WO2021253071A1

WO2021253071A1 - System, method and device for monitoring wellbeing

Info

Publication number: WO2021253071A1
Application number: PCT/AU2021/050569
Authority: WO
Inventors: Geoff Ayre; Adam Jahnke
Original assignee: Umps Health Pty Ltd
Priority date: 2020-06-16
Filing date: 2021-06-08
Publication date: 2021-12-23
Also published as: AU2021290445A1; GB202300619D0; GB2612480A

Abstract

A device for tracking at least one user, the device comprising: a housing; at least one socket accessible via the housing, the socket having apertures to receive pins of an electrical plug for connecting at least one load to a primary power source; at least one energy sensor coupled to the housing configured to measure a change in electrical energy use by the load; at least one presence sensor coupled to the housing configured to determine when the user is proximate to the device; and a processor coupled to the housing configured to resolve from at least a portion of signals from the energy sensor and the presence sensor one or more components.

Description

SYSTEM, METHOD AND DEVICE FOR MONITORING WELLBEING

Technical Field

[0001] The present invention generally relates to a system, method and device for monitoring the wellbeing of at least one user.

Background of Invention

[0002] Many older adults prefer to continue to live independently as they age. However, this presents a range of risks. A fall or major illness, if undetected, can have serious consequences. Older adults living alone are often supported by formal or informal carers, but these carers are not always present and may not identify there is a problem until it is too late.

[0003] As of the priority date, in Australia on average, one in three people over the age of 65 will fall each year, often with severe consequences; Australians over 65 are four times more likely to contract pneumonia than the general population and have twice the mortality rate from that illness; Prevalence of coronary heart disease (CHD) is 13 times higher in Australians aged over 75 than in those aged between 45 and 54. CHD is the leading cause of death for older Australians.

[0004] Similar statistics are observed worldwide across a range of health conditions.

[0005] Further, 25% of older Australians live alone, with up to a third reporting that they experience loneliness or social isolation. It follows that there is a significant risk of a person becoming incapacitated by illness or an incident such as a fall and not receiving prompt assistance from others. This can be fatal; lying on the floor for long periods following a fall is associated with a significant increase in mortality rate, for example, and delayed access or lack of access to antibiotics following an infection significantly increases mortality rate.

[0006] Illnesses often cause changes to activity in the home. For example, research has shown that there is a link between CHD and overnight sleep disturbances. Major incidents such as falls will result in a complete absence of normal activity if the person is incapacitated. By passively measuring an analysing a person’s activity in the home, their formal or informal care network can be quickly alerted in the event that person needs help. The more accurately and comprehensively key activities of daily living can be measured, the more quickly a decline in health or wellbeing can be identified.

[0007] Other demographics, such as people with certain disabilities or chronic health conditions, face similar challenges in living independently to older adults.

[0008] A widely used solution consists of an alert button (typically worn as a pendant or fastened around the wrist). The alert button typically sends an alert to an emergency responder when pressed. This solution has a number of shortcomings, for example: The solution requires the user to self-report their need for assistance, and research shows that in up to 83% of cases where a person falls, the button is not pressed — often because the person is unable to access it, does not remember its purpose, or is otherwise reluctant to use it. In practice, if the user lacks insight into their condition they will not press the button. Delirium is a common comorbidity with a range of serious health conditions in older adults and can result in the user being unable to recognise their need for assistance and activate alert systems. Dementia and cognitive impairment can similarly affect the user’s insight into their condition. Users are often reluctant to wear alert buttons, as they are perceived as an indicator of weakness and vulnerability; Some users also report high false positive rates, as alerts can be activated when the user inadvertently leans or sits on the button.

[0009] Other solutions include wearable devices with biometric sensors and/or accelerometers. These solutions can provide accurate insight into the user’s health and wellbeing and can detect falls without the user’s involvement, but also have a number of shortcomings, for example: users commonly do not wear the device at home — often because the device is perceived as unnecessary or intrusive, or simply because the user forgets to put it on; Batteries must be recharged frequently; Users with dementia or cognitive impairments may struggle to ensure the battery is always adequately charged; Many devices are not waterproof, meaning they cannot be used in the shower or bath. Research shows that more than 10% of falls occur in the bathroom, and these falls are more than twice as likely to result in an injury. [0010] Newer solutions involve passive monitoring of activity levels using combinations of motion sensors and door sensors. These address many of the shortcomings of the above described solutions, but also have a number of issues. A major weakness is that the data collected by these sensors gives an indication of overall activity levels in the home, but by itself is not sufficient to accurately identify activities which are indicators of health and wellbeing. For example, while a motion sensor can detect that the user is in the kitchen, it cannot detect whether they are preparing dinner or wandering around the room in confusion. This limits the system’s ability to accurately identify problems and raise timely alerts. Some widely used motion sensors are also prone to false positives; most solutions use passive infrared sensors, which can be triggered by draughts of warm or cool air, sunlight or bright reflections through windows, or warm objects such as radiators. Motion sensors are also highly susceptible to false positives from pets. The sensors require periodic battery changes, and since the sensors are generally located in high places to maximise their field of view, installation can be an inconvenient process for the user or their carer.

[0011] Other solutions involve the use of cameras to detect activity around the home. The video data from the cameras can by monitored by a human, an algorithm (e.g., computer vision), or some combination of the two. Camera-based systems can identify activity around the home and detect when the user needs help very quickly. However, such systems may compromise the user’s privacy and are generally perceived as highly intrusive.

[0012] Other solutions employ “smart plugs” to identify activity in the home and raise alerts to carers if the user needs assistance. Smart plugs are connected between a wall outlet and an appliance and detect usage of the appliance by measuring electrical current. Smart plug-based solutions can detect some activities around the home with high accuracy and a low false positive rate, but still have several issues: Smart plugs generate relatively sparse data, as an appliance will rarely be used more than a few times per day. This means that an absence in activity following an incident may take several hours to detect; The physical size of the smart plugs often prevents installation in some wall outlets and prevents the use of adjacent power outlets. Smart plugs can also be easily removed or inadvertently knocked out of the outlet by users; this can often occur if the user has impaired memory or cognition and does not remember the purpose of the plug. [0013] It would be desirable to provide a device and system which ameliorates or at least alleviates one or more of the above problems or to provide an alternative.

[0014] It would also be desirable to provide a device and system that ameliorates or overcomes one or more disadvantages or inconvenience of known systems for monitoring users.

[0015] A reference herein to a patent document or other matter which is given as prior art is not to be taken as an admission or a suggestion that the document or matter was known or that the information it contains was part of the common general knowledge as at the priority date of any of the claims.

Summary of Invention

[0016] According to an aspect of the present invention, there is provided a device for tracking at least one user, the device comprising: a housing; at least one socket accessible via the housing, the socket having apertures to receive pins of an electrical plug for connecting at least one load to a primary power source; at least one energy sensor coupled to the housing configured to measure a change in electrical energy use by the load; at least one presence sensor coupled to the housing configured to determine when the user is proximate to the device; and a processor coupled to the housing configured to resolve from at least a portion of signals from the energy sensor and the presence sensor one or more components.

[0017] In one or more embodiments, the processor is configured to resolve from the signals by way of a Fourier transform one or more frequency components each corresponding to electrical energy use associated with the user and proximity of the user to the device.

[0018] In one or more embodiments, the processor is configured to extract features from the energy sensor and the presence sensor components corresponding to a pattern of electrical energy use associated with the user and proximity of the user to the device.

[0019] In one or more embodiments, the housing comprises a general AC power outlet. The general AC power outlet may be configured to be mounted in a wall. [0020] In one or more embodiments, the device includes at least one switch coupled to the housing to selectively control the primary power source. The switch may be operated by the processor to selectively control the primary power source. Advantageously, in this particular form of the invention the outlet may be controllable and/or monitorable via a wired or wireless network (e.g., “a smart power outlet”).

[0021] In one or more embodiments, the switch comprises a solid state device. The solid state device may include a bidirectional triode thyristor (triac). The switch may be configured to have an electrically-open state and an electrically-closed state whether electromechanical or solid state.

[0022] In one or more embodiments, the device comprises a communication module, coupled to the housing. The communication module may be wireless. For example, communication can be carried out using any suitable communication protocols, including, but not limited to Wi-Fi 802.11 , 6LowPan/ZIGBEE™ 802.15, Ethernet 802.3, 802.11 and 802.15.4. It will also be appreciated that the communication can be carried out using Ethernet-over-Power to and from the devices, e.g., GPO power sockets or light fittings, and through the powerline itself.

[0023] In one or more embodiments, the processor is configured to transmit via the communication module to an analytics component, at least one of: at least a portion of signals from the energy sensor and the presence sensor resolved into one or more components; and features extracted from the energy sensor and the presence sensor components corresponding to a pattern of electrical energy use associated with the user and proximity of the user to the device.

[0024] In one or more embodiments, the analytics component is configured to extract features from the energy sensor and the presence sensor components corresponding to a pattern of electrical energy use associated with the user and proximity of the user to the device.

[0025] In one or more embodiments, the analytics component comprises a cloud- based host system configured to identify data indicative of interactions considered unusual between the user and the device based on the pattern. Unusual patterns may include a user missing a habitual or commonly occurring activity, such as turning on the kettle every morning. [0026] In one or more embodiments, the analytics component comprises a locally- based host system configured to identify data indicative of interactions considered unusual between the user and the device based on the pattern.

[0027] In one or more embodiments, the device further comprises at least one microphone coupled to the housing. The microphone converts sound into an electrical signal, including, but not limited to dynamic microphones, condenser microphones, piezoelectric microphones, fibre-optic microphones, laser microphones (e.g., a laser beam aimed at the surface of a window or other plane surface that is affected by sound) and MEMS microphones.

[0028] In one or more embodiments, the devices further comprises at least one sensor coupled to the housing selected from a group consisting of an ultrasonic sensor, a radar sensor, a motion sensor, an impact sensor, a camera, an infrared sensor, a capacitance sensor, a temperature sensor, a light sensor, a chemical sensor, a humidity sensor, a vibration sensor, a pressure sensor, an electrical field sensor, a bio sensor, a particulate sensor, a volatile organic compound sensor, a carbon monoxide sensor, a combustible gas sensor, and a carbon dioxide sensor.

[0029] According to another aspect of the present invention, there is provided a system for tracking at least one user, the system comprising: an analytics component configured to receive and store information from a plurality of devices, the plurality of devices comprising: a housing; at least one socket accessible via the housing, the socket having apertures to receive pins of an electrical plug for connecting at least one load to a primary power source; at least one energy sensor coupled to the housing configured to measure a change in electrical energy use by the load; at least one presence sensor coupled to the housing configured to determine when the user is proximate to the device; and a processor coupled to the housing configured to interact with the analytics component to extract features from the energy sensor and the presence sensor corresponding to a pattern of electrical energy use associated with the user and proximity of the user to the device, and identify signals indicative of interactions considered unusual between the user and the plurality of devices. [0030] In one or more embodiments, the plurality of devices include a communication module, coupled to the housing. The communication module may be wireless.

[0031] In one or more embodiments, the processor is configured to transmit via the communication module to an analytics component, at least one of: at least a portion of the signals from the energy sensor and the presence sensor resolved into one or more components; and features extracted from the energy sensor and the presence sensor components corresponding to a pattern of electrical energy use associated with the user and proximity of the user to the device.

[0032] In one or more embodiments, the analytics component is configured to extract features from the energy sensor and the presence sensor corresponding to a pattern of electrical energy use associated with the user and proximity of the user to the device.

[0033] In one or more embodiments, the analytics component comprises a cloud- based host system configured to identify data indicative of interactions considered unusual between the user and the device based on the pattern.

[0034] In one or more embodiments, the analytics component comprises a locally- based host system configured to identify data indicative of interactions considered unusual between the user and the device based on the pattern.

[0035] In one or more embodiments, the plurality of devices are homogeneous and cooperate with each other.

[0036] In one or more embodiments, the presence information includes an indication that the user has entered close proximity to the plurality of devices and an indication that the user has left close proximity to the plurality of devices.

[0037] In one or more embodiments, the change in energy usage measured by each of the plurality of devices includes a change in electrical power consumption by each of the loads.

[0038] According to another aspect of the present invention, there is provided a method of tracking at least one user, the method comprising: providing a plurality of devices for connecting a plurality of loads to a primary power source; receiving energy information from a plurality of energy sensors configured to measure a change in electrical energy use by the plurality of loads; receiving presence information from a plurality of presence sensors configured to determine when the user is proximate to the plurality of devices; extracting features from the energy information and the presence information; creating a pattern of electrical energy use associated with the user and proximity of the user to the plurality of devices; characterising interactions considered unusual between the user and the plurality of devices based on the pattern.

[0039] According to another aspect of the present invention, there is provided a system for tracking at least one user, the system comprising: an analytics component configured to receive and store information from a plurality of devices, the plurality of devices comprising: a housing including at least one of a general AC power outlet or a gang switch; at least one microphone coupled to the housing configured to receive a sound; a communication module, coupled to the housing and configured to communicate the sound with the analytics component.

[0040] In one or more embodiments the communication module may be wireless.

[0041] In one or more embodiments, the analytics component includes a multilateration component for determining a location of the sound using differences in received sound pressure level (SPL) relative to the plurality of devices.

[0042] In one or more embodiments, the analytics component incudes a time of arrival component for determining differences in arrival time of the sound at each of the plurality of devices.

[0043] In one or more embodiments, the analytics component includes a natural language processing (NLP) component for determining voice commands.

[0044] In one or more embodiments, the multilateration component is configured to iteratively update the location of the sound via adaptive filtering.

[0045] In one or more embodiments, the multilateration component includes an adaptive Kalman filter that uses the SPL to update and maintain current and previous locations of the sound and propagate current location to the analytics component. [0046] In one or more embodiments, the analytics component is further configured to extract features from the sound corresponding to a pattern of sounds associated with the user, wherein the features characterize interactions considered unusual between the user and the device based on the pattern.

[0047] In one or more embodiments, the features characterise distress. Distress may include a state of physical strain, difficulty and needing help.

[0048] In one or more embodiments, the sounds associated with the user includes sounds associated with appliances.

[0049] In one or more embodiments, the sounds associated with the user includes sounds associated with an event, including the acoustic signals associated with household fixtures.

[0050] In one or more embodiments, the analytics component comprises a cloud- based host system configured to identify sounds indicative of interactions considered unusual between the user and the device based on the pattern.

[0051] In one or more embodiments, the analytics component comprises a locally- based host system configured to identify data indicative of interactions considered unusual between the user and the device based on the pattern.

[0052] According to another aspect of the present invention, there is provided a method of tracking at least one user, the method comprising: receiving sound information from a plurality of devices; extracting features from the sound information; creating a pattern of sounds associated with the user; characterising interactions considered unusual between the user and the plurality of devices based on the pattern.

Brief Description of Drawings

[0053] The invention will now be described in further detail by reference to the accompanying drawings. It is to be understood that the particularity of the drawings does not superseded the generality of the preceding description of the invention.

[0054] FIG. 1 shows a simplified block diagram of a single device in accordance with an embodiment of the invention; [0055] FIG. 2 shows a block diagram of a single device in accordance with an embodiment of the invention;

[0056] FIG.3 shows an example of a sensor system in accordance with an embodiment of the invention;

[0057] FIG.4 shows four examples of a sensor system in accordance with an embodiment of the invention;

[0058] FIG. 5 shows a method implemented by a device or a plurality of devices in accordance with an embodiment of the invention;

[0059] FIG.6 shows an example of a sensor system in accordance with an embodiment of the invention; and

[0060] An example of a sensor system in accordance with an embodiment of the invention is shown in FIG. 7.

Detailed Description

[0061] The invention is suitable for analysing the activity of a person. In an embodiment, the activity is analysed in a house, and it will be convenient to describe the invention in relation to that exemplary, but non-limiting, application.

[0062] The term “house”, as used herein, may refer to a house, an apartment, a unit, a care facility, an independent living unit, a specialist disability accommodation facility, a nursing home, or a hospital and the like. In a care facility, a nursing home, or a hospital and the like the person need not be a permanent resident. For example, the invention is also suitable for determining if a carer has visited a person at a particular time.

[0063] An example of a single device in accordance with an embodiment of the invention is shown in FIG. 1. The device 100 includes a housing 102 and a socket 104 accessible via the housing 102. The socket 104 includes three pins 106 for connecting a load 118 (e.g., an appliance) to a primary power source 108 via power lines 116. It should be appreciated that the housing 102 and socket 104 may take the form of a general purpose alternating current (AC) power outlet (GPO), and those skilled in the art will recognise suitable designs for providing the stated functions. [0064] The GPOs may be supplied as a kit, or they may be installed in a house either at the time the house is built or retrofitted to existing wiring. Generally, more devices will be installed in larger houses than smaller houses. In a multi-room care facility, for example, it is envisaged that several hundred or several thousand devices will be installed.

[0065] A GPO may also be referred to as a power point, socket, outlet, wall outlet, wall mounted socket etc., and include one or more sockets for supplying electrical power, via a removable plug, to one or more loads. Sockets are generally used to supply AC electric power (e.g., 220V to 240V, 50Hz; 120V, 60Hz - unless otherwise indicated, references to AC voltages are understood to refer to substantially sinusoidal voltages, and voltage amplitudes are understood to refer to root mean square (RMS) values) via three apertures adapted to receive pins of the plug. Each aperture is separately connected to an active line, a neutral (returning) line and ground wire of a primary power source (e.g., mains power), although in some cases the ground connection may be omitted. Typically, sockets have three apertures to accommodate either two pin or three pin plugs.

[0066] Generally, the plug is the movable connector attached to a load, and the socket is a fixture on equipment or a building structure (e.g., a wall). Plugs have male circuit contacts, while sockets have female contacts. The plug has pins that fit into matching apertures in the socket. The geometrical arrangement of the socket and plugs varies from country to country according to the relevant national standard. For example, Australia New Zealand Standard AS/3112 defines two flat current-carrying blades orientated at 30° to the vertical to form an upside-down V-shape and a flat vertical grounding blade, as shown in FIG. 1 .

[0067] GPOs may also be fitted with a switch for switching or toggling the supply state of power to a socket (and plug) between on and off states. Inclusion of a switch (or switches) allows a user to selectively control power to a device, independent of any power switch contained within the device. The use of switches in GPOs is common in Australia.

[0068] The device 100 also includes an energy sensor 112 and a presence sensor 110 disposed within the housing 102. The energy sensor 112 and the presence sensor 110 are electrically connected to a processor 114 as shown and powered by the primary power supply 108. Suitable AC-to-DC transformers (not shown) may be employed, as necessary to deliver DC power at a specified voltage to the processor 114 and sensors 112, 110. For example, step-down transformers may transform AC power from high voltage levels suitable for transmission to levels that can be substantially directly applied to the processor 114. The energy sensor 112 is also electrically connected to the socket 104 to monitor electrical usage associated with the load 118.

[0069] In a number of embodiments, the energy sensor 112 is a current sensor coupled to the socket 104 so as to be operable to determine the current being drawn from the socket 104 by the load 118. A current amplifier may be used to amplify the output signal of the current sensor. The amplified signal is then received by the processor 114 and is compared with a threshold level. The threshold level will typically be set as a range of ampere, so that the current drawn by the appliance will be above the threshold only when the appliance is in use. When the appliance enters a state of use, the current drawn will rise above the threshold level, and the processor 114 may record a timestamp or similar against the activity in memory 120. While the appliance is in use, the processor continues to compare current draw with the threshold value. When the appliance enters a state of non-use, the current will fall below the threshold and the processor 114 will similarly record the event in memory 120.

[0070] In some embodiments, the detection of usage events may be achieved by comparing the first derivative of the current with a threshold; i.e. logging an event when the rate of change exceeds a predefined value. This technique is more robust in cases where the total current consumption of the load when the device is in use is unknown.

[0071] By way of non-limiting example, the current sensor 112 may be able to sense, a sharp increase in the detected flow of electrical current through the outlet 104 connected to a kettle so as to determine the activity “turning on the kettle”.

[0072] In a number of embodiments, the presence sensor 110 is a passive infrared detector (PIR). Presence sensors of this kind can be used, for example, as an occupancy detector to record when someone is present in the room. During operation, the presence sensor 110 monitors an area, such as the kitchen of a home. Specifically, the presence sensor 110 detects a change in the infrared energy radiating from regions in the area monitored by different sensing lobes of the PIR detector, which generally have a pass band within the 8-14 pm infrared range. If a person enters the monitored area, the person changes the amount of infrared energy being detected by one or more sensing lobes of the PIR detector. The difference in detected infrared energy at the sensing lobes is measured and compared with a predefined trigger threshold; if the difference in detected energy exceeds the threshold, a signal is provided to the processor 114. The processor 114 interprets the signal provided by the presence sensor 110, it being understood that other circuitry may be employed, depending on the preferences of the system designer including digital and/or analog logic. Moreover, while the presence sensor 110 has been described as a PIR detector, other types of presence sensors, such as ultrasonic sensors, radar sensors, motion sensors, microphones, cameras, infrared sensors, temperature sensors, light sensors, vibration sensors, electrical field sensors, may be used, if desired. In addition, while the device 100 has been described with reference to a single presence sensor 110, it is understood that the invention may be implemented using multiple sensors at the same or different locations in a room (including sensors remote from the housing 102). Employing sensors remote from the housing 102 may be advantageous in situations where the GPO is obstructed by an appliance, for example, a refrigerator or microwave. In this particular embodiment, the sensors may be mounted on the roof or adjacent to the appliance and connected wirelessly or by an auxiliary cable to the GPO.

[0073] Advantageously, the use of purely passive monitoring technologies means the system can raise an alert even if the user is incapacitated or lacks insight that they require help. The solution can raise an alert even if the user is unconscious or delirious. Moreover, the use of multiple sensors enables the system to detect activity throughout the home, rather than relying on limited data from a single location.

[0074] The processor 114 is configured to resolve from at least a portion of signals from the energy sensor 112 and the presence sensor 110 into one or more components relating to the load 118 or proximity of a user to the device 100. For example, the processor 114 may resolve from the signals by way of a Fourier transform one or more frequency components each corresponding to electrical energy use associated with the load 118 and proximity of the user to the device 100. It should also be appreciated that additional filtering methods can be used, such as those, but not limited to including, moving average filters, evenly weighted moving average filters, the like, or a combination of these filters, which may be particularly suitable for implementation in firmware.

[0075] In a number of embodiments, the processor 114 converts the signals from the energy sensor 112 and the presence sensor 110 into digital values and communicates messages based on those digital values to another device via powerlines 116 (e.g., data power-line networking) or saves them in memory 120 for retrieval by an external device.

[0076] Advantageously, by combining appliance usage data (derived from energy sensor 112) with presence sensor data (derived from presence sensor 110) the present invention can accurately detect specific activities of daily living and quickly identify abnormalities in activity levels. The low false positive rate of the appliance usage data combined with the high sensitivity of the presence sensor enables the activity to be detected and classified with greater accuracy than known solutions. The inclusion of additional sensors in the device 100 can further improve its ability to detect and classify activity as discussed with reference to FIG. 2.

[0077] The integration of the sensors 110, 112 into a GPO means that batteries are not required, removing the significant ongoing maintenance requirements of known motion sensor-based solutions. Since there are power outlets in almost every room of a typical home, the present invention provides a convenient way to outfit a home with a comprehensive sensor network with minimal additional wiring, installation costs, and visible hardware.

[0078] The device 100, designed to be positioned in a building structure (e.g., a wall), comprises a sensor housing 102 that allows for environmental damage resistance and those skilled in the art will recognises suitable materials for providing the stated functions, for example: elastomeric plastics materials such as polycarbonate and polytetrafluorethylene. It is generally preferred that the device 100 is provided as a sealed or substantially sealed module to limit or substantially preclude water or particle ingress (so that it can be installed in bathrooms) to the module. Mechanical isolation of the processor 114 or sensitive components, including the presence sensor 110, may also be provided, including resilient suspension means, or gel pads formed of slow response silicone or urethane foam, or the like. [0079] An example of a single device in accordance with an embodiment of the invention is shown in FIG. 2. The device 200 includes a housing 102 and a socket 104 accessible via the housing 100. The socket 102 includes three pins 106 for connecting a load 118 (e.g., an appliance) to a primary power source 108 via power lines 116. It should be appreciated that the housing 102 and socket 104 may take the form of a general purpose alternating current (AC) power outlet (GPO) as discussed with reference to FIG. 1 .

[0080] In the embodiment shown, the housing includes a mechanical switch 202 coupled to the socket 104. Inclusion of a switch (or switches) allows a user to selectively control power to the load (i.e., an appliance) 118, independent of any power switch contained within the load 118. The switch may be a mechanical switch which comprises mechanical contacts which open or close in response to manual activation. Such devices may include simple manual switches, push button switches, mechanical throws, knobs, toggles or dollys, rockers, dials, triggers or the like, with “on” and “off” settings.

[0081] The device 200 also includes a relay 204 electrically connected to the socket 104 and processor 114 to remotely supply power from the primary power supply 108 to the load 118 without user 210 intervention. The processor 114 may provide a low voltage control signal to control the relay 204, which is capable of handling and switching very high-voltage or very high-power circuits. The relay may include a protective diode connected in parallel with the primary power supply 108. The diode may be electrically connected across the relay coil to avoid back EMF created when the relay coil switches off. The relay 204, is a conventional mechanical relay. It should be appreciated that other switching means may be employed to switch load 118, such as, but not limited to: solid state relays (SSRs) with optical isolation, i.e., employing an internal optical coupler in combination with a semiconductor element to switch power; solid state relays with transformer isolation; reed relays; thyristors; and, field effect transistors (FETs), i.e., FET and MOSFET switches for relay-type functions.

[0082] In the embodiment shown the processor 114 is electrically connected to an antenna 208 and a wireless transceiver 206. The processor 114 may employ firmware or other circuitry for continuously monitoring the sensors 110, 112, 204 to n and controlling activation and deactivation of the antenna 208 accordingly. [0083] In addition to the energy sensor 112 and a presence sensor 110 disposed within the housing 102, a number of other sensors n may be used to assist in detecting specific activities or environmental conditions which could affect the user’s wellbeing including an ultrasonic sensor, a radar sensor, a motion sensor, an impact sensor, a camera, an infrared sensor, a capacitance sensor, a temperature sensor, a light sensor, a chemical sensor, a humidity sensor, a vibration sensor, a pressure sensor, an electrical field sensor, a bio sensor, a particulate sensor, a volatile organic compound sensor, a carbon monoxide sensor, a combustible gas sensor, and a carbon dioxide sensor.

[0084] As a non-limiting example, a vibration sensor (or microphone) may be disposed on a walking surface and in communication with the processor 114 to resolve from a signal representing vibrations by one or more footsteps date representative of whether the user 210 is proximate to the device 200. It will be appreciated, to identify steps or the vibrations associated with a fall, such events need to be resolved from the raw data containing the entire vibration signal by the processor 114. The noise, for example, may be modelled as a Gaussian distribution, with an anomaly detection method used to extract step or fall events.

[0085] As another non-limiting example, a radar module consisting of one or more transmitter and one or more receiver antennae may be used with the processor 114 configured to process signals reflected from the user 210. A continuous sinusoidal electromagnetic wave (typically with a frequency of 30GHz - 300GHz) is broadcast by at least one transmitting antenna, and the reflected signal is received by at least one receiving antenna. If the user 210 is in front of the sensor, the electromagnetic wave will reflect from their body or clothing. Any movement of the reflector will result in a Doppler shift in the reflected wave; the frequency of the reflected signal will vary slightly from that of the original signal. By mixing the original and received signals using a nonlinear electronic component such as a diode and filtering the result to remove high- frequency components (heterodyning), this Doppler shift (and hence, the user’s instantaneous movement speed) may be accurately measured. Frequency domain analysis of this data enables cyclic movements such as the human heartbeat and respiration to be readily identified; this enables highly accurate detection of the presence of a human being, and can provide valuable additional information about their health and wellbeing. For example, an irregular or unusually fast heartbeat can be identified.

[0086] As another non-limiting example, a microphone may be used to analyse sounds from a shower head, bath or skin faucet, water supply piping and the like. It will be appreciated, this may provide an indication of bathing activity if a device 200 is positioned in a bathroom. Similarly, a humidity sensor may be used, the output of which may be analysed together with the presence sensor (e.g., a user remained in the bathroom for a period of time indicative of a bath or a shower). To differentiate between bathing and acts associated with increased humidity in the bathroom, like washing the hands and the face with hot water or shaving, the device 200 (and presence sensor) may be placed in front of the sink, for example. In this example, a threshold may also be set on the time a user typically spends in the shower. It will be appreciated that if a humidity sensor output is high for thirty minutes, when a user typically only spends eight minutes in the shower, the shower activity may be categorised as unusual.

[0087] As another non-limiting example, a microphone may receive voice commands as well as passively tracking a user’s location using sound, as will be discussed with reference to FIG. 6 and FIG. 7.

[0088] Similarly to most embodiments, in the above described examples it is envisaged that signal processing is employed for radar and vibration signal analyses including sequential return signal integration, Fast Fourier Transform (FFT), and signal correlation. In each example, the result of the signal processing is to determine the activity of a user.

[0089] In a number of embodiments, the antenna 208 may be a patch antenna arranged in sensor housing 102. In another embodiment, the antenna 208 may be a simple dipole antenna with a radiation pattern in a desired alignment with the structure (i.e., the wall) the device 200 is installed in. For example, the radiation pattern of the antenna 208 may be toroidal with respect to a common axis generally perpendicular to the sensor housing 102. The antenna 208 may also include a shield (not shown) to provide attenuation of electromagnetic interference (EMI) generated by the sensors 110, 112, 204 to n, processor 1 14, the relay 204, and EMI generated from exterior sources such as the primary power supply 108 or the load 118. [0090] In a number of embodiments, the processor 114 also converts the sensor 110, 112, 204 to n readings into digital values and communicates messages based on those digital values to the wireless transceiver 206. It should also be appreciated that additional filtering methods can be used, such as those, but not limited to including, moving average filters, evenly weighted moving average filters, the like, or a combination of these filters, which may be particularly suitable for implementation in firmware.

[0091] The wireless transceiver 206 is also adapted to facilitate communication between a remote firmware update mechanism and the processor 114. As will be appreciated by those skilled in the art, the remote firmware update mechanism together with the processor 114 may be adapted to periodically check for updates from a remote repository, download firmware updates and to compare downloaded firmware to existing firmware to determine the necessity of installing the downloaded firmware and the like.

[0092] In a number of embodiments, the device 200 may be provisioned with position information. For example, the location of the device 200 may be stored in memory 120 during an initialisation process (e.g., device 1 = kitchen, device 2 = bathroom).

[0093] Advantageously, sensor data collected by the sensors 110, 112, 204 to n in the device 200 may be pre-processed locally. This may involve filtering the data to remove noise, performing statistical analyses to identify patterns or trends, or applying mathematical transforms to the data (e.g. applying a Fourier transform to enable frequency domain analysis). The data may then be transmitted via wireless transceiver 206 or to another device via powerlines 116 (e.g., data power-line networking) for processing by a cloud computing platform or directly by the base station as also discussed with reference to FIG. 3. Pre-processing data may also avoid some of the costs traditionally associated with sending data through a mobile network (if employed).

[0094] The remaining elements shown in FIG. 2 are identical to FIG. 1 and so share the same references.

[0095] An example of a sensor system in accordance with an embodiment of the invention is shown in FIG. 3. The sensor system 300 includes a plurality of devices 316 positioned in a house 302. Each device 316 is wirelessly connected to an access point 320 or another device acting as an access point 318. In a number of embodiments, the wireless connection is part of a mesh network, such as an IEEE 802.15.4 mesh network. However, it will be appreciated that the communication can be carried out using any suitable communication protocols, including, but not limited to Wi-Fi 802.11 , 6LowPan/ZIGBEE™ 802.15, Ethernet 802.3, 802.11 and 802.15.4. It will also be appreciated that the communication can be carried out using Ethernet-over-Power to and from the devices, e.g., GPO power sockets or light fittings, and through the powerline itself.

[0096] A mesh network is typically characterised by a local network topology in which the nodes connect directly, dynamically and non-hierarchically to as many other nodes as possible and cooperate with one another to efficiently route messages 338 to or from clients, such as base station 332 further described below. Each device 316 can act as a repeater extending the range of the network and can reconfigure to accommodate the failure of another device 316 in the network. In other embodiments, other wired or wireless networking topologies that provide for bi-directional communication between the devices 316 and the access point 320 may be used to connect the devices to an access point 320 or to a device acting as an access point 318.

[0097] The sensor system 300 includes one or more servers 328, 330 communicatively coupled to a cloud computing environment 310, “the cloud”.

[0098] Although “the cloud” has many connotations, according to embodiments described herein, the term includes a set of network services that are capable of being used remotely over a network, and some of the processes described herein may be implemented as a set of instructions stored in a memory and executed by a cloud computing platform 310. It will be appreciated by those skilled in the art, that a number of platforms may be used to host the set of instructions as a software application on a number of virtual machines 310, each on a corresponding virtual hard drive 322. The software application may provide a service to one or more servers 326 or support other software applications provided by a third party servers 330. Examples of services include a website (e.g. a dashboard with a person’s activity statistics conveniently displayed), a Network Time Protocol (NTP) server, a database, a geographic information service (GIS), a business intelligence or data visualisation service, or other web services. One or more routers 328 may be used to communicate information from the cloud to one or more remote users 334 (e.g., relatives or carers) equipped with base stations 332. Base stations 332 may include any one or more of a mobile communication device, tablet, desktop computer or the like. Information routed through one or more routers 328 is passed through a firewall 324 to a packet switching network, such as, the Internet. Information may be wirelessly routed to remote users directly through a network WLAN 336 and not require the Internet. In a number of embodiments, messages to and from each of the devices 316 may be transmitted via a cellular network 320 to a central data repository, for example on virtual hard drive 322 in a cloud computing environment 110.

[0099] Advantageously, using a cellular network 320 (e.g., via GSM, GPRS, EDGE, UMTS, W-CDMA, LTE, CDMA, TDMA, FDMA, EVDO, CDMA2000, UMB and WIMAX protocols) can simplify setup and the costs associated with a fixed internet connection (e.g., DSL or cable). However, it will be appreciated that a fixed internet connection may be used.

[00100] Throughout the description there are references to base stations 332 as example devices where information is moved to and from. In other embodiments, the implemented devices are computers, smartphones, tablets, laptop computers, desktop computers, server computers, among other forms of computer systems. In a number of embodiments, the base station 332 includes an analytic component for the storing, sorting and analysis of messages from device 316. That is, the base station 332 can process data directly without relying on the cloud 310, for example.

[0101] In a number of embodiments, the base station 332 is positioned in a convenient location in the house 302. Advantageously, the base station 332 directly receives data from one or more devices 316 (and optionally additional sensors in the home) and runs a software application in the analytics component locally. This software analyses the raw (or pre-processed) sensor data from one or more sensors and generates activity data, which describes specific activities the user 334 is performing around the home. [0102] In a number of embodiments, the access point 320 or device acting as an access point 318 is connected to base stations 332 via the cloud computing platform or the Internet 310 with a wireless connection such as an 802.11 Wi-Fi network or Bluetooth™. The sensor systems 300 is configured using a client application that enables users 334 to exchange information with each device 316 or group of devices 316a via the access point 320 or device acting as an access point 318. Advantageously, this may allow for a group of devices 316a to be configured in the same way, for example two devices in the bathroom 308 of the house 320 may each have the same identifier (e.g., device 1 = bathroom, device 2 = bathroom).

[0103] In a number of embodiments, the client application enables the user 334 to bind each device 316 or group of devices 316a to other services (i.e., implemented on servers 362, 330, for example). Once bound, the sensor system 300 can be configured with a baseline activity pattern based on, without limitation, the location within the house 302, time of day, or type of activity. The activity pattern can be used to identify patterns of activity from each device 316 to identify data indicative of interactions considered unusual between the user and the device based on that pattern.

[0104] In other embodiments, the user 334 (e.g., a carer) may independently collect activity information and make the information available to the devices 302 via the base station 332. For example, the user 334 may provide an indication of how lethargic the monitored person appears, or other information and the activity pattern may be adapted in order to yield better results (e.g., earlier detection of unusual patterns of activity, or fewer false positives).

[0105] Various alert system 336 control implementations are possible. In general, these implementations will issue alerts based on the similarity or dissimilarity of recent patterns of activity to stored activity patterns. In a number of embodiments, one or more algorithms may be used to detect the presence or absence of features in the data corresponding to aspects of everyday independent living, and calculate a score indicating the degree of abnormality relating to that feature. For instance, one or more algorithms may be used to identify disturbances in sleep patterns and return an overall ‘sleep disturbance score’ where 0.0 indicates no disturbance and 1.0 indicates significant disruption to normal sleeping routines, based on the data collected by multiple sensors throughout the home. A second algorithm may detect changes in appetite based on fridge and microwave usage, and a third may detect increases in sedentary behaviour using presence and television usage data. The scores from a number of such algorithms (not limited to the above examples) may then be input to a mathematical function (for example, a weighted average of the scores) and the result compared to a threshold, with an alert being raised if the result is greater than the threshold. In other embodiments, machine learning may be used to identify unusual patterns of activity. For instance, a recurrent neural network may be trained based on previously recorded patterns of activity, and used to identify and classify new data as it is received. In a number of embodiments, these algorithms are run on virtual servers hosted in the cloud 310 based on activity data stored in cloud databases 322. In other embodiments, some or all of these algorithms are run on a processor embedded in the access point 318.

[0106] It will be appreciated that an alert may be a notification on a base station, for example a notification on a smartphone or tablet, or an audible alarm or light on the base station itself.

[0107] Four examples of a sensor system in accordance with an embodiment of the invention are shown in FIG. 4. FIG. 4 generally shows three simplified versions of the embodiments discussed with reference to FIG. 3.

[0108] Referring firstly to FIG. 4 (a), the sensor system 400 includes a plurality of devices 402, for instance device 200 with reference to FIG. 2. Each device 402 is connected to a base station 404, which is in turn connected to “the cloud” 406. The cloud 406 is part of an TCP/IP network adapted to issue alerts 408 to a relative or carer.

[0109] As discussed above, activity data can be determined by the processor within the device 402 or within the base station 404 in an analytics component. Software or firmware in either device 402 or 404 may compare data from multiple sensors across each device 402 with expected activity patterns associated with specific activities. For example, an increase in temperature, increase in humidity and increase in power usage consistent with the use of an electric hair dryer transmitted from one or more devices 402 in a bathroom may be associated with the activity “showering”.

[0110] In the embodiment shown, the activity data is then transmitted to the cloud (e.g., to a virtual computer hosted in a cloud computing platform). Additional software may be run using the activity data (and optionally other sources of data) as an input. The data may be compared with expected patterns of which may be stored in memory or algorithmically derived from historical activity data.

[0111] If the activity data input matches certain patterns of behaviour associated with incidents or illnesses, the additional software may generate an alert. By way of non-limiting example, if the activity data shows that a user has used the bathroom significantly more than usual, the additional software may generate an alert as the present data matches activity patterns associated with a urinary tract infection.

[0112] Additionally, if the detected activity data does not match expected patterns of behaviour associated with high wellbeing and stable independent living the additional software may generate an alert. By way of non-limiting example, if the activity data shows that there has been no activity at a time when the user typically cooks dinner, eats, and turns on the TV, the additional software may generate an alert as the present data does not match expected patterns derived from historical data.

[0113] If an alert is generated, a message will be transmitted to another piece of software, another computer, or directly to a person, and those skilled in the art will recognise method of providing the stated function, including push notifications, SMS messages, emails, and the like.

[0114] Referring to FIG. 4 (b), the sensor system 400 includes a plurality of devices 402, for instance device 200 with reference to FIG. 2. Each device 402 is connected directly to base station 404. The bases station 404 is adapted to issue alerts 408 to a relative or carer.

[0115] In this embodiment, the activity analysis may be performed within the devices 402. Activity data is then transmitted from the outlets to the base station, where further analysis may be performed with additional software in the analytics component (as described with reference to FIG. 4 (a)).

[0116] In embodiments where at least one base station 404 is located within the user’s home, an alert indicator may be provided to indicate to the user when an alert has been raised. This could involve, for example, a flashing light and an alarm tone. In cases where this base station 404 is not responsible for analysing activity and generating alerts, the alert indicator is triggered by a message from another computer.

[0117] In one or more embodiments, 408 is a device located with a carer or family member. The alert indicator is intended for the user and positioned within their house. Typically, the alert indicator will include an alarm tone and/or flashing light on the base station (or hub) 404. However, it will be appreciated that the alert indicator may be a separate device. Optionally, there may also be provided one or more inputs on the hub 404 to enable the user to cancel and/or acknowledge the alert.

[0118] Referring to FIG. 4 (c), the sensor system 400 includes a plurality of devices 402 (or a single device), for instance device 200 with reference to FIG. 2. Each device 402 is connected directly to the cloud 406.

[0119] In this embodiment, the activity analysis may be performed within the devices 402. Activity data is then transmitted from the outlets directly to the cloud 406, where further analysis may be performed with additional software in the analytics component (as described with reference to FIG. 4 (a)). The cloud 406 is part of an TCP/IP network adapted to issue alerts 408 to a relative or carer.

[0120] Advantageously, this approach eliminates the requirement for a base station in the home - reducing hardware costs and eliminating the single point of failure. The proliferation of low cost, low power mobile data technologies such as NB-loT and 5G NR will likely make option 4(c) more attractive over time.

[0121] FIG. 5 shows a method 500 for implementing a device or a plurality of devices in accordance with an embodiment of the invention, including the embodiments discussed with reference to FIG. 2.

[0122] The method 500 starts at block 502. At step 504 a device or a plurality of devices for connecting a load or a plurality of loads to a primary power source are installed in a house. In one or more embodiments the devices are provided as GPOs. At step 506, energy information from a plurality of energy sensors configured to measure a change in electrical energy use by the plurality of loads is received by a processor. The loads may be appliances, including lighting, a television, a kettle, a refrigerator and the like. At step 508, presence information from a plurality of presence sensors configured to determine when the user is proximate to the plurality of devices is received by the processor. The processor may be in the device itself, in the house as a separate device, or remote from the house as a separate device or virtual device.

[0123] At step 510, features are extracted from the energy information and the presence information. In one or more embodiments the features relate to the activity of a user in the house. For example, the user spent a particular amount of time in front of a GPO, and the current consumed by the load rose above threshold for a particular amount of time.

[0124] At step 512, a pattern of electrical energy use associated with the user and proximity of the user to the device or plurality of devices is created. The pattern may be considered a baseline activity pattern of the user. In a number of embodiments, the baseline activity pattern is built up over several weeks or months. The baseline activity pattern may be representative of activities performed throughout the course of the day. For example, current is drawn from the kettle at approximately 6:00am most mornings or, a user walks past a particular GPO at 6:30am on the way to the bathroom.

[0125] At step 514, interactions considered unusual between the user and the device or plurality of devices are characterised based on the pattern identified at step 512. Interactions considered unusual may in this example be that current was not drawn from the kettle around 6:00am and the user did not walk past the GPO shortly thereafter.

[0126] The method returns over line 518 to step 506 for as long as sensing is necessary. In many embodiments the method occurs indefinitely i.e., for as long as power is supplied to the system. At step 516, the method terminates.

[0127] An example of a sensor system in accordance with an embodiment of the invention is shown in FIG. 6. The sensor system 600 includes a plurality of devices 602 and 608 positioned in a region. Each device includes at least one microphone 604 and at least one communications module 606 coupled to a housing. In one or more embodiments, the housing includes a GPO or gang switch as described above. A gang switch housing is typical of a housing where lighting controls are mounted (e.g., a switch and dimming controller). It will be appreciated that while only two devices 602 and 608 are shown, it is envisaged that more than two devices will be employed by the system when practically applied.

[0128] Each device 608 and 602 is wirelessly connected to the base station 610. In a number of embodiments, the wireless connection is part of a mesh network, such as an IEEE 802.15.4 mesh network. However, it will be appreciated that the communication can be carried out using any suitable communication protocols, including, but not limited to Wi-Fi 802.11 , 6LowPan/ZIGBEE™ 802.15, Ethernet 802.3, 802.11 and 802.15.4. It will also be appreciated that the communication can be carried out using Ethernet-over-Power to and from the devices, e.g., GPO power sockets or light fittings, and through the powerline itself.

[0129] In one or more embodiments, the microphones 604 continuously stream audio data back to a base station 610. The base station 610 is configured to process and analyse the received audio streams. As described with reference to FIG. 3 and FIG. 4, the base station 610 may be a physical device located in the house, or may be remote from the house, or a virtual device hosted in a cloud computing environment.

[0130] The base station 610 continually analyses received audio data to detect voice commands. As a non-limiting example, the received audio data may be divided into a series of short, partially overlapping samples and input to a feature extraction algorithm. This algorithm processes the sample and returns a feature vector characterising the audio sample. For example, it may compute a Fast Fourier Transform of the audio sample, apply a truncated discrete cosine transformation, and return a vector containing the mel-frequency cepstral coefficients of the sample. Each feature vector is then processed by an acoustic classification algorithm which identifies and classifies the base elements of speech (“phones”) in the sample. This may be achieved by computing the similarity of the received feature vector with a list of predefined feature vectors for specific phones, and selecting the phone which is most similar. Detected phones are then sequentially passed to a language model algorithm, which compares the phones to a dictionary of known words (and their component phones) and returns a sequence of identified words. In this way, the sentence spoken by the user is reconstructed. The sentence is then parsed by a command recognition algorithm, which computes the similarity of the sentence to a predefined list of sentences associated with known commands. It will be appreciated that alternative methods may be used for processing the received audio and identifying commands; for example, feature extraction, phone detection, language parsing, command recognition, or some combination of these may be performed by an artificial neural network trained for the purpose based on example data. The data from multiple microphones 604 may be combined to improve the range and accuracy of the system in the presence of other noise sources. For example, a beamforming algorithm may be used to increase sensitivity to sound from a particular direction.

[0131] In one or more embodiments, the devices 602 and 608 may include physical switches or buttons enabling the user 622 to temporarily or permanently disable monitoring in specific rooms if required.

[0132] An example of a sensor system in accordance with an embodiment of the invention is shown in FIG. 7. The sensor system 700 includes a four devices 702a, 702b, 702c and 702d distributed out a space 708.

[0133] In the embodiment shown, sound 704 is also used to track the position of users within the home. When the user makes a sound 704 (either intentionally, such as a voice command, or unintentionally, such a cough or footstep), it will be detected by several microphones e.g., within devices 702a, 702b, 702c and 702d. The base station (e.g., the base station 610 described above) identifies peaks in the detected sound pressure level (SPL) 706 at the microphones to identify when a “sound event” has occurred. Sound events may be defined as a local temporal maximum in received SPL 706, where the SPL is above a predefined threshold. When a sound is detected by any microphone, the detected SPL 706 at each microphone is compared. Using the relative sound levels and prior information or estimates about locations of each microphone, the system can generate relative positions of each microphone within the space and the source of the sound event via triangulation (see FIG. 7f).

[0134] At FIG. 7a, the system generally has limited information about the positional geometry of the microphones within the space 708. This represents the state of the system after it has been installed but before it has begun operation. A default layout may be initially assumed, such as a grid (as shown), line, or square. It is expected that the accuracy of the map generated by the system in these circumstances will be poor. [0135] Each time a map is generated (FIG. 7a to FIG. 7e), it is used as an input to an estimator algorithm, such as a Kalman filter. This estimator takes previous estimates in addition to the present measurements and provides an updated map of the system 700 compensating for measurement noise. In this way, the accuracy of the positional map will generally improve each time a new sound event 704 is detected by the system.

[0136] Since the accuracy of the system 700 can be expected to increase with each sound event 704, the system may initially be calibrated by deliberately generating a series of sound events 704 from different positions within the space 708. This can be achieved, for example, by walking around the space clapping loudly. Each clap will be identified as a sound event, triggering the generation of a new map and another iteration of the estimator algorithm.

[0137] In addition to or instead of the above process, the user may manually input a map of the location of each sensor module into the system. If this map is accurate, the time taken for the system to calibrate itself will be significantly reduced.

[0138] In addition to detecting commands from the occupant and passively identifying their location, the array of devices 702a, 702b, 702c and 702d can be used to identify activities around the home, including interactions with home appliances or fixtures. Most activities will generate a unique sound, which can be detected by the system. The sound will depend on the type of appliance or activity, as well as the geometry and acoustics of the house. The received audio at one or more devices is compared with stored data and classified based on its similarity or dissimilarity to known patterns. Filters or mathematical transforms may be applied to the audio data to improve the system’s 700 ability to identify and compare specific characteristics of the sound. The location of the source of the sound (measured using the approach described above) may also assist in classification. For example:

[0139] The user turns on the kitchen tap.

[0140] The sound of the water hitting the kitchen sink generates a unique sound. The sound is detected by devices 702a, 702b, 702c and 702d embedded in a light switch and two GPOs located in the kitchen. The devices 702a, 702b, 702c and 702d transmit the audio data back to a base station located elsewhere in the house over a wireless network. [0141] The audio data received from the three units is analysed by the base station.

[0142] The amplitude of the received signal at each unit is compared with previously detected signals. The system identifies that the amplitude (volume) is similar to previous usages of the kitchen tap.

[0143] A Fourier Transform is used to analyse the sound data in the frequency domain. The system 700 identifies that the frequency spectrum matches a pre-defined spectrum associated with running water, and detects subtle peaks in the spectrum previously identified as the resonant frequency of the kitchen water pipes and metal sink, for example.

[0144] The location of the audio source is estimated. The system 700 identifies that a kettle and microwave have previously been used in close proximity to the origin of this sound.

[0145] Based on this information, the system reports that the kitchen tap is being used.

[0146] A similar method of analysis may be used to identify if the user is distressed or calling for help. By analysing the acoustic features and content of the user’s speech or vocalisations, the system 700 can detect if the user is yelling, screaming, or otherwise indicating distress. This would enable, for example, an older adult with dementia to call for help without needing to remember the exact syntax to initiate a voice command.

[0147] For example, the received audio signal may be input to an algorithm which divides the signal into short, overlapping samples and identifies key acoustic features in each sample. These features may include the pitch, amplitude, and zero-crossing rate of the signal, as well as frequency information in the form of mel-frequency cepstral coefficients. The detected features are then analysed by a classifier algorithm which outputs the detected level of distress in the audio signal as a scalar value. In some embodiments, this classifier algorithm may use predefined rules and thresholds for each feature to identify the presence of distress in the vocal sample. In other embodiments, the classifier may use a machine learning model such as a Support Vector Machine (SVM) trained to identify distress using a set of labelled data. [0148] The detected level of distress may be combined with the voice command detection techniques described in [0130] to further improve detection accuracy. For example, reported distress level may be increased if the detected command spoken by the user includes words associated with an incident, such as “help”, “fall”, or “ambulance”.

[0149] In some embodiments, the distress value is then compared to a threshold, with an alert being raised as previously described if the distress level exceeds the threshold for more than a certain period of time.

[0150] Where the terms “comprise”, “comprises”, “comprised” or “comprising” are used in this specification (including the claims) they are to be interpreted as specifying the presence of the stated features, integers, steps or components, but not precluding the presence of one or more other features, integers, steps or components, or group thereof.

[0151] While the invention has been described in conjunction with a limited number of embodiments, it will be appreciated by those skilled in the art that many alternative, modifications and variations in light of the foregoing description are possible. Accordingly, the present invention is intended to embrace all such alternative, modifications and variations as may fall within the spirit and scope of the invention as disclosed.

Claims

The claims defining the invention are as follows

1 . A device for tracking at least one user, the device comprising: a housing; at least one socket accessible via the housing, the socket having apertures to receive pins of an electrical plug for connecting at least one load to a primary power source; at least one energy sensor coupled to the housing configured to measure a change in electrical energy use by the load; at least one presence sensor coupled to the housing configured to determine when the user is proximate to the device; and a processor coupled to the housing configured to resolve from at least a portion of signals from the energy sensor and the presence sensor one or more components.

2. The device of claim 1 , wherein the processor is configured to resolve from the signals by way of a Fourier transform one or more frequency components each corresponding to electrical energy use associated with the user and proximity of the user to the device.

3. The device of claim 1 or 2, wherein the processor is configured to extract features from the energy sensor and the presence sensor components corresponding to a pattern of electrical energy use associated with the user and proximity of the user to the device.

4. The device of any one of claims 1 to 3, wherein the housing comprises a general AC power outlet.

5. The device of any one of claims 1 to 4, further comprising at least one switch coupled to the housing to selectively control the primary power source.

6. The device of claim 5, wherein the switch is operated by the processor to selectively control the primary power source.

7. The device of claim 6, wherein the switch comprises a solid state device.

8. The device of claim 7, wherein the solid state device comprises a bidirectional triode thyristor (triac).

9. The device of any one of claims 1 to 8, further comprising a communication module, coupled to the housing.

10. The device of claim 9, wherein the communication module is wireless.

11. The device of claim 9 or 10, wherein the processor is configured to transmit via the communication module to an analytics component, at least one of: at least a portion of signals from the energy sensor and the presence sensor resolved into one or more components; and features extracted from the energy sensor and the presence sensor components corresponding to a pattern of electrical energy use associated with the user and proximity of the user to the device.

12. The device of claim 11 , wherein the analytics component is configured to extract features from the energy sensor and the presence sensor components corresponding to a pattern of electrical energy use associated with the user and proximity of the user to the device.

13. The device of any one of claims 11 or 12, wherein the analytics component comprises a cloud-based host system configured to identify data indicative of interactions considered unusual between the user and the device based on the pattern.

14. The device of any one of claims 11 or 12, wherein the analytics component comprises a locally-based host system configured to identify data indicative of interactions considered unusual between the user and the device based on the pattern.

15. The device of any one of claims 1 to 14, further comprising at least one microphone coupled to the housing.

16. The device of any one of claims 1 to 15, further comprising at least one sensor coupled to the housing selected from a group consisting of an ultrasonic sensor, a radar sensor, a motion sensor, an impact sensor, a camera, an infrared sensor, a capacitance sensor, a temperature sensor, a light sensor, a chemical sensor, a humidity sensor, a vibration sensor, a pressure sensor, an electrical field sensor, a bio sensor, a particulate sensor, a volatile organic compound sensor, a carbon monoxide sensor, a combustible gas sensor, and a carbon dioxide sensor.

17. A system for tracking at least one user, the system comprising: an analytics component configured to receive and store information from a plurality of devices, the plurality of devices comprising: a housing; at least one socket accessible via the housing, the socket having apertures to receive pins of an electrical plug for connecting at least one load to a primary power source; at least one energy sensor coupled to the housing configured to measure a change in electrical energy use by the load; at least one presence sensor coupled to the housing configured to determine when the user is proximate to the device; and a processor coupled to the housing configured to interact with the analytics component to extract features from the energy sensor and the presence sensor corresponding to a pattern of electrical energy use associated with the user and proximity of the user to the device, and identify signals indicative of interactions considered unusual between the user and the plurality of devices.

18. The system of claim 17, wherein the plurality of devices includes a communication module, coupled to the housing.

19. The system of claim 18, wherein the communication module is wireless.

20. The system of claim 19, wherein the processor is configured to transmit via the communication module to an analytics component, at least one of: at least a portion of the signals from the energy sensor and the presence sensor resolved into one or more components; and features extracted from the energy sensor and the presence sensor components corresponding to a pattern of electrical energy use associated with the user and proximity of the user to the device.

21. The system of claim 20, wherein the analytics component is configured to extract features from the energy sensor and the presence sensor corresponding to a pattern of electrical energy use associated with the user and proximity of the user to the device.

22. The system of any one of claims 17 to 21 , wherein the analytics component comprises a cloud-based host system configured to identify data indicative of interactions considered unusual between the user and the device based on the pattern.

23. The system of any one of claims 17 to 21 , wherein the analytics component comprises a locally-based host system configured to identify data indicative of interactions considered unusual between the user and the device based on the pattern.

24. The system of any one of claims 17 to 23, wherein the plurality of devices are homogeneous and cooperate with each other.

25. The system of any one of claims 17 to 24, wherein the presence information includes an indication that the user has entered close proximity to the plurality of devices and an indication that the user has left close proximity to the plurality of devices.

26. The system of any one of claims 17 to 25, wherein the change in energy usage measured by each of the plurality of devices includes a change in electrical power consumption by each of the loads.

27. A method of tracking at least one user, the method comprising: providing a plurality of devices for connecting a plurality of loads to a primary power source; receiving energy information from a plurality of energy sensors configured to measure a change in electrical energy use by the plurality of loads; receiving presence information from a plurality of presence sensors configured to determine when the user is proximate to the plurality of devices; extracting features from the energy information and the presence information; creating a pattern of electrical energy use associated with the user and proximity of the user to the plurality of devices; and characterising interactions considered unusual between the user and the plurality of devices based on the pattern.

28. A system for tracking at least one user, the system comprising: an analytics component configured to receive and store information from a plurality of devices, the plurality of devices comprising: a housing including at least one of a general AC power outlet or a gang switch; at least one microphone coupled to the housing configured to receive a sound; and a communication module, coupled to the housing and configured to communicate the sound with the analytics component.

29. The system of claim 28, wherein the communication module is wireless.

30. The system of claim 28 or 29, wherein the analytics component includes a multilateration component for determining a location of the sound using differences in received sound pressure level (SPL) relative to the plurality of devices.

31. The system of any one of claims 28 to 30, wherein the analytics component incudes a time of arrival component for determining differences in arrival time of the sound at each of the plurality of devices.

32. The system of any one of claims 28 to 31 , wherein the analytics component includes a natural language processing (NLP) component for determining voice commands.

33. The system of any one of claims 30 to 32, wherein the multilateration component is configured to iteratively update the location of the sound via adaptive filtering.

34. The system of any one of claims 30 to 33, wherein the multilateration component includes an adaptive Kalman filter that uses the SPL to update and maintain current and previous locations of the sound and propagate current location to the analytics component.

35. The system of any one of claims 28 to 34, wherein the analytics component is further configured to extract features from the sound corresponding to a pattern of sounds associated with the user, wherein the features characterize interactions considered unusual between the user and the device based on the pattern.

36. The system of claim 35, wherein the features characterise distress.

37. The system of claim 35, wherein the sounds associated with the user includes sounds associated with appliances.

38. The system of claim 35, wherein the sounds associated with the user includes sounds associated with an event, including the acoustic signals associated with household fixtures.

39. The system of any one of claims 35, wherein the analytics component comprises a cloud-based host system configured to identify sounds indicative of interactions considered unusual between the user and the device based on the pattern.

40. The device of any one of claims 35, wherein the analytics component comprises a locally-based host system configured to identify data indicative of interactions considered unusual between the user and the device based on the pattern.

41. A method of tracking at least one user, the method comprising: receiving sound information from a plurality of devices; extracting features from the sound information; creating a pattern of sounds associated with the user; and characterising interactions considered unusual between the user and the plurality of devices based on the pattern.