded to step 27, where the activity performed by the exercising user is determined as the activity defined by the movement pattern which the derived movement pattern matches. Furthermore, an output signal for a head mounted output device is generated in accordance herewith.

In the case that step 26 reveals that no match can be found between the derived movement pattern and any of the stored movement patterns, then the process is forwarded to step 22, and the activity is determined based on the determination performed at step 21. Fig. 6 is a flow chart illustrating a method according to a second embodiment of the invention. The process is started at step 28. At step 29 an exercising user is asked to start moving, i.e. to start interacting with an exercise device.

At step 30, a processing unit processes measurement data received from sensor units and determines an activity being performed by the exercising user, e.g. in the manner described above with reference to Fig. 5.

At step 31, at number of available experiences are presented to the user, and the user is requested to select one of them. In Fig. 5 five experiences 32 are shown. The experiences 32 represent various visual and/or audible outputs which can be provided to the exercising user as a reward for performing the activity. Accordingly, the available experiences 32 which are presented to the user at step 31, have been preselected from a larger group of experiences, based on the determined activity.

At step 33, the user may either select one of the presented experiences 32, or indicate that the activity has been incorrectly determined. In the case that the user selects one of the presented experiences 32, this is also a confirmation that the activity was determined correctly. In this case, the process is forwarded to step 34, where it is communicated to an artificial intelligence (AI) or machine learning (ML) engine that the activity was correctly determined. This information is used for improving or training the AI/ML model. Finally, the selected experience is loaded, in step 35.

In the case that the user, in step 33, indicated that the activity had been incorrectly determined, the process is forwarded to step 36, where this is communicated to the AI/ML engine. This information is also useful for improving or training the AI/ML model.

Furthermore, the process is forwarded to step 37, where the user is requested to enter the correct activity. Based on the entered activity, a new set of available experiences 32 is presented to the user, and the user is requested to select one of them, in the manner described above. Flowever, the new set of available experiences 32 is associated with the entered, correct activity.

Fig. 7 illustrates determining an activity performed by an exercising user in accordance with a method according to an embodiment of the invention. Inputs, xi, X2,..., XNO, in the form of measured data from sensor units are supplied to nodes, Yi ¹ , Y2 ¹ ,..., YNI ¹ , of a first hidden layer of the neural network.

The inputs xi, X2,..., XNO could, e.g. be in the form of acceleration along the x direction, acceleration along the y direction, acceleration along the z direction, three dimensions of orientation provided by a gyroscope, magnetometer measurements along the x, y and z direction, etc. Furthermore, the inputs xi, X2,..., XNO may originate from a sensor unit mounted on a movable part of an exercising device and/or from a sensor unit mounted on a head mounted output device.

The sensor data is processed by the nodes of the first hidden layer, and the processed data is supplied to nodes, Yi ² , Y2 ² ,..., YN2 ² , of a second hidden layer, where further processing is performed before the data is supplied to the next hidden layer, etc., until an output layer of the neural network is reached. For each layer of the neural network, deeper and deeper features are extracted from the data, thereby identifying patterns in the provided data, and the identified patterns may be compared to patterns identified in similar data obtained while users performed well defined exercising activities, and used for training the neural network.

The nodes, Yi ^k+1 , Y2 ^k+1 ,..., Yi\i ^k+1 , of the output layer output a number of final outputs in the form of values, each representing an activity and a confidence level, i.e. an indication regarding how likely it is that the determined activity is in fact the activity being performed by the exercising user. The confidence level thus reflects to which extend the patterns identified in the processed data match the corresponding patterns related to well defined exercising activities.

It should be noted that the activity being identified is not merely the kind of activity or the kind of exercising equipment being used, but also includes how the activity is being performed, e.g. in terms of speed, intensity, load, duration, etc.

Fig. 8 illustrates the node Yj ^k of the neural network of Fig 7. The node receives processed inputs, xi ^{k l} , X2 ^{k 1} ,..., XN ^{k l} , from the nodes from the previous layer. The inputs are provided with weights, wi, W2,..., WN, and the weighted inputs are processed at the node, using an activation function, z. The output of the node, F(z), is supplied to the nodes of the next layer of the neural network.