GB2610383A

GB2610383A - Posture sensing system

Info

Publication number: GB2610383A
Application number: GB2112403.7A
Authority: GB
Inventors: Ryan Joe
Original assignee: Vrgo Ltd
Current assignee: Vrgo Ltd
Priority date: 2021-08-31
Filing date: 2021-08-31
Publication date: 2023-03-08
Anticipated expiration: 2041-08-31
Also published as: GB202112403D0; WO2023031591A1; GB2610383B

Abstract

A posture sensing system, a method of determining the posture of a user, a computer program for the same, a method for improving the posture of a user and a method of training a machine learning algorithm for determining posture are disclosed. The posture sensing system 120 for determining the posture of a user sitting on a chair 210 comprises: a pressure sensor array 122 comprising a plurality of pressure sensors distributed in a zig-zag pattern across the seat of the chair; a controller 127 (Figure 1) coupled to the pressure sensing array, wherein the controller comprises a communication interface 128 (Figure 1); wherein the controller is configured to receive signals wherein the signals are indicative of the pressure from the plurality of sensors of the pressure sensing array, and send these signals to a remote device 150 (Figure 1) via the communication interface for determining the posture of the user. The system may further comprise an inertial measurement unit 126 (which may comprise a gyroscope or accelerometer) for determining the orientation of a back 240 of a chair.

Description

Posture sensing system

Field of the invention

The present disclosure relates to a posture sensing system and a method of determining the posture of a user.

Background

Repetitive strain injuries (RSIs) may be defined as an injury to a part of the body caused by long periods in a fixed position. Other names which may refer to RSIs are: repetitive stress disorders; cumulative trauma disorders (CTDs); and, overuse syndrome.

Spending extended periods of time (e.g. periods over 30 minutes) sedentary may cause symptoms of RSIs. For example, RSIs resulting from extended periods in a sedentary position may include pain in the muscles and/or tendons and/or nerves and/or soft tissue in any of the following body parts: neck; shoulders; elbows; forearms; wrists; hands; and, back.

Prevention of RSIs can be difficult. Treatment of RSIs can be difficult, time consuming and expensive.

There is a need to prevent and/or treat RSIs.

Summary of the invention

Aspects of the invention are as set out in the independent claims and optional features are set out in the dependent claims. Aspects of the invention may be provided in conjunction with each other and features of one aspect may be applied to other aspects.

An aspect of the disclosure provides a posture sensing system fordetermining the posture of a user sitting on a chair, the system comprising: a pressure sensor array comprising a plurality of pressure sensors distributed in a zig-zag pattern across the seat of the chair; a controller coupled to the pressure sensing array, wherein the controller comprises a communication interface; wherein the controller is configured to receive signals wherein the signals are indicative of the pressure from the plurality of sensors of the pressure sensing array, and send these signals to a remote device via the communication interface -2 -for determining the posture of the user.

The signals from the pressure sensors may be used to determine the posture of a user. Advantageously, the posture of a user may be determined with a relatively simple system (e.g. inexpensive components; low total cost of the system) relative to, for example, a lidar system (e.g. expensive components and data processing means).

For example, the signals received by the controller may be pressure signals. The controller may receive a plurality of pressure signals. Each pressure signal may be indicative of a pressure applied to a particular pressure sensor on the pressure sensor array. In other words, if there are X pressure sensors on the pressure sensor array, then the controller may receive X pressure signals wherein the nth pressure signal corresponds to the pressure applied to the nth pressure sensor.

In examples, a posture of user may be determined based on a relationship between pressure signals received from different pressure sensors. For example, a relative comparison between the pressure signal from a first sensor and the pressure signal from a second sensor may be used to determine the posture of a user (e.g. a posture which a user is sitting in wherein the posture is selected from a set of postures). For example, a relative comparison between the pressure signals from a first set of sensors and the pressure signals from a set of second sensors may be used to determine the posture of a user. For example, a plurality of relative comparisons between a pressure signal from a first sensor and a pressure signal from a second sensor may be used to determine the posture of a user.

Herein a relative comparison may refer to any of: a difference between pressure signals (e.g. pressure signals are not equal); a similarity between pressure signals (e.g. pressure signals within a preselected tolerance of one another); a first pressure signal being greater than a second pressure signal; a first pressure signal being less than a second pressure signal.

The pressure signals may be filtered by the controller.

The posture sensing system may be used by users (e.g. people sitting on a seat -3 -comprising the posture sensing system) to improve their posture and/or to break sedentary habits.

A good posture class may include: sitting up; standing up.

A bad posture class may include: leaning forward; leaning backward; leaning right leaning left; sitting cross-legged.

Therefore, a user of the posture sensing system may use the information output from the posture sensing system (e.g.: on screen feedback on a remote device; haptic feedback from a haptic feedback system on the controller; audible notifications; visual notifications) to improve their posture e.g. to adopt a good posture class fora greater period of time than a bad posture class. The posture sensing system may therefore form part of a feedback loop with the user wherein, for example: the posture sensing system provides information to the user regarding what bad posture classes they adopt (and for what time periods); then the user may use a new sitting posture in an attempt to adopt a good posture class and the posture sensing system can provide further feedback based on the new sitting posture. In this way the posture of the user may be improved by successive iterations of posture adjustment by the user and successive information provided to the user by the system.

The posture sensing system may be used in a variety of seated environments/applications. For example, the posture sensing system may be used with any of: office chairs; care home chairs; car seats; airplane seats; long distance lorry drivers' seats.

The pressure sensors may be distributed in relation to heat mapping data of average users e.g. the pressure sensors may be distributed in locations wherein an average users body meets the seat of the chair. For example, a survey may be conducted to determine the pressure distribution across a seat when users sit on a seat. In the survey, a plurality of users sit in different postures on a seat. For example the survey is for producing survey data, wherein the survey data comprises the pressure on the seat may be recorded as a function of position of the pressure on seat. The survey data may be used to generate an average pressure map of the seat e.g. the average pressure on the seat as a function of position of the pressure. The average pressure map may show on average where high -4 -pressure (e.g. pressure above a predetermined threshold) is applied to a seat.

The pressure sensing array may comprise two zig-zag arms, each zig-zag arm comprising a plurality of pressure sensors.

The pressure sensors may be disposed along two zig-zag arms based on the average pressure map. For example, pressure sensors may be disposed in locations where high pressure occurs on average. Disposing the pressure sensors in locations where high pressure occurs on average may result in pressure sensors disposed in two zig-zag arms.

The distance between each sensor is evenly distributed to each other so that the algorithm can calculate posture classification efficiently e.g. the distance along the zig-zag between neighbouring sensors is equal to permit the algorithm to determine a posture of user in a more efficient manner (e.g. less processing power required) in comparison to neighbouring sensors are not arranged in this manner.

Each of the zig-zag arms may permit flexion thereof in response to a user sitting on a seat and pressure sensing array. Therefore, the zig-zag arms may increase the lifetime of the pressure sensing array.

Herein the term zig-zag pattern refers to a pattern in which discrete elements (e.g. pressure sensors) are disposed offset on different sides of a straight line in an alternating fashion. The term zig-zag pattern may also be referred to as a serpentine pattern. For example, the term zig-zag pattern may refer to discrete elements disposed in a serpentine pattern e.g. the discrete elements are disposed on a serpentine path.

For example, a zig-zag pattern may refer to a pattern in which, moving along said straight line, a first discrete element is disposed on the left side of the straight line, a second discrete element on the right side of the straight line, a third discrete element is disposed on the left side of the straight line, a fourth discrete element is disposed on the right side of the straight line and so on.

For example, a zig-zag pattern may refer to a pattern in which, moving along said straight line, a first discrete element is disposed on the left side of the straight line, a second -5 -discrete element on the left side of the straight line, a third discrete element is disposed on the right side of the straight line, a fourth discrete element is disposed on the right side of the straight line, a fifth discrete element is disposed on the left side of the straight line, a sixth discrete element on the left side of the straight line, a seventh discrete element is disposed on the right side of the straight line, an eighth discrete element on the right side of the straight line, and so on.

In examples, the pressure sensor array may comprise two arms, one arm in which the sensors are disposed in a straight line (e.g. a straight line can pass through all of the sensors) and another arm in which the sensors are disposed in a serpentine or zig-zag pattern (e.g. a straight line cannot pass through all of the sensors and a serpentine path can pass through all of the sensors).

In examples, the pressure sensor array may comprise two arms wherein each arm comprises sensors disposed in a straight line (e.g. a straight line can pass through all of the sensors).

It has surprisingly been found to provide an improved map of the pressure distribution of a user, and thereby to increase the accuracy of the posture data generated (e.g. to increase the similarity between the determined posture of a user and the actual posture of the user), the pressure sensors on the pressure sensing array are preferably disposed in a non-linear manner, such as in a zig-zag.

The posture sensing system may comprise at least one inertial measurement unit, IMU, for determining the orientation of the pressure sensing array; and wherein the controller may be configured to receive sensor signals indicative of the orientation of the pressure sensor array from the at least one IMU and combine these with the signals indicative of the pressure and send these combined signals to the remote device via the communication interface.

In examples, the posture sensing system may comprise at least one inertial measurement unit, IMU, for determining any of: the orientation of the back of the chair and wherein the controller may be configured to receive sensor signals indicative of the orientation of the back of the chair from the at least one IMU and combine these with the signals indicative -6 -of the pressure and send these combined signals to the remote device via the communication interface.

In examples, wherein the IMU is configured for determining the orientation of the back of the chair, the system may be able to determine any of: motion of the user leaning back in the chair; the turning of the chair; the raising of the chair. In examples, the IMU may be used to determine if the user's chair is adjusted to the correct height for the user e.g. the system may be able to determine if the height of the seat of the chair is suitable for a user, based on a user's height.

In examples, the controller may not combine the signals indicative of the orientation of the pressure sensing array and the signals indicative of the pressure. Instead, the two types of signals may be sent separately to the remote device via the communication interface. The two types of signal may be used to train the classifier algorithm. Using two types of signal (orientation signal and pressure signal) may increase the accuracy of the classifier algorithm.

In examples, the controller may not combine the signals indicative of the orientation of the back of the chair and the signals indicative of the pressure. Instead, the two types of signals may be sent separately to the remote device via the communication interface. The two types of signal may be used to train the classifier algorithm. Using two types of signal (orientation signal and pressure signal) may increase the accuracy of the classifier algorithm.

The inertial measurement unit (IMU) may comprise an accelerometer and a gyroscope. The IMU may be configured to generate an orientation signal wherein the orientation signal may be indicative of the orientation of the pressure sensing array. In examples, the orientation signal may be indicative of a change in orientation of the pressure sensing array.

In examples, the IMU is disposed on the chair (e.g. separate from the pressure sensor array). The position at which the IMU is disposed relative to the pressure sensor array may be stored in the controller and used to determine the change in orientation of the pressure sensor array based on the orientation signal. -7 -

The communication interface (e.g. the controller communication interface) may be a short-range wireless communication interface, and wherein the controller is configured to send the signals to the remote device via the short-range wireless communication interface.

The posture sensing system of may comprise the remote device, wherein the remote device comprises a wireless communication interface for receiving signals from the controller, and wherein the remote device is configured to generate posture data wherein the posture data is indicative of the posture of the user based on the received signals from the controller. In examples, the signals may comprise any of: signals indicative of the pressure; filtered signals indicative of the pressure; sensor signals indicative of the orientation of the pressure sensor.

The remote device may be configured to send the posture data to a cloud computing platform, OCR, wherein the OCR is configured to provide a graphical user interface to the user to see their posture data as a function of time.

The user may use the posture data to inform in what posture they sit in the future. Iterative rounds of adjusting posture based on the posture data may improve a user's posture.

The OCR may be configured to determine a posture routine based on a function of the user's posture data over time. The posture routine may comprise a recommended routine to break sedentary sitting and improve the posture of the user.

The controller may comprise a haptic feedback means, and wherein the haptic feedback system is configured to be triggered in the event that the user is sedentary for longer than a selected threshold period of time.

The remote device may be configured to trigger the haptic feedback means in the event that the remote device determines that the user has been in the same posture for greater than a selected threshold period of time.

The CCP may be configured to trigger the haptic feedback. -8 -

A posture notification is configured to prompt a user of the system to change posture. For example, a posture notification may comprise a digital message configured to operate another piece of hardware (e.g. the remote device; the controller of posture sensing system; the haptic feedback system) to generate a physical notification which can be sensed by a user (e.g. visual notification; audible notification; haptic feedback; vibration).

Posture notifications may be generated when pressure signals and/or orientation signals from the posture sensing system indicate that the user has not moved from a sedentary position for a selected threshold period of time.

In examples, the controller may receive a posture notification from the remote device wherein the posture notification is sent to the controller in the event that the user is sedentary for longer than a selected threshold period of time. The haptic feedback system may be configured to trigger when the controller receives the posture notification.

In examples, the posture controller may be configured to generate any of: a posture datum; and, a posture notification, based on the pressure signals received from the pressure sensor array.

In examples, the COP may generate a posture notification and send the posture notification to the remote device.

In examples, the controller may make a determination (e.g. local to the controller) based on the pressure signals obtained from the pressure sensor array. For example, if the pressure signals are indicative that a user has been sat on the seat fora selected threshold period of time (e.g. an hour), then the controller may trigger the haptic feedback system.

In examples, the haptic feedback means comprises a vibration motor. The controller may comprise the haptic feedback means. The controller may be disposed on the chair (e.g. on an edge of the seat) so that in the event that the haptic feedback means is triggered, a vibration is passed through the seat.

The haptic feedback system may be configured to be triggered in the event that the user is sedentary for longer than a selected threshold period of time. In examples, the selected -9 -threshold period of time may be any of: 15 minutes 30 minutes; 45 minutes; 60 minutes.

The controller may be configured to filter the received signals indicative of the pressure from the plurality of sensors of the pressure sensor array before the signals are sent to the remote device.

Advantageously, filtering the signals (e.g. pressure signals) may maximise the accuracy and/or responsiveness of the posture sensing system. Advantageously, filtering the signals may reduce vibration pick-up.

Filtering may be performed by passing the pressure signals through a high-pass filter to obtained filtered pressure signals.

Filtering may be performed by passing the pressure signals through a low-pass filter to obtained filtered pressure signals. For example, the low-pass filter may be configured to remove frequencies greater than 10 Hertz.

An aspect of the disclosure provides: a chair comprising a posture sensing system described herein.

In examples, the pressure sensor array may be provided within the seat e.g. the plurality of sensors are an integral part of the seat. For example, the plurality of sensors may be disposed under a top layer of upholstery of the seat. In examples, the upholstery of the seat may comprise a smart fabric e.g. fabric configured for electronic components (e.g. a pressure sensor array) to be disposed therein.

In examples, the plurality of sensors may be provided as a separate device which, for use, is placed atop the seat e.g. such that, in use the plurality of sensors are disposed a user and the seat.

In examples, one or more additional pressure sensors may be provided on a back rest of the seat e.g. the backrest of a chair.

The pressure sensor array may be enclosed within a fabric. Preferably, the fabric is -10 -hardwearing and/or elastic. In examples, a polyether-polyurea copolymer fabric may be used e.g. Lycra (RTM). In examples, the fabric is a smart fabric e.g. fabric configured for electronic components (e.g. a pressure sensor array) to be disposed therein.

The pressure sensor array may comprise a top side and a bottom side. The bottom side is configured to face away from the user and the top side is configured to face toward the user.

In examples wherein the pressure sensor array is disposed atop the seat, the bottom side of the pressure sensor array faces toward and contacts the top of the seat. In examples, the bottom side of the pressure sensor may comprise one or more adhesive portions. Advantageously, providing one or more adhesive portions may reduce or prevent movement of the pressure sensor array relative to the seat. Reducing or preventing movement of the pressure sensor array relative to the seat may improve data consistency e.g. because a given sensor in the pressure sensor array will always be disposed in approximately the same position relative to the seat.

Herein the chair may be any of: a stool (e.g. a chair without a back); and, a chair (e.g. a seat with a back).

The remote device may be a desktop computer or laptop computer. The remote device may be a smart phone or a tablet computer.

An aspect of the disclosure provides a method of determining the posture of a user, the method comprising: obtaining a plurality of signals indicative of pressure from a pressure sensing array comprising a plurality of pressure sensors; sending the set of signals indicative of pressure to a remote device; determining a posture of the user by applying a trained machine learning model to the set of filtered signals indicative of pressure from the pressure sensing array.

An aspect of the disclosure provides a computer program product comprising instructions which, when the program is executed by a computer, cause the computer to carry out a method of determining the posture of a user, the method comprising: obtaining a plurality of signals indicative of pressure from a pressure sensing array comprising a plurality of pressure sensors; determining a posture of the user by applying a trained machine learning model to the set of signals indicative of pressure from the pressure sensing array.

In examples, the method may further comprise: sending the set of signals indicative of pressure to a remote device, The step of determining the posture of a user as a function of time may be executed by a method comprising: obtaining a plurality of signals indicative of pressure from a pressure sensing array comprising a plurality of pressure sensors; sending the set of signals indicative of pressure to a remote device; determining a posture of the user by applying a trained machine learning model to the set of signals indicative of pressure from the pressure sensing array.

An aspect of the disclosure provides a method for improving the posture of a user sitting on a chair, the method comprising: determining the posture of a user as a function of time; in the event that the posture of a user has not changed by more than a selected threshold for more than a selected threshold period of time, sending an alert to the user.

The method may further comprise: obtaining a plurality of signals indicative of pressure from a pressure sensing array comprising a plurality of pressure sensors; sending the set of signals indicative of pressure to a remote device; determining a posture of the user by applying a trained machine learning model to the set of signals indicative of pressure from the pressure sensing array.

Optionally, the method may further comprise a step of filtering the obtained signals indicative of pressure from the pressure sensing array to obtain a set of filtered signals indicative of pressure; Sending an alert to the user may comprise at least one of (i) providing haptic feedback to the user, and (ii) providing a notification to the user to alter their posture.

An aspect of the disclosure provides a method of training a trained classifier algorithm, the method comprising: (i) obtaining pressure signals from the posture sensing system when the user -12 -of the system is sat in a known posture; (ii) filtering the pressure signals to obtain filtered pressure signals; (iii) labelling the filtered pressure signals with an indication of the known posture to obtain a list of labelled filtered pressure signals; (iv) repeating steps (i) to (iii) for different known postures to obtain a list of labelled filtered pressure signals; (v) inputting the labelled filtered pressure signals into an untrained classifier algorithm to generate the trained classifier algorithm.

Herein the term user may refer to a person who sits upon a seat comprising the system. Herein the term pressure signals may refer to signals indicative of the pressure. Herein the term filtered pressure signals may refer to signals indicative of the pressure which have been filtered. Herein the term orientation signals may refer to signals indicative of the orientation of the pressure sensor array.

Each fork of the pressure sensor array may comprise any number of pressure sensors, for example, one, two, three, or, four. Advantageously, providing an increased number of pressure sensors may increase the accuracy and/or granularity of the raw pressure data generated by the pressure sensor array.

An aspect of the disclosure provides a computer readable non-transitory storage medium comprising a program for a computer configured to cause a processor to perform any of the methods described herein.

An aspect of the disclosure provides a computer system comprising a computer readable non-transitory storage medium, wherein the computer readable non-transitory storage medium comprising a program for a computer configured to cause a processor to perform any of the methods described herein.

Herein a posture datum may be indicative of a posture of the user when the signals used to generate the posture datum were generated. The posture datum may be indicative of the posture of the user, wherein the posture is selected from a list of postures recognised by an algorithm e.g. wherein the posture is selected from a list of postures comprising slouching; leaning forward; leaning backward leaning left; leaning right; sitting up; sitting -13 -cross-legged; standing up (e.g. not sitting on the seat).

The controller may comprise a Bluetooth Nordic nRF52 based module which provides all operational, sensing and Bluetooth communication functions. The controller may be configured for BLE (Bluetooth Low Energy). The controller may be configured to permit pairing with a remote device (e.g. a host PC). The controller may comprise a PCB. The controller may comprise any of: an IMU; an accelerometer; and, a gyroscope. The controller may comprise a power source. The power source may be configured to provide power to the controller (e.g. to provide energy to the components of the controller). The PCB may be configured to read any of: the orientation signals (e.g. signals from the IMU); accelerometer signals (e.g. signals from an accelerometer); gyroscopic signals (e.g. signals from a gyroscope); and, pressure signals (e.g. signals from the pressure sensors on the pressure sensor array). The controller may be configured to provide values over Bluetooth at a rate of greater than and/or equal to 10 Hertz.

An aspect of the disclosure provides a computer-implemented method of displaying posture data within a window displayed in a graphical user interface, the method comprising: obtain posture data from a remote device, wherein the posture data is indicative of the posture of a user of a posture sensing system; displaying the posture data within a window displayed in a graphical user interface.

The posture data may comprise a particular posture which a user has sat in most frequently.

In examples, the step of displaying the posture data within a window displayed in a graphical user interface may comprise any of: displaying historical posture data of the user e.g displaying the posture data of a user gathered between two points of time; displaying whether a user's posture during a selected time period (e.g. the previous day) meets a selected goal; displaying whether a user's posture during a plurality of selected time periods meets a selected goal e.g. the GUI may display goal progression for a user; displaying a posture rating for a user e.g. displaying how a user's posture over a -14 -period of time compares to an ideal posture.

Drawinas Embodiments of the disclosure will now be described, by way of example only, with reference to the accompanying drawings, in which: Figure 1 illustrates a schematic view of an arrangement comprising a posture sensing system; Figure 2 illustrates a plan view of a posture sensing system; Figure 3 illustrates a perspective view of a pressure sensing array of the posture sensing system of Figure 2; Figures 4A to 4F illustrate a view of a graphical user interface of the posture data sent from a cloud computing platform to a remote device operated by the user.

Specific description

Embodiments of the claims relate to a posture sensing system and method.

It will be appreciated from the discussion above that the embodiments shown in the Figures are merely exemplary, and include features which may be generalised, removed or replaced as described herein and as set out in the claims. The actual example actually shown in Figures 1 to 3 and also Figures 4A to 4F.

An arrangement 100 for improving the posture of a user of a chair is shown in Figure 1. The arrangement 100 comprises: a system 120; a remote device 150; and a cloud computing platform (CCP) 160.

The system 120 is (e.g. wirelessly) connected to the remote device 150. The remote device 150 is (e.g. wirelessly) connected to the COP 160.

The system 120 comprises: a pressure sensor array 122; an inertial measurement unit (I MU) 126; and, a controller 127.

The IMU 126 comprises an accelerometer and a gyroscope.

-15 -The controller 127 comprises: a controller communication interface 128. In the present example, the IMU 126 is disposed on the controller 127. However, in examples, the IMU 126 may be provided separate from the controller 127.

The controller 127 is connected to the pressure sensor array 122 (e.g. the controller is configured to receive, via the controller communication interface 128, signals from the pressure sensor array). The controller 127 is connected to the IMU 126 (e.g. the controller is configured to receive, via the controller communication interface 128, signals from the IMU) The remote device 150 comprises a remote device communication interface 158, a display 154, and, an input means 156.

CCP 160 comprises: a COP communication interface 168; a CCP processing means 164.

The pressure sensor array 122 comprises: a plurality of pressure sensors 124A to 124H; a top surface 122T; and a bottom surface 122B. The pressure sensors 124A to 124 H are disposed in forked zig-zag pattern shown in Figure 2 and Figure 3.

The forked zig-zag pattern comprises: a stem portion 310; a bifurcation point 315; a left fork 320L, and, a right fork 320R.

The step portion 310 is connected to the bifurcation point 315. The left fork 320L is connected to the bifurcation point 315. The right fork 320R is connected to the bifurcation point 315. The left fork 320L has a zig-zag pattern. The right fork has a zig-zag pattern.

A first pressure sensor 124A, a second pressure sensor 124B, a third pressure sensor 1240, and a fourth pressure sensor 124D are disposed on the left fork 320L. A fifth pressure sensor 124E, a sixth pressure sensor 124F, a seventh pressure sensor 124G, and an eighth pressure sensor 124H are disposed on the right fork 320R.

The first pressure sensor 124A is disposed closest along the left fork 320L to the bifurcation point 315. The fifth pressure sensor 124E is disposed closest along the right fork 320R to the bifurcation point 315.

-16 -The fourth pressure sensor 124D is disposed further along the left fork 320L from the bifurcation point 315 than the first, second, and third sensors 124A to 124C. The third pressure sensor 124C is disposed furtheralong the left fork 320L from the bifurcation point 315 than the first and second sensors 124A and 1248. The second pressure sensor 1248 is disposed further along the left fork 320L from the bifurcation point 315 than the first sensor 124A.

The eighth pressure sensor 124H is disposed further along the right fork 320R from the bifurcation point 315 than the fifth, sixth, and seventh sensors 124E to 124G. The seventh pressure sensor 124G is disposed further along the right fork 320R from the bifurcation point 315 than the fifth and sixth sensors 124E and 124F. The sixth pressure sensor 124F is disposed further along the right fork 320R from the bifurcation point 315 than the fifth sensor 124E.

Herein the first pressure sensor 124A and the fifth pressure sensor 124E may be referred to as complementary pressure sensors. For example, the complementary pressure sensor of the first pressure sensor 124 is the fifth pressure sensor 124E and vice versa. It will be appreciated that the term complementary pressure sensors may therefore referto any pair of pressure sensors which are disposed on opposite forks but occupy the same position along their respective forks from the bifurcation point. For example, in the example shown in Figure 3, the first pressure sensor 124A and the fifth pressure sensor 124 each have 3 other sensors between themselves and the bifurcation point 315.

In the present example, the complementary pressure sensor pairs are: * the first pressure sensor 124A and the fifth pressure sensor 124E; * the second pressure sensor 1248 and the sixth pressure sensor 124F; * the third pressure sensor 124C and the seventh pressure sensor 124G; * the fourth pressure sensor 124D and the eighth pressure sensor 124E.

The chair 210 comprises: a seat 220; a back portion 240; and, a pair of arms 260L and 260R. The seat 220 comprises: a sitting surface 222; a seat front edge 224; and, a seat back edge 226.

-17 -The system 120 is disposed on the chair 210. The pressure sensor array 122 of the system 120 is disposed on the sitting surface 222 e.g. the bottom surface 1228 of the pressure sensor array 122 is disposed on the sitting surface 222. In examples, the bottom face of the pressure sensor array may comprise one or more adhesive regions (e.g. regions of the bottom face 1228 comprises adhesive), wherein the adhesive regions are configured to prevent or reduce movement of the pressure sensor array relative to the sitting surface (e.g. the adhesive regions adhere the pressure sensor array to the sitting surface 222 to thereby reduce or prevent movement of the pressure sensor array relative to the sitting surface).

The pressure sensor array 122 is disposed on the sitting surface 222 such that the bifurcation point of the pressure sensor array is disposed closer to the back that the tips of the prong portions of the pressure sensor array (e.g. such as the arrangement shown in Figure 2).

The controller 127 is disposed on the chair 210. The IMU is disposed on the chair 210. In the example shown in Figure 2, the controller 127 and the IMU 126 are disposed on the back portion 240 of the chair 210. Forexample, the controller may comprise a fixing means for connection to the chair. The fixing means may comprise at least one of: a strap; complementary hooks and loops (e.g. Velcro(RTM)); and, an adhesive. In examples wherein the fixing means comprises complementary hooks and loops, a set of hooks may be disposed using adhesive to the chair and a set of loops may be disposed using an adhesive to the controller.

In examples, the controller may be disposed on an edge portion of the seat e.g. the back edge or the front edge or an edge connecting the front edge and the second edge.

The chair 210 and system 120 are configured to simultaneously enable a user to sit on the chair 210 and use the posture sensing system 120.

The system 120 is configured to be disposed upon a seat. Figure 2 shows the system 120 disposed on seat 210. The arrangement shown in Figure 2 is described in more detail below.

The pressure sensor array 122 is configured for disposal across a seat of a chair. The -18 -pressure sensor array 122 may be disposed either: atop a seat of a chair; or, within the seat of a chair.

In examples wherein the pressure sensor array 122 is disposed atop a seat of a chair (shown in Figure 2) the pressure sensor array 122 is disposed over an outermost surface of the seat. In such examples, in use, the pressure sensor array is disposed between the seat of the chair and a user of the chair and posture sensing system.

In examples wherein the pressure sensor array 122 is disposed within the seat of a chair the pressure sensor array 122 is disposed between a first part of the seat and a second part of the seat.

Each of the pressure sensors 124A to 124G of the pressure sensor array 122 is configured to generate a signal indicative of the pressure applied to the pressure sensor. For example, a signal indicative of the pressure applied to a pressure sensor may be referred to herein as a pressure signal.

For example, in use, a user sits on the seat of the chair and, therefore, also sits on the pressure sensor array 122. A user who sits on the seat of the chair exerts a pressure on the seat of the chair and also on one or more of the pressure sensors 124A to 124G. The pressure exerted by the user seated upon the seat is distributed across the seat 210 and the pressure sensors 124A to 124G in a manner corresponding to the posture of the user.

Pressure signals generated by the pressure sensors 124A to 124G of the pressure sensor array 122 are provided to the controller 127.

The I MU 126 is configured to generate a signal indicative of a change of orientation of the seat. For example, a signal indicative of a change of orientation of the seat may be referred to herein as an orientation signal.

The I MU 126 is disposed on a side face of the seat (see description below).

The controller 127 is configured to receive pressure signals from the pressure sensor array 122 (e.g. from any of the pressure sensors 124A to 124G). The controller 127 is configured -19 -to receive pressure signals from the I MU 126.

In examples, the controller may filter the pressure signals to generate filtered pressure signals. In such examples, the filtered pressure signals are sent to the remote via the controller communication interface.

The controller communication interface 128 of the controller 127 is configured to send signals to the remote device 150. Signals sent from the controller communication interface 128 of the controller 127 may comprise at least one of: pressure signals; filtered pressure signals; orientation signals.

In examples, the controller communication interface 128 may be a short-range wireless communication interface. Herein a short-range wireless communication interface may refer to an interface via which signals are sent via any of: Bluetooth (RTM); Zig Bee(RTM); WFi (RTM).

The remote device communication interface 158 of the remote device 150 is configured to receive signals from the controller 127. The remote device communication interface 158 of the remote device 150 is configured to receive at least one of: pressure signals; filtered pressure signals; and, orientation signals from the controller 127.

In examples, the posture sensing system and the remote device are disposed within close proximity to one another e.g. they are typically in the same room. In examples, in use the posture sensing system and the remote device are disposed within 2 meters of one another, more preferably within 1 meter of one another. Typically, the posture sensing system is disposed on a chair upon which a user sits to operate the remote device (e.g. an office chair used to operate an office computer (e.g. a remote device)).

In examples, the pressure sensors are configured to provide pressure signals to the controller at a selected sensing rate. For example, the pressure sensors may be configured to provide pressure signals at a sensing rate of 1 signal per pressure sensor per second.

The remote device 150 is configured to determine a posture of the user based on the signals (e.g. pressure signals and/or orientation signals) received by the remote device.

-20 -The remote device 150 comprises a classifier algorithm configured to determine a posture of the user based on the signals received by the remote device. The classifier algorithm is described in more detail herein.

The remote device is configured to determine the posture of a user and to generate posture data wherein the posture data is indicative of the posture of the user.

The COP communication interface 168 of the COP 160 is configured to receive signals and/or data from the remote device communication interface 158 of the remote device 150. The COP communication interface 168 of the OCR 160 is configured to provide signals and/or data received from the remote device communication interface 158 to the COP processing means 164.

The OOP processing means 164 is configured to generate posture notifications based on posture data. e.g. wherein the posture data is received from the remote device 150. The COP processing means 164 is configured to generate a posture notification based on at least one posture datum from the remote device.

A posture notification is configured to prompt a user of the system to change posture.

The OOP processing means 164 is configured to generate posture notifications based on the posture data received by the COP 160. In examples, if the posture data received by the COP 160 are indicative of a user in a given undesirable posture (e.g. any posture which is neither sitting up nor standing up) for a selected period of time, then the COP processing means 164 will generate a posture notification. In the event that the COP processing means 164 generates a posture notification, then the CCP communication interface 168 is configured to send the posture notification to the remote device 150.

In examples, posture notifications received by the remote device 150 may be sent to the controller 127 of the posture sensing system 120 e.g. via the remote device communication interface 158. The controller 127 may be configured to provide haptic feedback to the user in response to the posture notification. For example, the controller 127 may comprise a haptic feedback system configured to provide haptic feedback to the user of the system. The haptic feedback system may comprise a vibration motor configured to vibrate the chair -21 - (e.g. the system is disposed on the chair and vibrations from the vibration motor are transmitted to the user via the chair).

For example, posture notifications received by the remote device 150 are configured to visually notify the user to change posture via the display 156 of the remote device. In examples, the local device may comprise a speaker wherein the speaker. In such examples, the posture notifications received by the remote device 150 may be configured to audibly notify the user to change posture via the speaker of the remote device.

Furthermore, posture notifications may notify the user to perform a given activity (e.g. an exercise). For example, the exercise may be configured to reduce repetitive stress of the user's anatomy based on the type of posture they exhibit most frequently.

For example, if the user has a proclivity for leaning forward (e.g. they lean forward more than any other posture) then the posture notification may notify the user to stand up and stretch their arms above their heads to mitigate the negative health associations of leaning forward.

In examples, the controller 127 may be configured to generate posture notifications based on the posture data.

In examples, the COP processing means may be configured to receive signals from the pressure sensing array of the posture sensing system. The signals are sent from the pressure sensing array to the COP processing means via, the controller communication interface, the remote device communication interface, and the cloud computing communication interface.

The remote device display 156 is configured to display historical posture data. The remote device 150 is configured to obtain historical posture data from a memory of the remote device 150 and/or from the COP 160. The remote device 150 is configured to request historical posture data from the COP 160 e.g. the remote device 150 is configured to request historical posture data in response to an input to the remote device wherein the input is received via input means 156.

-22 -In use, a user of the pressure sensing system sits on the sitting surface 222 of the chair 210. The pressure sensor array 122 is disposed on the sitting surface 222 and, therefore, the user sitting on the sitting surface also provides a pressure (e.g. the user's weight) onto the pressure sensor array 122. The pressure sensors 124A to 124G of the pressure sensor array 122 are acted on by the pressure provided by the user. The magnitude of the pressure exerted on each of the pressure sensors 124A to 124G depends upon the posture of the user. The pressure sensors 124A to 124G each provide pressure signals indicative of the magnitude of the pressure applied to each of the respective pressure sensors 124A to 124G to the controller 127. The I MU 126 provides an orientation signal to the controller 127 The controller 127 provides, via the controller communication interface 128, the pressure signals to the remote device communication interface 158. The controller 127 generates, using a classifier algorithm, a posture datum based on the pressure signals and the orientation signal. The controller 127 provides, via the controller communication interface 128, the posture datum to the remote device communication interface 158.

The remote device communication interface 158 receives the posture datum from the controller communication interface 128. The CCP processing means 164 generates posture notifications based on the posture datum. The CCP sends, via the COP communication interface 168, the posture notification to the remote device 150.

In an example, the remote device 150 sends the posture notification to the controller 127. The posture notification triggers a haptic feedback system of the controller, thereby vibrating the chair 210.

A user may feel the vibration and stand up thereby breaking a period of extended sitting.

Supervised machine learning is a subcategory of machine learning and artificial intelligence. Supervised machine learning may be defined by its use of labelled datasets to train algorithms to classify data or predict outcomes accurately.

An untrained supervised machine learning classification algorithm is configured to receive training data. The untrained supervised machine learning algorithm is configured to adjust -23 -parameters therein (e.g. weights in functions), based on the training data, until an exit condition is satisfied. The exit condition may be when the algorithm is fitted to an acceptable degree and can be verified by a cross-validation process.

Training data for training a classification algorithm typically comprises: parameters; and, a label, wherein the label identifies the class of the parameters. The untrained supervised machine learning classification algorithm is configured to define each class by generating rules based on the parameters for a given class.

When the exit condition is satisfied the untrained supervised machine learning classification algorithm may be referred to as a trained supervised machine learning classification algorithm.

The trained supervised machine learning classification algorithm (referred to as the trained classification algorithm or trained classifier algorithm) is configured to receive test data and to separate the test data into the classes (e.g. based on the generated rules based on the parameters of each class).

The trained classification algorithm may be stored on the remote device.

In examples, the trained classification algorithm may be stored on the controller. In such examples, classification of a user's posture can be performed at the controller to generate posture data. The posture data can be sent, via the remote device, to the cloud computing platform.

Common classification algorithms are: linear classifiers; support vector machines (SVM), decision trees; k-nearest neighbour; and random forest.

In examples, the present disclosure generates a trained classification algorithm using the machine learning platform Azure ML (RTM).

The posture sensing system is used to obtain data (e.g. pressure signals from the pressure sensors) of users in pre-defined sitting postures for a selected period of time.

-24 -In examples, a method of obtaining training data comprises: (i) obtaining pressure signals from the posture sensing system when the user of the system is sat in a known posture; (H) filtering the pressure signals to obtain filtered pressure signals; (Hi) labelling the filtered pressure signals with an indication of the known posture to obtain a list of labelled filtered pressure signals; (iv) repeating steps (i) to (iii) for different known postures to obtain training data comprising a list of labelled filtered pressure signals.

The training data is configured to be input into an untrained machine learning classifier algorithm. A method of training a classifier algorithm to obtain a trained classifier algorithm comprises: (v) inputting the labelled filtered pressure signals into an untrained machine learning classifier algorithm to generate the trained classifier algorithm.

A method of training the classifier algorithm to obtain a trained classifier algorithm comprises: (i) obtaining pressure signals from the posture sensing system when the user of the system is sat in a known posture; (H) filtering the pressure signals to obtain filtered pressure signals; (Hi) labelling the filtered pressure signals with an indication of the known posture to obtain a list of labelled filtered pressure signals; (iv) repeating steps (i) to (Hi) for different known postures to obtain training data comprising a list of labelled filtered pressure signals; (v) inputting the labelled filtered pressure signals into an untrained machine learning classifier algorithm to generate the trained classifier algorithm.

In examples, the user may be instructed to sit in a given posture for a particular period of time and then to sit in another given posture for a particular period of time and so on (e.g. 60 seconds (s) of sitting in the user's current posture, then 60s of standing up, then 60s of leaning forward, then 60s of sitting upright in a proper posture, then 60s of leaning back, then 60s of slouching). The posture sensing system may be configured to obtain and filter pressure signals (e.g. steps (i) and (H)) forcorresponding periods of time in order to capture pressure signals when the user is sat in the given posture.

-25 -In examples, the classifier algorithm may be semi-trained prior to being trained for a specific user. For example, the untrained classifier algorithm may be trained using data from a plurality of users. In examples, the data from the plurality of users may be obtained using steps (i) to (iv) set out above. The data from the plurality of users is then inputted into an untrained machine learning classifier algorithm to generate a semi-trained classifier algorithm. Subsequently, the semi-trained classifier algorithm may be trained for a specific user using method steps (i) to (v) set out above, in particular, taking a semi-trained classifier algorithm and applying method sets (i) to (v) set out above to obtain a trained classifier algorithm, wherein the trained classifier algorithm is specific to the user.

During data collection, each of the pressure sensors 124A to 124H is configured to obtain data at 10 Hertz.

For example a classifier model is trained using a data set comprising a list pressure sensor data and a corresponding list describing the posture of the user when the data was obtained. An example data set fortraining a classifier model forthe posture sensing system 100 is set out below.

Col. 1 Col. 2 Col. 3 Col. 4 Col. 5 Col. 6 Col. 7 Col. 8 Col. 9 Slouching M M H M M H M M Leaning M H H L L H H M forward Leaning backward L L M L L M L L Leaning left H H H M L L L L Leaning right L L L L M H H H -26 -Sitting up M M H M M H M M Sitting L H H H L L M L cross-legged Standing up L L L L L L L L Wherein: * Col. 1 corresponds to the known posture of the user when the data was generated; * Col. 2 corresponds to the pressure signal output from the first sensor 124A when the user is sat in the known posture; * Col. 3 corresponds to the pressure signal output from the second sensor 124B when the user is sat in the known posture; * Col. 4 corresponds to the pressure signal output from the third sensor 124C when the user is sat in the known posture; * Col. 5 corresponds to the pressure signal output from the fourth sensor 124D when the user is sat in the known posture; * Col. 6 corresponds to the pressure signal output from the fifth sensor 124E when the user is sat in the known posture; * Col. 7 corresponds to the pressure signal output from the sixth sensor 124F when the user is sat in the known posture; * Col. 8 corresponds to the pressure signal output from the seventh sensor 124G when the user is sat in the known posture; * Col. 9 corresponds to the pressure signal output from the eighth sensor 124H when the user is sat in the known posture.

The above table characterises the pressure signals in a relational manner e.g. if the pressure signal has a relatively low magnitude (L), a relatively medium magnitude (M), or a relatively high magnitude (H).

In examples, each of the pressure sensors may output an impedance proportional to the pressure applied to the pressure sensor. The impedance may be on the order of KiloOhms, -27 -kn.

In examples, the above table(s) may be generated as Comma-Separated Values (CSV) files.

The trained classification algorithm is configured (e.g. by training) to classify test data into a plurality of different postures (e.g. posture classes) of the user. In the present example, the posture determining algorithm is configured (e.g. trained) to recognise eight distinct postures of the user described in more detail herein.

In examples, the CCP processing means may be configured to generate a posture datum based on the signals received from the CCP communication interface.

The CCP processing means may be configured to provide the posture datum to the CCP communication interface.

The CCP communication interface may be configured to provide the posture datum to the remote device communication interface.

Posture classes The eight posture classes (also referred to herein as simply as postures) are: 1. slouching; 2. leaning forward; 3. leaning backward; 4. leaning left; 5. leaning right; 6. sitting up; 7. sitting cross-legged; 8. standing up (e.g. not sitting on the seat).

1. Slouching Typically, slouching is characterised by the pressure sensors disposed closer to the front edge of the seat receiving a pressure of greater magnitude than the pressure signals further from the front edge of the seat. In other words, when pressure signals provided by -28 -the pressure sensors disposed closer to the front edge of the seat are indicative of a greater pressure applied thereto than pressure signals provided by the pressure sensors disposed further from the front edge of the seat, then this is an indication of slouching.

In the example shown in Figure 2 and Figure 3, when a user of the system slouches on the seat, the pressure from each pressure sensor may fulfil the inequalities below: PA > PB > Pc > PD; PE > PF > PG > PH; Wherein PA represent the magnitude of the pressure applied to pressure sensor 124A etc. 2. Leaning forward Typically, leaning forward is characterised by the pressure sensors disposed closer to the front edge of the seat receiving a pressure of greater magnitude than the pressure signals further from the front edge of the seat. In other words, when pressure signals provided by the pressure sensors disposed closer to the front edge of the seat are indicative of a greater pressure applied thereto than pressure signals provided by the pressure sensors disposed further from the front edge of the seat, then this is an indication of leaning forward.

In the example shown in Figure 2 and Figure 3, when a user of the system leans forwards on the seat, the pressure from each pressure sensor may fulfil the inequalities below: PA > PB > PC > PO; PE > PF > PG > PH; To distinguish between slouching and leaning forward, the system may also use orientation data e.g. obtained from the I MU. For example, if the conditions (e.g. inequalities) on pressure set out above are fulfilled determine: * if the orientation data is indicative that the chair is leaning forward (e.g. front edge lower than the back edge) then a determination of leaning forward is made; * if the orientation data is indicative that the chair is leaning backward (e.g. back edge lower than the front edge) then a determination of slouching is made.

3. Leaning backward Typically, leaning backward is characterised by the pressure sensors disposed closer to -29 -the back edge of the seat receiving a pressure of greater magnitude than the pressure signals further from the back edge of the seat. In other words, when pressure signals provided by the pressure sensors disposed closer to the back edge of the seat are indicative of a greater pressure applied thereto than pressure signals provided by the pressure sensors disposed furtherfrom the back edge of the seat, then this is an indication of leaning backward.

In the example shown in Figure 2 and Figure 3, when a user of the system leans backward on the seat, the pressure from each pressure sensor may fulfil the inequalities below: PA < Pe < Pc < Po; PE < PF < PG < PH, 4. Leaning left Typically, leaning left is characterised by the pressure sensors disposed on the left fork 320L of the pressure sensor array 122 receiving a pressure of greater magnitude than their respective corresponding pressure sensors on the right fork 320R. In other words, when pressure signals provided by the pressure sensors disposed on the left fork 320L are indicative of a greater pressure applied thereto than pressure signals provided by the corresponding pressure sensors disposed on the right fork 320R, then this is an indication of leaning left.

In the example shown in Figure 2 and Figure 3, when a user of the system leans left on the seat, the pressure from each pressure sensor may fulfil the inequalities below: PA > PE, PB > PF, Pc > PG; Po > PH, 5. Leanina riaht Typically, leaning right is characterised by the pressure sensors disposed on the right fork 320R of the pressure sensor array 122 receiving a pressure of greater magnitude than their respective corresponding pressure sensors on the leftfork 320L. In other words, when pressure signals provided by the pressure sensors disposed on the right fork 320R are indicative of a greater pressure applied thereto than pressure signals provided by the corresponding pressure sensors disposed on the left fork 320L, then this is an indication -30 -of leaning right.

In the example shown in Figure 2 and Figure 3, when a user of the system leans right on the seat, the pressure from each pressure sensor may fulfil the inequalities below: PA < PE, Pe < PF; Pc < PG; Po < PH, 6. Sitting up Typically, sitting up is characterised by the pressure sensors disposed on the right fork 320R of the pressure sensor array 122 receiving a pressure of equal magnitude than their respective corresponding pressure sensors on the left fork 320L. In other words, when pressure signals provided by the pressure sensors disposed on the right fork 320R are indicative of a pressure applied thereto which is equal to the pressure signals provided by the corresponding pressure sensors disposed on the left fork 320L, then this is an indication of sitting up.

In the example shown in Figure 2 and Figure 3, when a user of the system is sitting up on the seat, the pressure from each pressure sensor may fulfil the equalities below: PA = PE, PB = PF; Pc = Pc; PD = PH; In examples, the above equalities need only be met within a selected tolerance amount, PT, e.g.: PA = PE ± PT; PB = PF ± PT; Pc = PG ± PT; Po = PH ± PT, 7. Sittina cross-leaped Typically, sitting cross-legged is characterised by the system detecting leaning forward -31 -and sitting up simultaneously.

Typically, sitting cross-legged is characterised by: the pressure sensors disposed closer to the front edge of the seat receiving a pressure of greater magnitude than the pressure signals further from the front edge of the seat; and, the pressure sensors disposed on the right fork 320R of the pressure sensor array 122 receiving a pressure of equal magnitude than their respective corresponding pressure sensors on the left fork 320L In other words, when both: pressure signals provided by the pressure sensors disposed closer to the front edge of the seat are indicative of a greater pressure applied thereto than pressure signals provided by the pressure sensors disposed further from the front edge of the seat; and, pressure signals provided by the pressure sensors disposed on the right fork 320R are indicative of a pressure applied thereto which is equal to the pressure signals provided by the corresponding pressure sensors disposed on the left fork 320L; then this is an indication of sitting cross-legged.

In the example shown in Figure 2 and Figure 3, when a user of the system is sitting cross-legged on the seat, the pressure from each pressure sensor may fulfil the inequalities and equalities below: PA > PB > Pc > PE); PE > PF > PG > PH; PA = PE; PB = PF; Pc = PG; PD = PH; 8. Standing up Typically, standing up is characterised by all of the pressure sensors receiving no pressure. In other words, when the pressure signals provided by the pressure sensors are indicative of no pressure (e.g. zero pressure) applied thereto, then this is an indication that the user is standing up (e.g. not sitting on the seat).

-32 -In the example shown in Figure 2 and Figure 3, when a user of the system is standing up, the pressure from each pressure sensor may fulfil the equalities below: PA,B,C,D,E,F,G,H = 0; In examples, instead of or in addition to the relationships between the amount of pressure applied to each pressure sensor (indicated by the magnitude pressure signals), relationships between outputs of given sensors may be used to determine the posture class of a user. For example, pressure sensors (e.g. force sensing resistors) can drift e.g. the sensors may react differently depending on how much a user sits on them. For example, the output of the pressure sensors in response to a given pressure X applied thereto may be different as the pressure sensors age. Determining (at least in part) the posture of a user using relationships between given sensors rather than the magnitude of the pressure signals may allow the drift to be compensated for.

The pressure sensors described herein (e.g. such as pressure sensors 124A to 124H) may comprise force sensing resistors (FSRs). FSRs comprise: a first membrane; a second membrane; and, a spacer element. The spacer element is disposed between the first membrane and the second membrane to provide an air gap therebetween.

In examples, the first membrane and the second membrane have a circular shape and the spacer element has an annular shape. The first membrane comprises two sets of interdigitated electrically conductive traces which are electrically isolated from one another.

-33 -The second membrane comprises a conductive ink disposed thereupon. The first membrane is configured to be moveable into contact with the second membrane in response to a pressure applied to the first membrane.

The FSR is configured to short circuit and provide an electrical impedance thereacross when the first membrane is moved into contact with the second membrane. The FSR is configured to provide an electrical impedance thereacross which is proportional to the pressure applied to move the first membrane into contact with the second membrane.

Figures 4A to 4F illustrate a view of a graphical user interface (GUI) of the posture data sent from a cloud computing platform to a remote device operated by the user.

Figure 4A illustrates a view of a GUI which informs a user of a particular posture which they have been sitting in most frequently. In the example shown in Figure 4A the posture is leaning forward, but the particular posture may be any of the plurality of postures which the system is able to classify (e.g. any of the posture classes). In the example shown in Figure 4A, tips or recommendations are listed on screen which recommend ways in which a user may improve their posture. In this way, the present disclosure provides a feedback loop which may help a user improve their posture.

For example, the present disclosure provides a feedback loop wherein: the posture sensing system obtains data indicative of the user's posture; the posture sensing system sends the data to the cloud computing platform which determines the posture of the user using a classifier algorithm; the determined posture is sent from the cloud computing platform to the remote device; the remote device conveys (e.g. displays on a display of the remote device) recommendations as to how a user may improve their posture; then, the posture sensing system obtains data indicative of the user's posture again and so on.

Figures 4B to 4F illustrates a view of a GUI which informs a user of their historical posture data. The GUI shows the user whether their posture of a previous day met a selected goal. For example, the GUI shows the user what the average number of minutes per hour of good posture (e.g. sitting up) the user exhibited on previous days. The GUI also shows a breakdown (e.g. as a percentage of the total sitting time) of which posture classes the user sat in during a specific period of time (e.g. over hours of a specific day and also over -34 -multiple days).

The GUI may display goal progression for a user e.g. how the user's posture over a period of time compares to a preselected goal. For example, displayed goal progression may comprise displaying how many minutes a day a user sits in a sitting up posture and simultaneously displaying how many minutes a day a user targets to be in a sitting up posture for a day.

The GUI may display a posture rating for a user e.g. how a user's posture over a period of time compares to an ideal posture. For example, the posture rating may be a 'star rating' e.g. a rating out of 5 stars. The ideal posture may comprise sitting up for 5 hours a day. The cumulative period of time which the user is in a sitting up posture in a day may be compared to the 5 hours of sitting up in the case of the ideal posture. The fraction, F, of: the cumulative period of time the user is in a sitting up posture over a day divided by 5 hours (from the ideal case) may be used to determine a star rating. For example, the star rating may be calculated using the formula below: number of stars = 5F The GUI may display tips which encourage a user to perform stretches and/or to perform a particular exercise.

Claims

-35 -CLAIMS: 1. A posture sensing system for determining the posture of a user sitting on a chair, the system comprising: a pressure sensor array comprising a plurality of pressure sensors distributed in a zig-zag pattern across the seat of the chair; a controller coupled to the pressure sensing array, wherein the controller comprises a communication interface; wherein the controller is configured to receive signals wherein the signals are indicative of the pressure from the plurality of sensors of the pressure sensing array, and send these signals to a remote device via the communication interface for determining the posture of the user.
2. The posture sensing system of claim 1 wherein the pressure sensing array comprises two zig-zag arms, each zig-zag arm comprising a plurality of pressure sensors.
3. The posture sensing system of claim 1 or 2 comprising at least one inertial measurement unit, I MU, for determining the orientation of a back of a chair; and wherein the controller is configured to receive sensor signals indicative of the orientation of the back of the chair from the at least one accelerometer and send the signals indicative of pressure and the signals indicative of orientation to the remote device via the communication interface.
4. The posture sensing system of any of the previous claims wherein the communication interface is a short-range wireless communication interface, and wherein the controller is configured to send the signals to the remote device via the short-range wireless communication interface.
5. The posture sensing system of any of the previous claims further comprising the remote device, wherein the remote device comprises a wireless communication interface for receiving signals from the controller, and wherein the remote device is configured to generate posture data wherein the posture data is indicative of the posture of the user based on the received signals from the controller.
6. The posture sensing system of claim 5 wherein the remote device is configured to -36 -send the posture data to a cloud computing platform, OCR, wherein the OCR is configured to provide a graphical user interface to the user to see their posture data as a function of time.
7. The posture sensing system of claim 6 wherein the OCR is configured to determine a posture routine based on a function of the user's posture data over time.
8. The posture sensing system of claim 7 wherein the posture routine comprises a recommended routine to break sedentary sitting and improve the posture of the user.
9. The posture sensing system of any of the previous claims wherein the controller comprises a haptic feedback means, and wherein the haptic feedback system is configured to be triggered in the event that the user is sedentary for longer than a selected threshold period of time.
10. The posture sensing system of claim 9 as dependent on claim 5 or any claim as dependent thereon, wherein the remote device is configured to trigger the haptic feedback means in the event that the remote device determines that the user has been in the same posture for greater than a selected threshold period of time.
11. The posture sensing system of claim 9 as dependent on claim 8 wherein the OCR is configured to trigger the haptic feedback.
12. The posture sensing system of any of the preceding claims wherein the controller is configured to filter the received signals indicative of the pressure from the plurality of sensors of the pressure sensor array before the signals are sent to the remote device.
13. A chair comprising the posture sensing system of any of the preceding claims.
14. A method of determining the posture of a user, the method comprising: obtaining a plurality of signals indicative of pressure from a pressure sensing array comprising a plurality of pressure sensors; sending the set of signals indicative of pressure to a remote device; determining a posture of the user by applying a trained machine learning model to -37 -the set of signals indicative of pressure from the pressure sensing array.
15. A computer program product comprising instructions which, when the program is executed by a computer, cause the computer to carry out the method of claim 14.
16. A method for improving the posture of a user sitting on a chair, the method comprising: determining the posture of a user as a function of time; in the event that the posture of a user has not changed by more than a selected threshold for more than a selected threshold period of time, sending an alert to the user.
17. The method of claim 14 wherein determining the posture of a user comprises: obtaining a plurality of signals indicative of pressure from a pressure sensing array comprising a plurality of pressure sensors; determining a posture of the user by applying a trained machine learning model to the set of signals indicative of pressure from the pressure sensing array.
18. The method of claim 14 or 15 wherein sending an alert to the user comprises at least one of (i) providing haptic feedback to the user, and (ii) providing a notification to the user to alter their posture.
19. The method of improving the posture of a user sitting on a chair according to any of claims 15 to 17, wherein the step of: determining the posture of a user as a function of time; is executed in accordance to claim 14.
20. A method of training a trained classifier algorithm, the method comprising: (i) obtaining pressure signals from the posture sensing system when the user of the system is sat in a known posture; (ii) filtering the pressure signals to obtain filtered pressure signals; (iii) labelling the filtered pressure signals with an indication of the known posture to obtain a list of labelled filtered pressure signals; (iv) repeating steps (i) to (iii) for different known postures to obtain a list of labelled filtered pressure signals; -38 - (v) inputting the labelled filtered pressure signals into an untrained classifier algorithm to generate the trained classifier algorithm.