Water-safety system with biometric monitoring and automatic buoyancy activation

Abstract

A water-safety system is disclosed comprising a wearable device including a housing with a controller, at least one biometric sensor, and a buoyancy actuator. The controller operates in a detection mode to monitor biometric signals, enters an alarm mode when distress is indicated, and actuates the buoyancy actuator in an activation mode if no cancellation signal is received. Baselines for biometric signals are established using rolling averages to distinguish between normal activity and abnormal conditions. Optional features include adaptive response periods, multi-sensor confirmation, pattern detection of physiological events, reciprocating motion detection, manual activation interface, and wireless communication for remote cancellation, activation, and event notification.

Description

Water-Safety System with Biometric Monitoring and Automatic Buoyancy ActivationField of the Invention

[0001] The invention relates to personal safety equipment. More particularly, it concerns wearable water-safety systems that monitor biometric and motion signals of a user and selectively actuate a buoyancy device in response to detected distress conditions.Background of the Invention

[0002] Water safety systems have been developed to address the risk of drowning, particularly in situations where an individual becomes incapacitated and is unable to maintain buoyancy unaided. A range of prior art arrangements has sought to provide automatic or semi-automatic flotation assistance in such circumstances. These arrangements often combine physiological or environmental monitoring with mechanisms for inflating buoyancy devices when a potential emergency is identified.

[0003] For example, US 2021 / 0403132 A1 (Pourmasiha, 30 December 2021) discloses a personal aquatic safety system incorporating a device that monitors vital signs of the wearer. When one or more monitored parameters falls outside predetermined limits, the device emits an alarm and transmits a wireless distress signal. The system may be integrated into a garment such as an inflatable vest and configured to automatically inflate bladder pockets upon detecting distress from the vital-sign monitors. The device may also deploy a tracker element that ascends to the water surface to emit visual and audible alarms while transmitting a wireless distress signal.

[0004] WO 2015 / 087330 A1 (Schechter, 18 June 2015) discloses a water-flotation garment incorporating a bladder and associated electronics configured to identify emergency conditions such as drowning from sensor data. Upon detection of such conditions, the garment triggers inflation of the bladder to provide buoyancy and assist the wearer to the surface.

[0005] WO 2016 / 191821 A1 (Ecocraft Systems, 8 December 2016) discloses a selfinflating personal safety device having a wearable component and a buoyancy device arranged to adopt a substantially non-buoyant rest state and a buoyant active state. An actuator selectively converts the buoyancy device to the active state in response to a controller determining that a condition detected by one or more sensors has been satisfied.

[0006] These prior arrangements demonstrate recognised advantages, including automated assistance in emergencies, improved detection of physiological or environmental changes, and integration of flotation mechanisms with wearable garments. However, such systems often rely on relatively rigid threshold settings or single-sensor triggers, which can lead to false alarms under ordinary variations in activity or physiology. In other cases, delays in activation may occur if the system does not adequately account for rapid changes in user condition. There may also be limitations in distinguishing between different types of motion or physiological responses, which can affect reliability in real-world aquatic environments.

[0007] The present invention seeks to provide a way to overcome or substantially ameliorate at least some of the deficiencies of the prior art, or to at least provide an alternative.

[0008] It is to be understood that, if any prior art information is referred to herein, such reference does not constitute an admission that the information forms part of the common general knowledge in the art, in Australia or any other country.Summary of the Disclosure

[0009] According to one aspect, there is provided a water-safety system comprising a wearable device having a water-resistant housing containing a controller operatively coupled to a buoyancy actuator and at least one biometric sensor, the controller being configured to operate in a detection mode in which it monitors a biometric signal from the at least one biometric sensor, to enter an alarm mode when distress is detected, and to enter an activation mode in which the buoyancy actuator is actuated if a cancellation signal is not received within a predetermined period, wherein the controller determines a baseline for the biometric signal as a rolling average anddetects distress by evaluating deviation of the biometric signal with reference to the rolling-average baseline.

[0010] In certain embodiments, the controller may apply both a short-term rolling average and a longer-term rolling average, allowing abrupt departures from the baseline to be identified with greater sensitivity while still compensating for natural variability. This arrangement enables more reliable distinction between ordinary exertion and abnormal events.

[0011] In some embodiments, the controller may adjust the response period before activation according to the magnitude of deviation detected. Larger deviations or more rapid changes can result in shortened activation delays, while smaller or gradual deviations can allow longer delays. This proportional response ensures that urgent conditions receive earlier intervention.

[0012] In further embodiments, the controller may require concurrent confirmation from more than one biometric sensor before transitioning to alarm or activation modes. For example, a fall in heart rate may be corroborated by a simultaneous loss of motion. This reduces the likelihood of unwanted deployment caused by artefacts from a single sensor.

[0013] In some embodiments, the controller may additionally maintain a cumulative history of biometric and motion data for the wearer. This history may be used to establish a profile of normal activity and physiological parameters, which can then be referenced in conjunction with the rolling-average baselines. Preferably, the use of such a history allows detection thresholds to adapt to individual differences and longterm trends, thereby improving discrimination between normal exertion and genuine distress.

[0014] In some embodiments, the biometric sensor may comprise a heart-rate sensor, with the rolling average baseline representing the user’s normal heart rate. The controller may be configured to detect a sequence in which the heart rate first rises significantly above the baseline and then falls below it within a short interval, which can indicate a critical physiological event. In such cases, the controller may bypass the alarm mode and proceed directly to activation.

[0015] In other embodiments, the biometric sensor may comprise an accelerometer. The controller may detect reciprocating motion of a limb at or above a set frequency and enter the alarm mode on that basis. Such motion may be indicative of repeated attempts to gain attention or maintain buoyancy.

[0016] In further embodiments, the controller may bypass the alarm mode and proceed directly to activation when a reduction in motion is observed relative to an activity baseline, particularly following a period of elevated activity. Preferably, the activity baseline is computed as a rolling average of accelerometer data, allowing sudden inactivity to be distinguished from a gradual rest condition.

[0017] In some embodiments, the wearable device may include a haptic actuator and a sound emitter configured to deliver countdown alerts during the alarm mode. Audible and tactile warnings provide redundancy in noisy or low-visibility environments, increasing the probability that the user can respond to a pending activation.

[0018] In further embodiments, a manual activation interface may be provided to bypass the controller and mechanically trigger the buoyancy actuator. Preferably, this interface takes the form of a pull-cord that directly actuates the puncture valve of the gas canister. Such a feature ensures reliable operation even if electronic components fail or if battery charge is depleted.

[0019] In some embodiments, the system may include a water sensor with exposed electrodes to detect immersion and to initiate or reinforce entry into detection mode. This ensures that activation functions are appropriately limited to aquatic conditions.

[0020] In other embodiments, the system may comprise an audio transducer configured to identify acoustic characteristics of water entry, providing an additional confirmation of immersion.

[0021] In certain embodiments, the wearable device may further comprise a wireless transceiver operable to communicate with an electronic device, such as a mobile phone. The controller may receive a remote cancellation signal during the alarm mode, or a remote activation command, and may also transmit notifications includingdevice status and location data. This arrangement permits external supervision and intervention where the wearer is unable to act.

[0022] In further embodiments, the wearable device may be configured with a power source scaled according to anticipated use. For example, in configurations intended for recreational swimming, the battery capacity may be reduced to support only a typical session duration, such as two hours, thereby allowing a lighter and less obtrusive housing while still providing adequate operational time.

[0023] According to another aspect, there is provided a method for automated flotation assistance comprising, in a wearable device, establishing a rolling-average baseline for at least one biometric signal, monitoring the biometric signal in a detection mode, entering an alarm mode when a deviation from the baseline indicates distress, entering an activation mode if no cancellation signal is received within a predetermined period, and actuating a buoyancy actuator in the activation mode.

[0024] In certain embodiments of the method, activation may require confirmation from multiple different sensors, or may occur directly when a defined pattern of elevated and reduced heart rate is detected. In further embodiments, activation may be based on reciprocating motion detection, transitions from high to low activity, or remote intervention via wireless communication.

[0025] Other aspects of the invention are also disclosed.Brief Description of the Drawings

[0026] Notwithstanding any other forms which may fall within the scope of the present invention, preferred embodiments of the disclosure will now be described, by way of example only, with reference to the accompanying drawings in which:

[0027] Figure 1 illustrates an embodiment of a wearable water-safety system 100 including housing 102, controller 110, buoyancy actuator 120, flotation bladder 122, biometric sensors 130, wireless transceiver 140, power source 150, haptic actuator 160, sound emitter 162, manual activation interface 170, water sensor 180, audio transducer 182, CO2canister 190, puncture valve 192, and pressure transducer 194.

[0028] Figure 2 illustrates an operational flow diagram of the controller 110 comprising steps 201-213 for detection, alarm and activation of the buoyancy actuator 120.Description of Embodiments

[0029] Figures 1 and 2 illustrate embodiments of a water-safety system 100 configured as a wearable device for detecting signs of distress in a user and for automatically actuating a buoyancy actuator 120 in the event of non-cancellation. Figure 1 shows the principal hardware architecture of the system, while Figure 2 provides a flow diagram of the operational sequence of steps 201 -213 as performed by the controller 110. The system 100 is intended for continuous wear during aquatic activity and is dimensioned and styled to resemble a rash vest garment, being UV- resistant, chlorine-resistant and saltwater-resistant to permit use in pool and open water environments. The garment fit may be maintained by an adjustable chest strap to ensure the biometric sensors 130 remain in intimate contact with the user’s body without hindering swimming movement.

[0030] The system 100 comprises a water-resistant housing 102 secured to the garment 101. The housing may define a detachable console so that the garment 101 may be laundered separately from the electronic components. Within the housing 102 is located the controller 110, one or more biometric sensors 130, a wireless transceiver 140, a rechargeable power source 150, and switching electronics operatively coupled to the buoyancy actuator 120. The buoyancy actuator 120 may comprise a canister of compressed gas arranged to inflate a deployable flotation bladder 122 such as through a specially designed release zipper that allows controlled deployment from the garment structure. The flotation bladder 122 may be dimensioned to provide approximately 100 Newtons of buoyancy, sufficient to support an unconscious wearer in water, and is configured to turn the wearer into a face-up orientation with head and neck support.

[0031] The biometric sensors 130 may include an electrical heart rate monitor configured as an ECG / EKG electrode arrangement disposed within the chest region of the garment. The preferable snug fit of the garment and chest strap ensures reliableelectrical contact even during vigorous swimming activity. The controller 110 preferably monitors the heart rate signal and establishes a baseline value using a rolling-average computation. Preferably, each new sample of the heart rate signal is incorporated into the baseline while the oldest sample is discarded, ensuring that the baseline always reflects the user’s most recent physiological condition. By way of example, the baseline may be calculated over a window of 60 to 120 seconds, although longer or shorter windows may be employed depending on the activity level of the user. This dynamic updating enables the system 100 to adapt to natural changes in heart rate associated with exertion, while still being sensitive to deviations indicative of distress.

[0032] The rolling-average baseline technique is preferably applied not only to heart rate but also to motion signals acquired from accelerometers, gyroscopes, or other movement sensors. For instance, an activity baseline may be computed by averaging accelerometer magnitude over a two-minute interval. If the wearer is swimming vigorously, the baseline will rise accordingly, meaning that elevated activity is not misinterpreted as distress. Conversely, if activity suddenly falls away compared to the established baseline, this may be interpreted as a potential drowning event. In another example, a short-term rolling average of heart rate may be compared against a longer-term baseline; a sharp elevation followed by a rapid fall may be treated as an abnormal pattern consistent with cardiovascular collapse or loss of consciousness.

[0033] In operation, the controller 110 runs a detection mode in which biometric signals are continuously compared to their respective rolling-average baselines. When deviations exceed threshold limits, or when abnormal patterns are identified relative to the baseline, the controller enters an alarm mode and initiates alerting sequences. The alerts may include audible tones and haptic vibration signals, prompting the wearer to consciously cancel the alarm if no distress is present. If the user does not cancel the alarm within a predetermined period, the controller 110 transitions into an activation mode, at which point the buoyancy actuator 120 is triggered to inflate the flotation bladder 122.

[0034] The predetermined response period may be variable and may also be adjusted in proportion to the severity of the deviation. For example, a moderate elevation of heart rate above baseline might permit a 20-second countdown before activation, giving the user ample opportunity to cancel. By contrast, a sudden collapse of heart rate below baseline, combined with loss of motion, may shorten the response period to only a few seconds. This adaptive timing ensures that serious events are addressed with greater urgency while still allowing for recovery or cancellation in less severe cases. The entire process, from monitoring through to actuation, is summarised in the sequence of steps shown in the flow diagram of Figure 2.

[0035] In some preferred but non-limiting embodiments, the controller 110 may compute more than one rolling-average baseline in order to increase sensitivity to certain conditions while maintaining robustness against false triggers. For instance, the controller 110 may calculate a short-term rolling average over a window of 10 to 20 seconds to represent the most recent state of the biometric signal, and compare this to a longer-term rolling average computed over a window of 60 to 120 seconds. By comparing the short-term value against the longer-term baseline, the system can detect sudden departures from a user’s established condition, such as an abrupt spike or fall in heart rate, even where the longer-term baseline still reflects a steady state. This dual-average approach provides a useful balance between responsiveness and stability.

[0036] Optionally, the short-term and long-term baselines may be weighted differently according to activity type or user profile settings. For example, a younger or highly active swimmer may have the system 100 configured to respond to more rapid changes, whereas an older or less active user may rely on longer-term averages to avoid unnecessary alarms. The ability to tailor these averaging intervals enhances the adaptability of the system to different users and aquatic environments.

[0037] In some embodiments, the controller 110 may further store an all-time history of the wearer’s biometric and activity data. This stored history may be used to establish a personalised profile reflecting long-term activity patterns, physiological responses, and resting parameters. The controller 110 may then utilise this profile inconjunction with the rolling-average baselines to refine detection sensitivity and improve discrimination between normal and abnormal conditions. For example, a user whose historical profile indicates consistently elevated heart rates during swimming may have a higher nominal baseline applied, thereby reducing the probability of unnecessary alarms. Conversely, where the history indicates lower exertional capacity, thresholds may be tightened to ensure earlier intervention. The use of historical data therefore allows the system to adapt to individual differences beyond the immediate rolling-average baselines.

[0038] In further embodiments, the predetermined period allowed for alarm cancellation before buoyancy activation may be automatically adjusted in proportion to the severity of the detected deviation. The controller 110 may calculate a severity metric as the difference between the instantaneous biometric signal and the rollingaverage baseline, or as a rate of change of the signal relative to baseline. A larger deviation or more rapid change would correspond to a higher severity metric. The controller 110 may then reduce the countdown time accordingly. For example, if a user’s heart rate falls 30% below baseline within a few seconds, the response period might shorten to 5 seconds or less. If, by contrast, the deviation is modest, such as a gradual 10% increase above baseline during exertion, the response period might extend to 20-30 seconds, giving the user time to recover or cancel the alarm.

[0039] Such adaptive timing improves the practical reliability of the device. It ensures that life-threatening conditions are met with urgent intervention, while reducing the likelihood of false deployments caused by transient exertion or harmless variations. This adaptability is particularly useful in aquatic sports or recreational swimming, where rapid heart rate changes are common but not necessarily dangerous.

[0040] In further optional embodiments, the controller 110 may require corroboration from more than one biometric signal before advancing to the alarm or activation modes. In this arrangement, the system 100 evaluates concurrent indications of distress from two or more sensors 130 within a defined evaluation window. For example, the controller may confirm that a drop in heart rate relative to the rollingaverage baseline is accompanied by a simultaneous reduction in limb motion detectedby the accelerometer. Only when both conditions are satisfied does the controller escalate to the alarm mode. This approach reduces the incidence of false positives arising from isolated sensor artefacts, such as temporary loss of heart-rate signal due to electrode displacement or spurious accelerometer noise caused by water turbulence.

[0041] Optionally, the evaluation window during which multiple signals are compared may be adjustable by user profile or activity type. A shorter evaluation window, such as 5 seconds, may be appropriate in high-risk conditions such as open water swimming, while a longer evaluation window, such as 20 seconds, may be suitable for recreational pool use.

[0042] In yet other embodiments, the system 100 may be configured to recognise specific abnormal sequences of heart-rate behaviour that are strongly indicative of distress. For instance, the controller 110 may detect a pattern in which the heart rate first elevates significantly above the established baseline, consistent with sudden exertion or panic, and is then followed by a rapid fall below the baseline, consistent with cardiovascular collapse, loss of consciousness, or the onset of drowning. In such cases, the controller may be programmed to bypass the ordinary alarm mode entirely and proceed directly to activation of the buoyancy actuator 120. This ensures that no time is lost in situations where delay could be critical.

[0043] Preferably, the thresholds for defining “elevated” and “reduced” relative to baseline are configurable. For example, an increase of more than 25% above baseline within 10 seconds, followed by a fall of more than 30% below baseline within the subsequent 20 seconds, may be treated as a high-priority distress sequence. The precise values may be adapted according to the wearer’s age, medical history, or user-defined safety profile.

[0044] The preferable use of multi-sensor verification and pattern-recognition logic enables the controller 110 to reduce false activations arising from isolated sensor artefacts and to initiate buoyancy deployment promptly when defined combinations or sequences of physiological signals indicative of distress are detected.

[0045] In some embodiments, the biometric sensors 130 include an accelerometer configured to detect reciprocating motion of a user limb. The controller 110 may analyse accelerometer outputs to identify periodic motion above a defined frequency threshold. Detection of such reciprocating motion can be used as an indicator of distress, for example where the frequency corresponds to involuntary thrashing or repeated attempts to maintain buoyancy. Upon detection of reciprocating motion exceeding the threshold, the controller 110 may enter the alarm mode.

[0046] In other embodiments, the controller 110 may bypass the alarm mode and proceed directly to activation when the motion signal indicates a transition from elevated activity to low activity relative to an established activity baseline. This condition may be treated as indicative of loss of consciousness or fatigue following vigorous activity.

[0047] Preferably, the activity baseline is determined as a rolling average of accelerometer magnitude values calculated over a defined time interval, such as 30 to 120 seconds. The use of a rolling average compensates for variations in swimming style or tempo and allows detection of abrupt departures from the prevailing level of activity.

[0048] In some embodiments, the system 100 includes a haptic actuator 160 and a sound emitter 162 operatively connected to the controller 110. When the controller 110 enters the alarm mode, countdown alerts may be generated by the haptic actuator 160 and the sound emitter 162. The haptic actuator 160 may be a vibration motor arranged to deliver tactile pulses through the garment structure, while the sound emitter 162 may be positioned proximate the neck region to provide audible warnings that remain perceptible in a water environment.

[0049] In further embodiments, the system 100 includes a manual activation interface 170 operatively coupled to the buoyancy actuator 120. The manual activation interface 170 provides a direct actuation path that bypasses the controller 110. Preferably, the manual activation interface 170 is in the form of a pull-cord mechanically linked to the gas release mechanism of the buoyancy actuator 120. Inthe event of controller failure or other malfunction, actuation of the pull-cord results in immediate inflation of the flotation bladder 122.

[0050] In some embodiments, the system 100 further comprises a water sensor 180 positioned on the housing 102 or garment surface. The water sensor 180 may include a pair of exposed electrodes arranged to detect conductivity when immersed. Detection of water contact may be used to initiate the detection mode of the controller 110 or to confirm immersion before entering the alarm or activation modes.

[0051] In other embodiments, the system 100 may include an audio transducer 182 configured to detect acoustic signatures associated with submersion events. The audio transducer 182 may be arranged to identify characteristic frequencies or amplitude patterns produced when the device is immersed. Output from the audio transducer 182 may be provided as an additional input to the controller 110 to confirm a water entry condition.

[0052] The system 100 may also include a wireless transceiver 140 arranged for communication with an external electronic device such as a mobile phone running a paired application. In such embodiments, the controller 110 may receive a remote cancellation signal during the alarm mode or a remote activation command for the buoyancy actuator 120. The wireless transceiver 140 may also be configured to transmit event notifications, including device status and geolocation data obtained from an integrated positioning module 142, to the paired device or to predesignated contacts.Exemplary system architecture

[0053] The controller 110 may be realised by a low-power microcontroller with integrated wireless capability to reduce part count. Suitable examples include: (a) a Nordic Semiconductor nRF52840 / nRF5340 SoC providing a Cortex-M4F / M33 core with BLE 5.x, sufficient RAM for rolling-average buffers, hardware cryptography (AES- CCM, ARM TrustZone on the M33 variant), and multiple SPI / I2C / UART peripherals; (b) a Microchip SAM L21 / SAM E54 paired with a Murata BLE module; or (c) an STMicroelectronics STM32WB55 series with dual-core architecture (application core plus radio core). The controller 110 executes time-deterministic tasks for sensoracquisition, baseline computation, state-machine control (detection, alarm and activation), haptic / audio output and wireless communication.

[0054] The biometric sensors 130 may include an ECG front end coupled to textile or snap-on electrodes integrated into the garment. Example ECG / PPG front ends include the Analog Devices AD8232 or Maxim Integrated MAX86150 (ECG + optical PPG). The ECG leads are routed via flexible, salt- and chlorine-resistant cable to the housing 102, with series resistors and ESD protection diodes to meet IEC / ESD robustness. The motion sensors may include a Bosch Sensortec BMI270 or InvenSense ICM-42688 accelerometer / gyroscope connected over l2C / SPI with an interrupt line to timestamp step / reciprocation events. The water sensor 180 may be a pair of 316 stainless electrodes on the garment exterior using an AC excitation scheme from the controller 110 to avoid plating; a resistive divider and comparator or an MCU ADC channel detects conductivity. The audio transducer 182 may be a waterproof MEMS microphone (e.g., Knowles SPH0645 l2S) mounted within the housing 102 behind a hydrophobic acoustic membrane.

[0055] The buoyancy actuator 120 may comprise a CO2canister 190 (for example 24 g to 33 g for small to larger sizes) retained in a corrosion-resistant carrier with a puncture pin valve 192. The valve 192 may be driven by a latching solenoid or motorised cam under the control of the controller 110 via a protected MOSFET driver with current sensing. A manual activation interface 170 is provided as a mechanical pull-cord that directly drives the puncture pin independently of the controller 110. The flotation bladder 122 is stowed within the garment behind a release zipper and / or frangible seam; the zipper slider and stops are configured to avoid accidental release under swimming loads yet open reliably under inflation pressure. An optional pressure transducer 194 in the inflation pathway may be used to verify successful deployment and to log fill dynamics for diagnostics.

[0056] The wireless transceiver 140 may use BLE 5.2 / 5.3 with LE Secure Connections. A GATT profile may expose services including a control service (arming state, alarm cancellation, activation command), a telemetry service (heart-rate samples, motion metrics, battery and temperature) and a diagnostics service (eventlogs, firmware version, self-test status). Pairing is initiated by a concealed user button on the housing 102 with an LED pattern indicating pairing, connected state and fault states. All over-the-air commands that change safety-critical state (alarm cancellation or activation) are authenticated and require an encrypted link; the controller 110 rejects unauthenticated or replayed packets using nonces and session counters.

[0057] The power source 150 may be a lithium-polymer cell sized to provide at least 12 hours of continuous operation under typical acquisition and advertising intervals. In alternative embodiments, the battery capacity may be scaled to match the expected duration of typical swimming sessions in the target user age group. By way of example, many recreational swimming sessions last approximately two hours, and therefore a reduced-capacity cell may be employed to achieve a smaller, lighter housing 102 while still supporting the intended use. A battery gauge IC (e.g., Tl BQ27441) reports state-of-charge, and a charger IC (e.g., Tl BQ51050B for Qi wireless charging, or BQ24074 for wired LISB-C) manages charging. The housing 102 includes a Qi-compatible charging coil and ferrite shield, with conformal coating (e.g., silicone or parylene) on the controller PCB. A buck / boost regulator with low quiescent current supplies 3.0-3.3 V rails. Battery status LED indicators are presented externally through a lightpipe and are readable in daylight.

[0058] The power source 150 may be a lithium-polymer cell sized to provide at least 12 hours of continuous operation under typical acquisition and advertising intervals. A battery gauge IC (e.g., Tl BQ27441) reports state-of-charge, and a charger IC (e.g., Tl BQ51050B for Qi wireless charging, or BQ24074 for wired USB-C) manages charging. The housing 102 includes a Qi-compatible charging coil and ferrite shield, with conformal coating (e.g., silicone or parylene) on the controller PCB. A buck / boost regulator with low quiescent current supplies 3.0-3.3 V rails. Battery status LED indicators are presented externally through a lightpipe and are readable in daylight.

[0059] The haptic actuator 160 may be an eccentric rotating mass or linear resonant actuator driven by a closed-loop haptic driver (e.g., Tl DRV2605) for consistent pulse amplitudes through the garment. The sound emitter 162 may be a waterproof piezo sounder positioned near the neck seam and driven by an H-bridge to achieve highSPL for countdown alerts while submerged. Alert patterns are generated from timers on the controller 110 to provide deterministic cadence.

[0060] The positioning module 142 may be an ultra-low-power GNSS receiver (e.g., u-blox M10 family) integrated on the PCB or provided via the paired device. When integrated, acquisition is event-gated: GNSS remains off in normal swimming to conserve power and is enabled when the alarm mode starts or when water immersion is detected, with hot-start aiding via retained ephemeris when available.

[0061] The firmware architecture on the controller 110 follows a cooperative real-time loop or an RTOS (e.g., Zephyr RTOS or FreeRTOS) with tasks for sensor acquisition, signal processing, state control, I / O and radio. A finite state machine defines at least the following states: idle, detection, alarm, activation and post-activation. Interrupt service routines timestamp sensor edges (accelerometer data-ready, microphone buffer full, ECG sample ready) and push fixed-point samples into ring buffers.

[0062] Signal acquisition on the controller 110 may sample ECG at 100-250 Hz with a digital band-pass filter (e.g., 0.5-40 Hz). R-R intervals are derived using a Pan- Tompkins-type detector or simpler thresholding with refractory period. The rollingaverage baseline for heart rate is calculated over a long-term window (for example 60-120 s) using a fixed-point exponential moving average with coefficient aL chosen so that the effective window corresponds to the desired seconds; a short-term average uses a higher aS to emphasise the last 10-20 s. Motion magnitude is computed from > / (ax2+ay2+az2) with DC bias removal; a long-term activity baseline uses the same exponential averaging scheme and a short-term activity estimate uses a shorter time constant. All averages are maintained per user profile, reset at power- on, and optionally seeded after a brief calibration period when the garment is donned.

[0063] Deviation metrics are derived as normalised differences between instantaneous or short-term values and their respective long-term baselines. A severity metric may be computed as a weighted sum of normalised heart-rate deviation, normalised activity deviation and trend terms (first derivative), with weights configured per user profile. Thresholds for alarm entry and for direct activation arestored in non-volatile memory and may be updated by a paired application, subject to authenticated writes.

[0064] The alarm mode countdown is generated from the severity metric using a monotonic mapping. For example, countdown T may be computed as T = clamp(Tmin, Tmax, Tmax - k S), where S is severity and k is a gain; Tmin and Tmax are profilespecific bounds. The controller 110 produces haptic and audio patterns during the countdown and concurrently listens for local cancellation (button press or specific gesture detected by the accelerometer) and remote cancellation via the wireless transceiver 140. If cancellation is received, the controller 110 reverts to detection and logs the event. If the countdown expires without cancellation, the controller 110 energises the actuator driver to puncture the canister 190 and monitors pressure feedback 194 to verify inflation.

[0065] Data logging is implemented in a circular flash log that records timestamped events: immersion detect, alarm start, severity S, countdown T, cancellation source, activation start, pressure rise and completion status. Logs are downloadable via the diagnostics service for post-event analysis. A bootloader supports authenticated firmware updates over BLE (e.g., MCUBoot with image signing) and a hardware watchdog supervises the application; a brown-out detector protects flash integrity.

[0066] The mobile application communicates with the controller 110 via BLE and provides user profile configuration (age band, activity level, sensitivity), electrode check, sensor self-tests, alarm / activation history and ICE contact management. On alarm start or activation, the application may request foreground execution to display a countdown Ul and to send SMS / notification payloads to pre-configured contacts with location. Where regulations allow, an emergency-services API call may be supported; otherwise, the application prepares a call or message template for the user. The application stores keys for secure pairing and rotates session keys on reconnect. For privacy, raw ECG is not transmitted by default; instead, derived metrics and event flags are sent unless the user opts-in to share raw data.

[0067] Mechanical design of the housing 102 targets at least IP68. The PCB is secured on elastomeric standoffs to mitigate shock, and high-corrosion-riskcomponents are isolated from saltwater ingress paths. Connectors between the detachable console and the garment use sealed pogo pins. The garment integrates an adjustable chest strap routed to maintain electrode contact; all seams in the bladder compartment are radio-frequency welded. The release zipper tape and sliders are selected for chemical resistance and cycle life; a colour tag on the zipper pull differentiates the manual activation interface 170 from ordinary garment features.

[0068] Manufacturing and test provisions include a SWD / JTAG header concealed inside the housing 102 for programming the controller 110, a magnetic reed or Hall sensor to place the device in factory test mode, and a bed-of-nails fixture that exercises all outputs (haptic, audio, actuator driver) and verifies current draw and radio performance. Calibration routines guide electrode impedance checks and motion sensor bias calibration on first use.

[0069] Power-management firmware places the controller 110 in deep sleep between sensor reads. ECG and accelerometer sample rates and BLE advertising intervals are increased only when the water sensor 180 indicates immersion or when motion exceeds an activity threshold. Qi charging is detected via the charger IC status pins; while charging, high-draw tests (haptic / audio) may be run to verify transducers without materially draining the cell.

[0070] Environmental and garment requirements include UV-stable fabrics, chlorine / salt resistance and thermal operation across a defined range suitable for swimming environments. The detachable console allows laundering of the garment while protecting the electronics. The user interface includes an on / off button placed to avoid accidental operation but accessible by the wearer; LED indications provide battery level, pairing status and fault codes via distinct temporal patterns.Exemplary use case

[0071] An exemplary use case scenario is now described with reference to the system 100 and the operational sequence of Figure 2. In step 201 , the system is initialised and powered on while the user wears the garment. In step 202, the controller 110 establishes a rolling-average baseline for heart rate signals obtained from the ECG electrodes 131. For example, during jogging prior to swimming, the user’s heart raterises from 85 beats per minute to 140 beats per minute. A long-term rolling-average baseline of approximately 110 beats per minute is maintained, while a short-term rolling average over 15 seconds reflects the immediate elevation. Because the increase is gradual and consistent with the long-term baseline, the system does not progress beyond step 204, which involves comparing current values against the baseline. This illustrates how the adaptive baseline approach accommodates natural increases in heart rate due to exercise, avoiding a false alarm that might occur with static thresholding.

[0072] When the user later enters the water, step 203 continues monitoring with confirmation of immersion by the water sensor 180. The accelerometer 132 detects reciprocating arm motion at a frequency of approximately 1 Hz. This is incorporated into an activity baseline. Under step 205, the controller distinguishes normal swimming strokes from abnormal motion such as thrashing. If the user intentionally waves an arm rapidly back and forth, this pattern is detected as a distress gesture, prompting entry into the alarm mode at step 206.

[0073] In another sequence, the user’s heart rate rises abruptly from a baseline of 130 beats per minute to 165 beats per minute within 10 seconds, then falls rapidly to 70 beats per minute, while accelerometer activity drops below the activity baseline. At step 205, this is recognised as a critical distress pattern. The severity metric calculated exceeds a threshold, and the controller bypasses a prolonged alarm countdown. At step 206, the alarm mode is entered but, due to the high severity, the response period defined in step 209 is shortened to three seconds.

[0074] During alarm mode, countdown alerts are issued via the haptic actuator 160 and sound emitter 162 in step 206. At step 207, the controller checks for a cancellation signal from the wearer or a paired mobile device. If no cancellation is received, the controller proceeds at step 209 to activation. In step 210, the buoyancy actuator 120 punctures the CO2canister 190 to inflate the flotation bladder 122, which rotates the user into a face-up orientation with head and neck supported.

[0075] In the event of controller failure or battery depletion, step 211 allows the wearer to trigger the manual activation interface 170, directly actuating the puncture valve 192 and inflating the flotation bladder 122 without controller intervention.

[0076] At step 212, the wireless transceiver 140 transmits event notifications including device status and geolocation data to the paired electronic device, which may forward these to emergency contacts. Remote activation or cancellation commands received during step 207 are verified before execution.

[0077] Finally, in step 213, once the event cycle is complete, the system either resumes monitoring biometric signals for subsequent events or powers down as required.

[0078] The foregoing description, for purposes of explanation, used specific nomenclature to provide a thorough understanding of the invention. However, it will be apparent to one skilled in the art that specific details are not required in order to practise the invention. Thus, the foregoing descriptions of specific embodiments of the invention are presented for purposes of illustration and description. They are not intended to be exhaustive or to limit the invention to the precise forms disclosed as obviously many modifications and variations are possible in view of the above teachings. The embodiments were chosen and described in order to best explain the principles of the invention and its practical applications, thereby enabling others skilled in the art to best utilize the invention and various embodiments with various modifications as are suited to the particular use contemplated. It is intended that the following claims and their equivalents define the scope of the invention.

Claims

Claims1. A water-safety system comprising a wearable device having a water-resistant housing containing a controller operatively coupled to a buoyancy actuator and at least one biometric sensor, the controller being configured to operate in a detection mode in which it monitors a biometric signal from the at least one biometric sensor, to enter an alarm mode when distress is detected, and to enter an activation mode in which the buoyancy actuator is actuated if a cancellation signal is not received within a predetermined period, wherein the controller determines a baseline for the biometric signal as a rolling average and detects the distress by evaluating deviation of the biometric signal with reference to the rolling-average baseline.

2. The system of claim 1 , wherein the controller applies a first rolling average to obtain a short-term value of the biometric signal and compares the short-term value to the rolling-average baseline computed over a longer time window.

3. The system of claim 1 , wherein the controller adjusts the predetermined period inversely with a severity metric that depends on the magnitude of deviation between the biometric signal and the rolling-average baseline.

4. The system of claim 1 , wherein the controller enters the activation mode only when concurrent indications of distress are detected from at least two different biometric sensors within an evaluation window.

5. The system of claim 1 , wherein the biometric sensor comprises a heart-rate sensor and the baseline heart rate is calculated as a rolling average for the user.

6. The system of claim 5, wherein the controller bypasses the alarm mode and enters the activation mode upon detecting a sequence comprising an elevated heart rate relative to the baseline followed by a lowered heart rate relative to the baseline within a predetermined interval.

7. The system of claim 1 , wherein the biometric sensor comprises an accelerometer and the controller is configured to enter the alarm mode when detecting reciprocating motion of a limb above a threshold frequency.

8. The system of claim 7, wherein the controller bypasses the alarm mode and enters the activation mode when detecting low activity relative to an activity baseline following a period of elevated activity.

9. The system of claim 8, wherein the activity baseline is calculated as a rolling average of accelerometer activity.

10. The system of claim 1 , wherein the wearable device comprises a haptic actuator and a sound emitter arranged to deliver countdown alerts in the alarm mode, the sound emitter being positioned proximate a neck region in a vest configuration.

11. The system of claim 1 , further comprising a manual activation interface mechanically coupled to the buoyancy actuator so as to bypass the controller.

12. The system of claim 11 , wherein the manual activation interface comprises a pullcord.

13. The system of claim 1 , further comprising a water sensor with exposed electrodes configured to detect water conductivity and to cause entry into the detection mode.

14. The system of claim 1 , further comprising an audio transducer configured to detect audio signatures indicative of submersion and to cause entry into the detection mode.

15. The system of claim 1 , wherein the wearable device comprises a wireless transceiver paired with an electronic device running an application, the controller being configured to receive, via the transceiver, a remote cancellation signal during the alarm mode and / or a remote activation command for the buoyancy actuator, and to transmit event notifications including device status and location to the electronic device.

16. The system of claim 1 , wherein the controller is configured to store a history of biometric and motion data for the wearer and to establish a profile of normal parameters based on the stored history, the profile being referenced in conjunction with the rolling-average baseline.

17. A method for automated flotation assistance comprising, in a wearable device, establishing a rolling-average baseline for at least one biometric signal, monitoring the biometric signal in a detection mode, entering an alarm mode when a deviation from the baseline indicates distress, entering an activation mode if no cancellation signal is received within a predetermined period, and actuating a buoyancy actuator in the activation mode.

18. The method of claim 17, comprising requiring concurrent indications of distress from at least two different sensors before actuating the buoyancy actuator.

19. The method of claim 17, comprising detecting a heart-rate pattern of elevation above the baseline followed by reduction below the baseline within a window and bypassing the alarm mode to actuate the buoyancy actuator.

20. The method of claim 17, comprising detecting reciprocating motion using an accelerometer to trigger the alarm mode and detecting a transition from elevated activity to low activity relative to an activity baseline to bypass the alarm mode and actuate the buoyancy actuator.

21. The method of claim 17, comprising communicating with a paired electronic device to accept a remote cancellation signal during the alarm mode and / or a remote activation command, and sending an event notification including device status and location data to the paired device.

22. The method of claim 17, comprising storing a history of biometric and motion data for the wearer, establishing a profile of normal parameters based on the stored history, and referencing the profile in conjunction with the rolling-average baseline when monitoring for distress.

Images