Intelligent assistant and emergency situation recognition and determination system using a real-time temporal artery pulse-based multimodal sensor on-device ai ambient-noise-supprssion mono-headset-type smart wearable device, and control method thereof
The mono-headset wearable device integrates biosensors and environmental detectors with on-device AI for real-time emergency judgment and secure internal network processing, addressing integration and response delays in existing technologies.
Patent Information
- Application Number
- KR1020260086441
- Authority / Receiving Office
- KR · KR
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2026-05-13
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2046-05-13
AI Technical Summary
Existing wearable devices in industrial settings lack integration of voice communication headsets and biosensors into a single mono-headset type smart wearable device, struggle to classify emergency causes using synchronized biological and environmental data, and rely on external networks that can compromise security and delay emergency responses.
A mono-headset type smart wearable device with integrated biosensors, environmental sensors, and motion detectors, utilizing an on-device AI station for real-time data analysis and emergency judgment, prioritizing voice communication during emergencies, and processing data on an internal network.
Enables real-time classification of emergency causes and prioritized voice communication, enhancing data security and reducing reliance on external networks for immediate emergency responses.
Smart Images

Figure 112026058005404-PAT00001_ABST
Abstract
Description
Technology Field
[0001] The present invention relates to a multimodal sensor-based on-device AI smart wearable device system, and more specifically, to a system that provides work safety and work convenience to workers engaged in various industrial sites such as manufacturing, construction, logistics, and chemical industries by integrally collecting and analyzing bio-data including pulse signals in the temporal artery region, environmental data, and motion data. More specifically, the invention relates to a system and a control method thereof that includes a mono-headset type smart wearable device that acquires multimodal sensor data including temporal artery pulse signals in real time, and an on-device AI station that is connected to the device via an internal network to enable communication, thereby providing real-time voice communication between workers, an AI-based intelligent assistant, and a service for recognizing and judging multiple emergency situations in a high-noise industrial environment. Background Technology
[0002] In industrial sites, walkie-talkies, headsets, mobile devices, camera equipment, and various sensor devices are used to respond to emergencies and guide workers on on-site work procedures. Voice communication devices are essential, particularly in noisy environments or sites where workers must use both hands. Workers perform tasks while communicating in real-time with managers or other workers. Furthermore, since the bringing of personal devices into security-sensitive industrial sites is prohibited, it is common practice to verify work procedures or communicate with other workers using authorized dedicated equipment.
[0003] Meanwhile, in work environments involving high temperatures, confined spaces, heights, chemical handling, or equipment inspection, it is necessary to monitor both the worker's vital signs and the surrounding environment. Biometric data such as heart rate, body temperature, and oxygen saturation can be used to identify individual abnormalities, while environmental data such as temperature, humidity, atmospheric pressure, VOCs, gases, and CO2 can be used to assess the risk level at the work site.
[0004] In addition, by checking changes in the worker's posture or impact conditions together, it is possible to determine whether an accident has occurred due to a fall or sudden movement.
[0005] In this regard, Korean Registered Patent No. 10-1866677 discloses a safety management system and method at a construction site based on a wearable device.
[0006] The above registered patent includes a configuration in which a wearable device mounted on a worker's body measures health data including body temperature, heart rate, blood pressure, oxygen saturation, and pressure of both feet, collects location information and altitude information, and a server calculates a risk score and a total risk score using the health data, location information, and altitude information, and transmits a notification signal according to the risk level.
[0007] However, the above-mentioned conventional technology relates to a structure that calculates risk using a worker's health data and location or altitude information, and has limitations in integrating a voice communication headset structure and a biosensor structure actually worn by a worker at an industrial site into a single mono-headset type smart wearable device.
[0008] In other words, a wearable structure equipped with a unilateral speaker that provides sound to only one ear of the worker, while attaching the biosensor part to the temporal artery region by means of a support member on the other side of the main body, has not been sufficiently considered.
[0009] Furthermore, since conventional technology focuses on calculating risk levels using health data, location information, or altitude information, there are limitations in classifying emergency causes into individual physiological abnormalities, environmentally induced abnormalities, and complex abnormalities by synchronizing time with biological data and environmental data such as temperature, humidity, atmospheric pressure, VOCs, gases, and CO2 around the worker, as well as motion data from an inertial measurement unit.
[0010] Consequently, it is difficult to distinguish whether an increase in heart rate and body temperature is due to thermal abnormalities caused by a hot and humid environment, whether a decrease in oxygen saturation is due to environmentally induced hypoxia resulting from elevated CO2 or VOCs, or whether shock signals and changes in posture are due to a fall accompanied by sudden changes in atmospheric pressure.
[0011] In addition, while workers in industrial settings may receive voice-based work instructions corresponding to equipment, processes, or worker identification information, in the event of an emergency, a connection to a manager must be automatically performed based on the emergency assessment result rather than work instructions.
[0012] Since conventional technology focuses on biometric-based risk calculation and notification signal transmission, it has limitations in controlling QR code-based work guidance, offline emergency keyword recognition, voice communication between workers, and priority connection to manager terminals as a single priority system.
[0013] Furthermore, since safety data from industrial sites may include workers' biometric data, environmental data, motion data, voice signals, work details, and log data, transmitting it to an external cloud can lead to security issues and communication delays. In particular, in sites where external network connectivity is limited, such as underground areas, confined spaces, plants, and factory interiors, relying on an external cloud makes it difficult to respond immediately to emergencies.
[0014] Therefore, there is a demand for on-device AI-based intelligent assistant and emergency situation recognition and judgment technology that can stably acquire a worker's biometric data, environmental data around the worker, and motion data from a single mono-headset type smart wearable device while maintaining voice communication functions, analyze this data by synchronizing it with time on an internal network and an edge computing server, and control emergency response actions to take priority over work guidance in the event of an emergency. Prior art literature
[0015] Korean Patent Publication No. 10-1866677 (Registered June 4, 2018) "Safety management system and method at a construction site based on a wearable device" The problem to be solved
[0016] In accordance with the above needs, the present invention aims to provide a mono-headset type smart wearable device that acquires biometric data, environmental data, and motion data together through a mono-headset type smart wearable device worn by an industrial site worker, analyzes them by time-synchronizing them on an edge computing server of the site's internal network to classify the cause of an emergency as a personal physiological abnormality, an environment-induced abnormality, or a complex abnormality, and controls the system so that emergency voice recognition and connection to a manager take priority over work guidance when an emergency situation is detected, as well as an intelligent assistant and an emergency situation recognition and judgment system using the same and a control method thereof. means of solving the problem
[0017] To achieve the above objectives, the present invention provides a mono headset type wearable device (100) comprising a main body (110) worn on the head of a worker, a single speaker (120) that provides sound to only one ear of the worker, a microphone (130), a support member (140) including a tension providing part (141) inside, a biosensor part (150) that is in close contact with the temporal artery area of the worker by the support member (140), an environment sensor part (160) that measures the environment around the worker, an inertial measurement part (170) that generates motion data of the worker, a voice communication part (180), and a data communication part (190).
[0018] In addition, the present invention provides a monoheadset type smart wearable device (100) that forms a functional layer (155) including an adhesive layer (156), a conductive layer (157), and a waterproof layer (158) on the skin contact surface side of the biosensor part (150) of the monoheadset type wearable device (100), thereby maintaining the contact state of the biosensor part (150) with the temporal artery region even in situations of movement or sweating during work and increasing the measurement stability of biodata.
[0019] In addition, the present invention provides an intelligent assistant and emergency situation recognition and judgment system that includes an on-device AI station (200) that is informationally connected to the mono headset type smart wearable device (100), and the on-device AI station (200) includes an internal network (230), a data collection and transmission unit (210), a QR recognition camera unit (220), and an edge computing server (240) to collect, store, analyze, and control and process biometric data, environmental data, motion data, voice signals, and QR code information within an industrial site.
[0020] In addition, the present invention provides an intelligent assistant and emergency situation recognition and judgment system in which an edge computing server (240) includes an internal storage unit (241), a state determination unit (250), an emergency voice recognition unit (260), a voice communication processing unit (270), a QR interpretation and work guidance unit (280), an event management unit (290), and a reference state learning unit (295), and determines any one of emergency scenarios such as cardiac arrest, stroke, fainting, hypoxia or suffocation, heatstroke or thermal abnormality, fall, arrhythmia, hyperventilation or panic attack, or exposure to harmful gases or chemicals by mapping biological abnormal signs, environmental risk levels, and motion data in time synchronization.
[0021] In addition, the present invention provides a control method for an intelligent assistant and emergency situation recognition and judgment system comprising a data acquisition step (S10), a data transmission step (S20), a reference state learning step (S30), a state determination step (S40), an emergency voice recognition step (S50), a work guidance output step (S60), and an event control step (S70), wherein when an emergency determination or emergency keyword recognition occurs, the work guidance output is stopped and the voice communication processing unit (270) is switched to a priority connection channel for the administrator terminal. Effects of the invention
[0022] According to the present invention, a worker's biometric data, environmental data, and motion data can be acquired together while maintaining voice communication functions through a mono-headset type smart wearable device. In addition, by analyzing the data with time synchronization, the cause of the emergency can be classified into individual physiological abnormalities, environmentally induced abnormalities, or complex abnormalities.
[0023] Furthermore, in the event of an emergency determination or recognition of an emergency keyword, the output of work guidance can be stopped and the connection channel can be switched to the administrator terminal's priority channel, allowing for the rapid execution of intelligent assistant functions and emergency response.
[0024] In addition, since the original data and inference results are stored and processed on the internal network and edge computing servers, dependence on the external network can be reduced and on-site data security can be enhanced. Brief explanation of the drawing
[0025] FIG. 1 is an image illustrating the outwardly visible configuration of the mono-headset type smart wearable device and on-device AI station of the present invention. FIG. 2 is a cross-sectional view illustrating a multilayer structure of a functional layer. Specific details for implementing the invention
[0026] Various embodiments and / or aspects are now disclosed with reference to the drawings. For illustrative purposes, numerous specific details are disclosed in the following description to aid in a general understanding of one or more aspects. However, it will be apparent to those skilled in the art that these aspects may be practiced without such specific details. The following description and the accompanying drawings describe specific exemplary aspects of one or more aspects in detail. However, these aspects are exemplary, and some of the various methods in the principles of the various aspects may be used, and the descriptions are intended to include all such aspects and their equivalents. Specifically, terms such as “exemplary,” “example,” “aspect,” and “example” as used herein may not be interpreted as implying that any described aspect or design is superior or advantageous over other aspects or designs.
[0027] Hereinafter, identical or similar components are assigned the same reference numeral regardless of drawing symbols, and redundant descriptions thereof are omitted. Furthermore, in describing the embodiments disclosed in this specification, detailed descriptions of related prior art are omitted if it is determined that such detailed descriptions may obscure the essence of the embodiments disclosed in this specification. Additionally, the attached drawings are intended only to facilitate understanding of the embodiments disclosed in this specification, and the technical concept disclosed in this specification is not limited by the attached drawings.
[0028] Although terms such as first, second, etc. are used to describe various elements or components, it goes without saying that these elements or components are not limited by these terms. These terms are used merely to distinguish one element or component from another. Therefore, it goes without saying that the first element or component mentioned below may be the second element or component within the technical scope of the present invention.
[0029] Unless otherwise defined, all terms used in this specification (including technical and scientific terms) may be used in a meaning that is commonly understood by those skilled in the art to which the present invention pertains. Additionally, terms defined in commonly used dictionaries are not to be interpreted ideally or excessively unless explicitly and specifically defined otherwise.
[0030] Furthermore, the term "or" is intended to mean an implicit "or" rather than an exclusive "or." That is, unless otherwise specified or evident from the context, "X uses A or B" is intended to mean one of the natural implicit substitutions. In other words, if X uses A; if X uses B; or if X uses both A and B, "X uses A or B" may apply to any of these cases. Additionally, the term "and / or" as used herein should be understood to refer to and include all possible combinations of one or more of the enumerated related items.
[0031] Additionally, the terms “comprising” and / or “comprising” should be understood to mean that such features and / or components are present, but not to exclude the presence or addition of one or more other features, components, and / or groups thereof. Furthermore, unless otherwise specified or clearly evident from the context to indicate a singular form, the singular in this specification and claims should generally be interpreted to mean “one or more.”
[0032] In addition, the terms "information" and "data" as used in this specification may often be used interchangeably.
[0033] When it is stated that a component is "connected" or "connected" to another component, it should be understood that it may be directly connected or connected to that other component, or that there may be other components in between. On the other hand, when it is stated that a component is "directly connected" or "directly connected" to another component, it should be understood that there are no other components in between.
[0034] The suffixes "module" and "part" used for components in the following description are assigned or used interchangeably solely for the ease of drafting the specification and do not inherently possess distinct meanings or roles.
[0035] The purpose and effects of the present disclosure, and the technical configurations for achieving them, will become clear by referring to the embodiments described in detail below in conjunction with the accompanying drawings. In describing the present disclosure, if it is determined that a detailed description of known functions or configurations might unnecessarily obscure the essence of the present disclosure, such detailed description will be omitted. Furthermore, the terms described below are defined considering their functions in the present disclosure, and these may vary depending on the intentions or practices of the user or operator.
[0036] However, the present disclosure is not limited to the embodiments disclosed below but may be implemented in various different forms. These embodiments are provided merely to make the present disclosure complete and to fully inform those skilled in the art of the scope of the disclosure, and the present disclosure is defined only by the scope of the claims. Therefore, such definitions should be based on the content throughout this specification.
[0038] Hereinafter, a monoheadset type smart wearable device according to the present invention will be described.
[0039] A mono headset-type smart wearable device (100) according to one embodiment of the present invention is worn on the head of an industrial site worker and performs voice communication, biometric data measurement, environmental data measurement, and motion data measurement together.
[0040] Unlike a headset that covers both ears of a worker, a mono headset-type smart wearable device (100) has a wearing structure that provides sound to only one ear of a worker, and accordingly, the worker can hear voice guidance or communication voice through a single speaker (120) while also perceiving on-site warning sounds, equipment operation sounds, or the voice of a nearby worker through the other ear.
[0041] This mono headset-type smart wearable device (100) includes a main body (110), a single-sided speaker (120), a microphone (130), a support member (140), a biosensor unit (150), an environment sensor unit (160), an inertial measurement unit (170), a voice communication unit (180), and a data communication unit (190).
[0042] The main body (110) forms a supporting frame that is worn on the worker's head. The main body (110) has a curved frame shape that extends along the upper, side, or back of the worker's head, and can maintain the wearing state even when the worker moves or turns their head at an industrial site.
[0043] The main body (110) provides a basic structure that supports a single-sided speaker (120), a microphone (130), a support member (140), an environment sensor unit (160), an inertial measurement unit (170), a voice communication unit (180), and a data communication unit (190).
[0044] A single-sided speaker (120) is provided on one side of the main body (110) and outputs sound at a position corresponding to one ear of the worker, and is provided in a form that does not completely block the ear, so that the worker can hear the output sound while simultaneously perceiving external sounds from the site, and such a single-sided acoustic structure provides a wearing state that maintains voice communication functions in an industrial site without blocking surrounding dangerous sounds.
[0045] The single-sided speaker (120) is coupled to one side of the main body (110) and has an acoustic output surface facing one ear of the worker, and since the opposite ear is open, it is desirable to have an ear cup structure that is generally difficult to apply in industrial sites due to its sealing properties.
[0046] Accordingly, the single-sided speaker (120) outputs a manager's voice, work guidance voice, or emergency guidance voice transmitted from the voice communication unit (180) to one ear of the worker, while not blocking the other ear of the worker and the surrounding auditory path, so the worker perceives equipment warning sounds, equipment operation sounds, rear approach sounds, or the voices of surrounding workers together.
[0047] A microphone (130) is provided in the main body (110) and faces the area around the worker's mouth or a direction capable of receiving voice, and receives a voice signal spoken by the worker. The voice signal input to the microphone (130) is used for voice communication with other workers or managers through the voice communication unit (180) and is also provided for emergency keyword recognition by the emergency voice recognition unit (260) to be described later.
[0048] The support member (140) has an arm shape extending from the other side of the main body (110) toward the temporal region of the worker and supports the biosensor part (150) so that the skin contact surface of the biosensor part (150) comes into contact with the temporal artery region of the worker, and a tension providing part (141) is provided inside the support member (140) to provide elastic force or compressive force to the support member (140).
[0049] The tension providing part (141) is provided inside the support member (140) and applies a restoring force or a pressing force to the support member (140).
[0050] In one embodiment, the tension providing member (141) includes an elastic piece, a plate spring, a coil spring, or an elastic wire inserted along the longitudinal direction of the support member (140), and one end of the tension providing member (141) is supported on the side of the main body (110), and the other end transmits force to the sensor support end on the side of the biosensor member (150).
[0051] Therefore, when a worker wears the mono headset-type smart wearable device (100) on their head, the support member (140) spreads outward and then attempts to return inward due to the restoring force of the tension providing part (141), so the skin contact surface of the biosensor part (150) is continuously pressed toward the temporal artery area.
[0052] However, there is a possibility that the worker may feel discomfort when the pressure is strong, and since the present invention is designed to be fixed to the temporal artery region by including an adhesive layer (156), the main purpose of the tension providing part (141) may be to ensure comfort of wearing rather than providing strong pressure, while allowing the biosensor part (150) to be positioned in the temporal artery region.
[0053] The biosensor unit (150) includes a PPG sensor (151) and a body temperature sensor (152).
[0054] The PPG sensor (151) includes a light-emitting part that emits light toward the skin contact surface of the biosensor part (150) and a light-receiving part that detects reflected light that changes according to changes in blood flow in the temporal artery area, and the light-emitting part and the light-receiving part are formed at positions adjacent to each other with respect to the skin contact surface.
[0055] Since the PPG sensor (151) irradiates light through the skin contact surface in close contact with the temporal artery area and detects reflected light, the PPG sensor (151) generates an optical biosignal to calculate the worker's heart rate and oxygen saturation (SpO2).
[0056] A body temperature sensor (152) is provided on the skin contact surface of the biosensor part (150) or around it to measure the temperature of the temporal artery area or the surrounding skin.
[0057] The body temperature sensor (152) is formed together with the PPG sensor (151) in the same sensor housing or sensor pad, and generates body temperature data based on a temporal artery region substantially identical to the optical measurement location of the PPG sensor (151), and obtains heart rate, oxygen saturation (SpO2), and body temperature together based on a single temporal artery contact point without measuring them separately at different wearing sites.
[0058] A functional layer (155) is formed on the skin contact surface side of the biosensor part (150).
[0059] The functional layer (155) includes an adhesive layer (156), a conductive layer (157), and a waterproof layer (158), and is formed between the measuring surface of the biosensor part (150) and the worker's skin to assist in the sensor contact state.
[0060] The adhesive layer (156) provides adhesion to the skin so that the skin contact surface of the biosensor part (150) does not slip in the temporal artery area, the conductive layer (157) stabilizes the biosignal transmission path between the biosensor part (150) and the skin, and the waterproof layer (158) prevents sweat or external moisture from directly entering the measurement surface of the PPG sensor (151) and the body temperature sensor (152).
[0061] The specific stacking structure of this functional layer (155) will be described later.
[0062] The environmental sensor unit (160) is integrally provided in the main body (110) to measure the surrounding air conditions where the worker is located. Unlike the bio-sensor unit (150) which is in close contact with the worker's skin, the environmental sensor unit (160) must come into contact with the outside air, so it is coupled to the outer surface, front side, side side, or exposed surface where an air inlet is formed of the main body (110).
[0063] In one embodiment, the environmental sensor unit (160) is inserted into a sensor receiving groove formed in the outer housing of the main body (110), and an opening or porous protective cover through which air passes is formed on the outer side of the sensor receiving groove, thereby protecting the sensor from dust or external impact while allowing air around the worker to reach the temperature sensor (161), humidity sensor (162), atmospheric pressure sensor (163), VOC sensor (164), gas sensor (165), and CO2 sensor (166).
[0064] In other words, the environmental sensor unit (160) includes a temperature sensor (161), a humidity sensor (162), an atmospheric pressure sensor (163), a VOC sensor (164), a gas sensor (165), and a CO2 sensor (166), and each sensor generates a measurement value based on the ambient air entering through the external air contact area of the main body (110).
[0065] The temperature sensor (161) and humidity sensor (162) measure the temperature and humidity around the worker, the pressure sensor (163) measures changes in atmospheric pressure around the worker, and the VOC sensor (164), gas sensor (165), and CO2 sensor (166) measure volatile organic compounds, harmful gases or specific gases, and CO2 concentrations, respectively. Accordingly, the environmental sensor unit (160) generates field environmental data based on the worker's movement location.
[0066] The inertial measurement unit (170) is coupled to the main body (110) or the support member (140) to measure 3-axis acceleration and 3-axis gyroscope. In one embodiment, the inertial measurement unit (170) is fixed to the internal substrate of the main body (110) or the base of the support member (140), and detects the movement of the operator's head, changes in posture, impact, and changes in rotation based on the movement of the main body (110), and the inertial measurement unit (170) includes a 3-axis acceleration sensor (171) and a 3-axis gyroscope sensor (172).
[0067] Since the inertial measuring unit (170) is coupled to the main body (110) or support member (140) worn on the worker's head rather than the wrist or tool, it may be possible to generate motion data closer to changes in the worker's upper body posture and falls than to local movements caused by the operation of work tools or wrist movements.
[0068] The voice communication unit (180) is provided within the internal circuit board of the main body (110) or within a communication housing coupled to the main body (110), and is electrically connected to a single-sided speaker (120) and a microphone (130). The voice communication unit (180) converts a voice signal input from the microphone (130) into a communication signal and transmits it to another worker or manager, and transmits a voice signal received from the outside to the single-sided speaker (120).
[0069] Separately, the data communication unit (190) is electrically connected to the biosensor unit (150), the environment sensor unit (160), and the inertial measurement unit (170) to transmit the biodata, environment data, and motion data generated from each sensor to the data collection and transmission unit (210) or internal network (230) to be described later.
[0070] This mono headset-type smart wearable device (100) combines an acoustic output structure having a single speaker (120), a temporal artery region contact structure having a support member (140) and a tension providing part (141), and a composite sensing structure having a biosensor part (150), an environment sensor part (160), and an inertial measurement part (170) into a single main body (110), so that a worker can obtain voice signals, biometric data, environmental data, and motion data required during industrial site work through the mono headset-type smart wearable device (100) at a single wearing point without wearing a voice communication device, a biometric measurement device, an environment measurement device, and a motion measurement device separately.
[0071] In other words, the real-time temporal artery pulse application in the present invention means acquiring a PPG raw signal while the biosensor unit (150) is in close contact with the temporal artery area of the worker, and generating biodata corresponding to heart rate, oxygen saturation (SpO2), heart rate variability (HRV), or respiratory rate (RR) using the blood flow change or pulse component included in the PPG raw signal, and the multimodal sensor base means acquiring biodata, environmental data, and motion data together through the biosensor unit (150), environmental sensor unit (160), and inertial measurement unit (170), and processing different types of sensor data together based on the same worker and the same time interval.
[0073] The functional layer (155) is formed on the skin contact surface side of the biosensor part (150) and is a layer that stabilizes the contact state between the measurement surface of the biosensor part (150) and the temporal artery area of the worker, and has a structure in which an adhesive layer (156) and a conductive layer (157) are laminated, and a waterproof layer (158) surrounds the outer periphery of the adhesive layer (156) and the conductive layer (157).
[0074] At this time, the adhesive layer (156) maintains skin contact between the measurement surface of the biosensor part (150) and the worker's skin, the conductive layer (157) assists in increasing the efficiency of biosignal transmission, and the waterproof layer (158) prevents sweat, moisture, or external moisture from penetrating to the edges of the adhesive layer (156) and the conductive layer (157).
[0075] The adhesive layer (156) may be formed into a skin-contacting hydrogel layer comprising catechol group-introduced polyvinyl alcohol, polyvinylpyrrolidone, glycerin, and betaine.
[0076] The catechol group-introduced polyvinyl alcohol provides skin contact force so that the adhesive layer (156) does not easily slip even on a moist skin surface, polyvinylpyrrolidone forms a polymer matrix that retains moisture to reduce drying shrinkage of the adhesive layer (156), glycerin provides flexibility and moisturization to the adhesive layer (156) to prevent the skin contact surface from hardening when worn for a long time, and betaine stabilizes the skin contact condition by relieving the moisture balance of the adhesive layer (156) even in the presence of sweat or moisture.
[0077] At this time, the adhesive layer (156) may be formed from an adhesive layer forming composition comprising 0.1 to 10 weight% of catechol group-introduced polyvinyl alcohol, 1 to 20 weight% of polyvinylpyrrolidone, 1 to 30 weight% of glycerin, 0.1 to 10 weight% of betaine, and the remainder being purified water.
[0078] If the content of the catechol group-introduced polyvinyl alcohol in the above range is excessively low, the contact retention force on the temporal artery area is weakened, and if it is excessively high, the adhesive layer (156) may become excessively sticky or the residue may increase when removed from the skin.
[0079] In addition, polyvinylpyrrolidone and glycerin control the hydrogel framework and flexibility of the adhesive layer (156), and betaine reduces the rapid change in the moisture state of the adhesive layer (156) due to sweat.
[0080] The conductive layer (157) may be formed as an ion-conducting hydrogel layer comprising polyvinyl alcohol, polyvinylpyrrolidone, sodium chloride, choline chloride, and glycerin.
[0081] Polyvinyl alcohol provides a transparent film-forming framework for the conductive layer (157), polyvinylpyrrolidone increases the moisture retention of the conductive layer (157) so that the ion transport pathway is not easily broken, sodium chloride and choline chloride impart ion conductivity to the conductive layer (157) to assist in signal transmission between the biosensor part (150) and the skin, and glycerin maintains the flexibility of the conductive layer (157) to reduce cracking even with repeated deformation during wear.
[0082] At this time, the conductive layer (157) may be formed from a conductive layer forming composition comprising 2 to 20 weight% polyvinyl alcohol, 1 to 15 weight% polyvinylpyrrolidone, 0.1 to 3 weight% sodium chloride, 0.1 to 10 weight% choline chloride, 1 to 30 weight% glycerin, and the remainder being purified water.
[0083] The conductive layer (157) can be formed with a thickness of 30 to 300 μm so as not to affect the optical measurement of the PPG sensor (151), and the light transmittance of the area corresponding to the light-emitting part and the light-receiving part can be adjusted to be 70% or more, and accordingly, the conductive layer (157) simultaneously assists in thermal contact of the body temperature sensor (152) and the acquisition of an optical signal of the PPG sensor (151).
[0084] The waterproof layer (158) can be formed into an elastic waterproof layer comprising polydimethylsiloxane, silsesquioxane-based silicone resin and polyurethane-based elastomer.
[0085] Polydimethylsiloxane provides flexibility and moisture barrier properties to the waterproof layer (158), silsesquioxane-based silicone resin increases the surface water resistance and film strength of the waterproof layer (158), and polyurethane-based elastomer ensures that the waterproof layer (158) maintains an outer periphery seal even if it is repeatedly deformed by the pressure of the support member (140) and the movement of the worker.
[0086] The waterproof layer (158) is formed in a ring shape or border shape that surrounds the outer periphery of the adhesive layer (156) and the conductive layer (157), rather than covering the entire surface of the adhesive layer (156) and the conductive layer (157), thereby suppressing moisture penetration at the edges while maintaining the skin contact area and sensor measurement area in the central part.
[0087] At this time, the waterproof layer (158) may be formed from a waterproof layer forming composition comprising 40 to 95 weight% polydimethylsiloxane, 1 to 30 weight% silsesquioxane-based silicone resin, 1 to 30 weight% polyurethane-based elastomer, and 0.1 to 10 weight% silica.
[0088] Silica controls the flowability of the waterproof layer (158) to reduce the spreading of the waterproof layer forming composition into the central measurement area when applied to the outer periphery, and maintains the edge shape of the waterproof layer (158), so that the waterproof layer (158) is formed along the sides of the adhesive layer (156) and the conductive layer (157), thereby blocking the path for sweat, washing water, or external moisture to penetrate between the layers.
[0089] The method for manufacturing the functional layer (155) can be performed as follows.
[0090] First, the skin contact surface of the biosensor part (150) is cleaned with isopropyl alcohol or purified water, and after removing surface moisture using dry air, plasma treatment or corona treatment is performed on the skin contact surface of the biosensor part (150) to form a surface condition in which the conductive layer (157) can be stably attached, and at this time, a transparent protective film is left on the light-emitting part and light-receiving part of the PPG sensor (151), or a masking member is temporarily attached so that the light path is not blocked.
[0091] Next, a conductive layer forming composition is prepared.
[0092] Specifically, polyvinyl alcohol is added to purified water and stirred at 80 to 95°C for 1 to 3 hours to dissolve the polyvinyl alcohol, and then polyvinylpyrrolidone, sodium chloride, choline chloride, and glycerin are sequentially added while the temperature is lowered to 40 to 60°C to prepare a uniform conductive layer forming composition, and then the conductive layer forming composition is coated on the skin contact surface of the biosensor part (150) with a wet thickness of 50 to 200 μm, and dried at 30 to 60°C for 10 to 60 minutes to form a conductive layer (157).
[0093] Next, an adhesive layer forming composition is prepared.
[0094] Specifically, polyvinylpyrrolidone is added to purified water and dissolved, then catechol group-introduced polyvinyl alcohol, glycerin, and betaine are added and stirred at 25 to 50°C for 30 minutes to 2 hours to prepare an adhesive layer forming composition, and the prepared adhesive layer forming composition is coated on the skin-side surface of the conductive layer (157) with a wet thickness of 50 to 500 μm, and dried at 25 to 50°C for 30 minutes to 4 hours to form an adhesive layer (156).
[0095] Next, a waterproof layer (158) is formed along the outer periphery of the adhesive layer (156) and the conductive layer (157).
[0096] Specifically, a waterproof layer forming composition is prepared by mixing polydimethylsiloxane, silsesquioxane-based silicone resin, polyurethane-based elastomer, and silica, and after applying it in a ring shape or border shape to the outer periphery of the adhesive layer (156) and the conductive layer (157) using a dispenser or screen mask, curing and drying are performed at 40 to 80°C for 10 minutes to 2 hours to form a waterproof layer (158).
[0097] Finally, the masking member is removed, and the light transmission state of the area corresponding to the light-emitting part and light-receiving part of the PPG sensor (151), the skin contactability of the adhesive layer (156), the ion conductivity of the conductive layer (157), and the outer circumference sealing state of the waterproof layer (158) are checked.
[0098] In this process, the total thickness of the functional layer (155) can be adjusted to a range of 100 to 800 μm so as not to create an excessive gap when the biosensor part (150) is in close contact with the temporal artery area, and the waterproof layer (158) is formed limited to the outer periphery so as not to cover the central measurement area.
[0099] The functional layer (155) manufactured in this manner maintains the skin contact state of the biosensor part (150) even when a worker moves or sweats in an industrial setting, and increases the measurement stability and durability of heart rate, body temperature, and oxygen saturation (SpO2).
[0101] Meanwhile, the voice communication unit (180) and the data communication unit (190) process voice signals and sensor data through separate communication paths.
[0102] The voice signal corresponds to a worker's speech sound, a manager's call sound, a communication sound between workers, or an emergency situation guidance sound input through the microphone (130), and the sensor data corresponds to biometric data, environmental data, and motion data generated by the biometric sensor unit (150), environmental sensor unit (160), and inertial measurement unit (170).
[0103] As such, voice signals and sensor data differ in the nature of the data, transmission cycle, and delay tolerance, so the mono headset type smart wearable device (100) does not mix the two types of data in one path but processes them through separate communication paths.
[0104] Specifically, the voice communication unit (180) is connected to a single-sided speaker (120) and a microphone (130) to transmit and receive voice signals of a worker, and the data communication unit (190) is connected to a bio-sensor unit (150), an environment sensor unit (160), and an inertial measurement unit (170) to transmit data generated from each sensor.
[0105] At this time, the voice communication unit (180) and the data communication unit (190) use different wireless communication protocols or use different channels, sessions, or identification information even within the same communication method to distinguish between voice signals and sensor data.
[0106] For example, the voice communication unit (180) uses a Bluetooth audio or voice-only wireless channel suitable for real-time voice communication, and the data communication unit (190) uses a BLE or sensor data-only channel suitable for periodic transmission of biometric data, environmental data, and motion data.
[0107] According to this communication path separation structure, even if the amount of data transmitted by the biosensor unit (150), the environment sensor unit (160), and the inertial measurement unit (170) increases, the phenomenon of the voice signal of the voice communication unit (180) being delayed or interrupted is reduced.
[0108] In addition, even when a worker makes a voice call through the microphone (130) in an emergency situation or an administrator delivers an emergency notice through the single-sided speaker (120), the sensor data transmission and voice communication do not conflict with each other, thereby maintaining a stable real-time communication state.
[0109] Therefore, the mono headset type smart wearable device (100) performs voice communication and sensor data transmission functions together within a single wearable device, while simultaneously ensuring the immediacy of voice signals and the continuity of sensor data, thereby reducing communication delays or data loss even in situations where communication between workers, calling of managers, measurement of biological conditions, measurement of environmental conditions, and measurement of operating conditions are performed in parallel at an industrial site.
[0111] Meanwhile, the biosensor unit (150) generates biodata corresponding to heart rate, heart rate variability (HRV), and respiratory rate (RR) based on the PPG raw signal output from the PPG sensor (151).
[0112] The PPG raw signal includes an optical signal that changes according to changes in blood flow in the temporal artery region, and the biosensor unit (150) calculates the heart rate using the peak interval of the PPG raw signal and calculates the heart rate variability (HRV) using the variability of the peak interval. In addition, the biosensor unit (150) can calculate the respiratory rate (RR) using changes in the amplitude or baseline of the PPG raw signal.
[0113] At this time, the output data of the environment sensor unit (160) and the output data of the inertial measurement unit (170) are time-synchronized with the output data of the biosensor unit (150) and transmitted to the data communication unit (190). To this end, the mono headset type smart wearable device (100) assigns a measurement time, data sequence number, or common timestamp to the output data of the biosensor unit (150), the environment sensor unit (160), and the inertial measurement unit (170), and ensures that the biodata, environment data, and motion data generated in the same time interval correspond to each other.
[0114] In this way, the mono headset type smart wearable device (100) calculates biometric data at one wearing position and transmits biometric data, environmental data, and motion data sorted by time, so that the intelligent assistant and emergency situation recognition and judgment system described later can analyze the worker's state and the surrounding environment state at the same time.
[0116] Below, prior to the specific description of the intelligent assistant and emergency situation recognition and judgment system according to the present invention, an overview of the system will be described.
[0117] The intelligent assistant and emergency situation recognition and judgment system of the present invention is implemented as a computing device-based system for collecting, storing, analyzing, and controlling a worker's biometric data, environmental data, motion data, voice signals, and QR code information within the site.
[0118] The above computing device-based system is implemented using known hardware devices including a known processor, memory, storage device, communication interface, sensor interface, voice input / output interface, video input interface, and power supply circuit, and said hardware devices are electrically and informationally connected through a wired or wireless communication path, which will be obvious to those skilled in the art.
[0119] The above processor executes program instructions stored in memory to perform sensor data collection, data preprocessing, time information assignment, data storage, status determination, voice signal processing, QR code interpretation, work guidance generation, and event priority control.
[0120] The above memory stores program instructions executed by the processor and temporary data generated during operations, and the above storage device stores biometric data, environmental data, motion data, voice signals, QR code information, inference results, and log data, etc.
[0121] The above communication interface transmits and receives information that may include biometric data, environmental data, motion data, voice signals, and QR code information between devices within the site.
[0122] The sensor interface receives electrical signals or digital data generated from biosensors, environmental sensors, and inertial sensors, the voice input / output interface processes voice signals input from a microphone and voice signals output from a speaker, and the image input interface receives image data including QR codes or site identification information.
[0123] The above computing device-based system may be implemented as a single physical device, or it may be implemented in a form in which a device worn by a worker and a field processing device carried by the worker or mounted on work clothes or protective gear are connected to each other.
[0124] In addition, each functional unit and module is implemented using separate dedicated circuits, firmware, software, or a combination thereof, and a single processor executes program instructions corresponding to multiple functional units and modules, or multiple processors perform functional operations in a distributed manner.
[0125] In the present invention, on-device processing refers to a method of performing data collection, storage, analysis, inference, and control processing on a computing device within an industrial site, rather than a method of transmitting original data to an external cloud and receiving a judgment result back from an external server.
[0126] Accordingly, the intelligent assistant and emergency situation recognition and judgment system according to the present invention performs worker status judgment, emergency situation recognition, work guidance generation, and voice communication control within the site without relying on the external network connection status.
[0127] Based on such a computing device-based implementation, a specific embodiment is described below in which a mono headset-type smart wearable device (100), an on-device AI station (200), an internal network (230), an edge computing server (240), and their sub-components are electrically and informationally interconnected to perform intelligent assistant and emergency situation recognition and judgment.
[0129] The intelligent assistant and emergency situation recognition and judgment system of the present invention includes a mono headset-type smart wearable device (100) and an on-device AI station (200).
[0131] A mono headset-type smart wearable device (100) is worn on a worker's head to acquire biometric data, environmental data, motion data, and voice signals, and an on-device AI station (200) is informationally connected to the mono headset-type smart wearable device (100) to collect, store, analyze, and control the data within the site.
[0132] At this time, the on-device AI station (200) is provided in a form that is mounted on the worker's body, work clothes, or protective gear, or carried by the worker, so it can perform on-site data processing along with the worker's movement location, and by executing an artificial intelligence inference model on the edge computing server (240) to determine the cause of an emergency, an emergency scenario, and an event priority from a combination of biometric data, environmental data, motion data, and voice signals, it provides an on-device processing environment that does not rely on an external cloud.
[0133] The on-device AI station (200) includes an internal network (230), a data collection and transmission unit (210), a QR recognition camera unit (220), and an edge computing server (240).
[0134] The internal network (230) forms a data transmission and reception path between the data collection and transmission unit (210), the QR recognition camera unit (220), and the edge computing server (240) within the on-device AI station (200).
[0135] The internal network (230) may include wired communication lines, internal buses, Ethernet, serial communication, short-range wireless communication, or a combination thereof, and provides an information path through which biometric data, environmental data, motion data, voice signals, and QR code information are transmitted to an edge computing server (240).
[0136] The data collection and transmission unit (210) is connected to an internal network (230) and collects sensor data from the data communication unit (190) of a mono headset-type smart wearable device (100) and transmits it to the internal network (230).
[0137] Here, the sensor data includes bio-data generated by the bio-sensor unit (150), environmental data generated by the environmental sensor unit (160), and motion data generated by the inertial measurement unit (170).
[0138] The data collection and transmission unit (210) is provided with an MCU-based circuit to perform the reception, buffering, data format conversion, data sequence maintenance, and transmission to the internal network (230), and the sensor data generated from the mono headset-type smart wearable device (100) is processed by the edge computing server (240) as data for the same worker and the same time interval.
[0139] The QR recognition camera unit (220) is connected to an internal network (230) and acquires QR codes displayed on equipment, process charts, work charts, entrances, danger zones, or worker identification targets within the industrial site.
[0140] The QR recognition camera unit (220) is provided as a camera device that is integrated into the on-device AI station (200) or connected to the on-device AI station (200) via wired or wireless connection, and transmits the acquired QR code image or QR code recognition data to the edge computing server (240) through the internal network (230).
[0141] For example, when a worker photographs a QR code attached to the equipment to be inspected, the QR recognition camera unit (220) obtains identification information of the equipment, and the edge computing server (240) can retrieve an inspection procedure, precautions for handling hazardous materials, or maintenance history corresponding to the identification information from the internal storage unit (241) and provide it to the worker as voice guidance.
[0142] The edge computing server (240) is informationally connected to the data collection and transmission unit (210), the QR recognition camera unit (220), and the voice communication unit (180) of the mono headset type smart wearable device (100) through the internal network (230).
[0143] At this time, the edge computing server (240) is provided as a computing device that performs calculations and executes an artificial intelligence inference model within the on-device AI station (200), and includes an internal storage unit (241), a status determination unit (250), an emergency voice recognition unit (260), a voice communication processing unit (270), a QR interpretation and work guidance unit (280), and an event management unit (290).
[0144] The edge computing server (240) analyzes biometric data, environmental data, motion data, voice signals, and QR code information within the site using program instructions stored in the processor and memory and an artificial intelligence inference model stored in the memory, and performs data storage, status determination, emergency keyword recognition, voice communication control, QR-based work guidance generation, and event priority control.
[0145] The internal storage unit (241) is provided in the edge computing server (240) and stores biometric data, environmental data, motion data, voice signals, inference results, and log data.
[0146] The internal storage unit (241) may include a semiconductor storage device, flash memory, SSD, eMMC, memory card, or an equivalent storage medium, and stores an artificial intelligence model, sensor data received through the data collection and transmission unit (210), QR code information obtained through the QR recognition camera unit (220), voice signals processed by the emergency voice recognition unit (260), and the judgment result of the status determination unit (250) together with time information.
[0147] Accordingly, the on-device AI station (200) manages the worker's biometric data and field work information internally at the site without transmitting them to an external cloud.
[0148] The state determination unit (250) determines the cause of an emergency and the emergency scenario using bio-data from the bio-sensor unit (150), environmental data from the environmental sensor unit (160), and motion data from the inertial measurement unit (170).
[0149] This state determination unit (250) does not determine only whether a single sensor value is exceeded, but also processes together how bio-data, environmental data, and motion data appear in relation to each other in the same time interval.
[0150] To this end, the state determination unit (250) includes a biomarker analysis module (251), an environmental indicator analysis module (252), a cross-fusion judgment module (253), and a scenario classification module (254).
[0151] The biomarker analysis module (251) analyzes bio-data corresponding to heart rate value, body temperature value, oxygen saturation value, PPG raw signal, heart rate variability value, or respiratory rate value transmitted from the biosensor unit (150) to calculate signs of abnormality in the body.
[0152] That is, the biomarker analysis module (251) generates signs of biological abnormalities such as a decrease or increase in heart rate, a decrease in oxygen saturation, irregular changes in heart rate intervals, abnormal heart rate variability, an increase in respiratory rate, or loss of respiratory signals.
[0153] The environmental indicator analysis module (252) calculates the environmental risk level by analyzing environmental data corresponding to the temperature value, humidity value, atmospheric pressure value, VOC sensor output value, gas sensor output value, and CO2 concentration value transmitted from the environmental sensor unit (160).
[0154] That is, the environmental indicator analysis module (252) generates an environmental risk level indicating high temperature and humidity conditions, sudden changes in atmospheric pressure, VOC increase, gas anomaly, CO2 increase, or poor ventilation conditions.
[0155] The cross-fusion judgment module (253) maps the above biological abnormality signs, the above environmental risk level, and the operation data of the inertial measurement unit (170) by time synchronization, and classifies the cause of the emergency into one of personal physiological abnormalities, environmentally induced abnormalities, and complex abnormalities.
[0156] The scenario classification module (254) determines an emergency scenario based on the classification result of the cross-fusion judgment module (253), determines an emergency scenario corresponding to any one of cardiac arrest, stroke, fainting, hypoxia or suffocation, heatstroke or heat abnormality, fall, arrhythmia, hyperventilation or panic attack, or exposure to harmful gas or chemical substances, and transmits the determined emergency scenario to the event management unit (290).
[0157] The emergency voice recognition unit (260) recognizes an emergency keyword offline from a voice signal received through the voice communication unit (180) and internal network (230) from the microphone (130) of the mono headset type smart wearable device (100).
[0158] The emergency voice recognition unit (260) does not transmit voice signals to an external voice recognition server, but performs voice feature extraction, keyword matching, or voice recognition operations within the edge computing server (240).
[0159] Accordingly, when a worker utters a voice requesting rescue or indicating a dangerous situation at an industrial site, an emergency voice event is generated inside the on-device AI station (200).
[0160] The voice communication processing unit (270) controls real-time voice communication between workers by linking with the voice communication unit (180) of the mono headset type smart wearable device (100).
[0161] The voice communication processing unit (270) manages the transmission path of the voice signal input from the microphone (130) and the reception path of the voice signal to be output to the single-sided speaker (120), controls the communication connection status between multiple workers or managers, and also switches between the normal communication status between workers and the manager priority connection status in emergency situations according to the control of the event management unit (290).
[0162] In an additional embodiment of the present invention, the voice communication processing unit (270) may perform worker-adaptive voice preprocessing to mitigate the phenomenon of the first syllable being cut off in the speech start section for the worker's speech content.
[0163] Here, the voice communication processing unit (270) determines the utterance start section by reflecting weights learned based on voice characteristics accumulated from the worker's repeated utterances and applying multiple voice preprocessing modules having different threshold values.
[0164] In this case, a frequency range that is detected only by a sensitive preprocessing module with a relatively low threshold and not by a general preprocessing module with a relatively high threshold can be determined as a candidate for the first syllable cut off at the start of the utterance.
[0165] For example, when a worker pronounces 'toothpaste,' since 'ㅊ' is a voiceless consonant of the dental series, it may be difficult to exceed the threshold for general speech recognition; however, the sensitive preprocessing module can detect the corresponding frequency range.
[0166] Accordingly, the voice communication processing unit (270) can reduce the omission of the first syllable of the speech start section even in a noisy environment by complementarily combining the first syllable candidate section detected by the sensitive preprocessing module with the low-noise voice section detected by the general preprocessing module.
[0167] The QR interpretation and work guidance unit (280) interprets the QR code obtained by the QR recognition camera unit (220) to extract identification information corresponding to a work target, work process, or worker within the industrial site, and generates work guidance (voice guidance) corresponding to the identification information.
[0168] The QR interpretation and work guidance unit (280) corresponds the equipment identification information, process identification information, or worker identification information extracted from the QR code with the work manual, production schedule, or work status stored in the internal storage unit (241), and converts the guidance content required for the corresponding worker or equipment into voice guidance data.
[0169] The generated voice guidance is output to a single speaker (120) through the voice communication unit (180) of the mono headset type wearable device (100), so the worker can listen to the work guidance without operating a separate portable terminal with their hands.
[0170] The event management unit (290) controls the operation priority of the status determination unit (250), emergency voice recognition unit (260), voice communication processing unit (270), and QR interpretation and work guidance unit (280).
[0171] The event management unit (290) monitors the emergency determination result of the status determination unit (250) and the emergency keyword recognition result of the emergency voice recognition unit (260), and when an emergency event occurs, it prioritizes the administrator connection or emergency guidance output of the voice communication processing unit (270) over QR interpretation and the provision of work guidance by the work guidance unit (280).
[0172] This intelligent assistant and emergency situation recognition and judgment system provides QR-based work guidance and voice communication between workers during normal operations, and in emergency situations, can prioritize emergency response actions based on status assessment and emergency keyword recognition results.
[0173] In other words, in the present invention, the intelligent assistant refers to a function that performs voice guidance, work procedure guidance, emergency keyword recognition, and communication connection control based on voice signals, QR code information, and work-related data between multiple workers and an on-device AI station (200); the recognition and judgment of an emergency situation refers to a function that analyzes biometric data, environmental data, and motion data by synchronizing them in time, distinguishes whether the abnormal condition of the worker corresponds to a personal physiological abnormality, an environment-induced abnormality, or a complex abnormality, and determines an emergency scenario based on the result.
[0175] Meanwhile, the cross-fusion judgment module (253) maps the biological abnormality signs calculated by the biomarker analysis module (251), the environmental risk level calculated by the environmental indicator analysis module (252), and the operation data of the inertial measurement unit (170) based on the same time interval.
[0176] Here, the same time interval is determined based on the measurement time, data sequence number, or timestamp assigned to the biometric data, environmental data, and motion data, and the cross-fusion judgment module (253) distinguishes whether the biometric abnormality sign appears alone, together with the environmental risk level, or together with the abnormal pattern of the motion data.
[0177] The scenario classification module (254) determines any one of the emergency scenarios based on the mapping results, such as cardiac arrest, stroke, fainting, hypoxia or suffocation, heatstroke or heat abnormality, fall, arrhythmia, hyperventilation or panic attack, or exposure to hazardous gases or chemicals.
[0178] Cardiac arrest scenarios are determined based on a rapid decrease or loss of heart rate, SpO2, HRV, and RR values.
[0179] The scenario classification module (254) determines cardiac arrest when multiple conditions among the following conditions occur together in the same time interval: the heart rate value decreases to 30 bpm or less, the heart rate signal is lost for 10 seconds or more, the SpO2 value decreases to 80% or less, the HRV value appears close to 0 for 5 minutes, and the respiration signal is lost for 20 seconds or more.
[0180] This is intended to classify a state in which heart rate activity, oxygen supply, autonomic nervous response, and respiratory activity are all reduced as a cardiac arrest scenario.
[0181] Stroke scenarios are identified based on irregular changes in heart rate interval (IBI), sudden changes in HRV values, and abnormal head tilt observed in the raw PPG signal.
[0182] The scenario classification module (254) determines that it is a stroke if irregular changes in IBI, atrial fibrillation (AF) pattern, or an IBI coefficient of variation exceeding 25% appear in the original PPG signal, the RMSSD (root mean square difference of adjacent heart rate intervals) among the HRV values changes by more than ±50% compared to the reference, and the head tilt in the motion data of the inertial measurement unit (170) is maintained at a state of 15 degrees or more for more than 10 seconds.
[0183] This is intended to classify a condition in which cardiac rhythm instability and the abnormal maintenance of head posture occur together as a neurological abnormality scenario.
[0184] The fainting scenario is determined based on the rapid change in heart rate values and the posture change pattern of the inertial measurement unit (170).
[0185] The scenario classification module (254) determines fainting when the heart rate value changes by more than 30 bpm within 1 minute or decreases rapidly by more than 20 bpm, and when a free fall signal, an impact signal, and a horizontal posture maintenance appear sequentially in the motion data of the inertial measurement unit (170), and when the RMSSD among the HRV values decreases rapidly or the SpO2 value temporarily decreases to less than 94%.
[0186] This is intended to classify a state in which changes in heart rate and falling motions occur together, unlike simple changes in posture, as a syncope scenario.
[0187] Hypoxia or suffocation scenarios are determined based on a decrease in SpO2 values and an increase in CO2 concentration values.
[0188] The scenario classification module (254) determines the hypoxia warning state when the SpO2 value drops to 90% or less, determines the hypoxia danger state when the SpO2 value drops to 85% or less, and determines the hypoxia due to poor ventilation when the drop in SpO2 value continues for 30 seconds or more and the CO2 concentration value rises to 1,000 ppm or more.
[0189] In addition, if the CO2 concentration rises above 5,000 ppm and the RR value increases beyond 25 beats / min and then decreases, or if the heart rate value changes from a state exceeding 110 bpm to a state below 50 bpm, it is determined to be asphyxiation.
[0190] This is intended to classify a condition in which reduced oxygen supply and deterioration of the respiratory environment occur together as an environment-induced hypoxia or suffocation scenario.
[0191] Heatstroke or heat abnormality scenarios are determined based on conditions of increased body temperature and heart rate, as well as increased temperature and humidity.
[0192] The scenario classification module (254) determines that if the body temperature rises above 38.5℃, it is in a thermal abnormality warning state, if the body temperature rises above 40℃, it is in a thermal abnormality danger state, and if the heart rate rises above 100 bpm for more than 10 minutes, it is in a continuous tachycardia state.
[0193] In addition, if the WBGT (Wet Bulb Globe Temperature Index) calculated using temperature and humidity values is 28℃ or higher, it is determined to be a high temperature and high humidity warning environment, and if the WBGT is 31℃ or higher, it is determined to be a high temperature and high humidity danger environment, and if the SDNN (Standard Deviation of Normal Heart Rate Interval) among HRV values drops to less than 20ms, it is determined to be a thermal burden accompanied by a decrease in autonomic nervous system function.
[0194] This is intended to classify conditions in which an increase in body temperature and heart rate occurs along with a hot and humid environment as heatstroke or heat-related abnormality scenarios.
[0195] The fall scenario is determined based on the sequential occurrence of the free fall signal, impact signal, and attitude change of the inertial measurement unit (170).
[0196] The scenario classification module (254) determines that it is a fall if, in the motion data of the inertial measurement unit (170), a free fall signal of 0.3g or less occurs, followed by an impact signal of 4g or more, and then a change in posture or maintenance of a horizontal posture occurs.
[0197] In addition, if the state of inactivity is maintained for more than 10 seconds after impact, it is determined to be a fall accompanied by a decrease in consciousness; if the heart rate value rapidly increases to over 120 bpm after impact, it is determined to be a fall accompanied by a vital response after the accident; and if a sudden change in atmospheric pressure occurs, it is determined to be a fall accompanied by a change in height or a fall while working on the upper part of the equipment.
[0198] Arrhythmia scenarios are determined based on irregular changes in IBI and abnormalities in HRV values identified in the raw PPG signal.
[0199] The scenario classification module (254) determines that it is an arrhythmia when the IBI coefficient of variation exceeds 25% in the original PPG signal, a tachycardia pattern or a bradycardia pattern appears, and an RMSSD surge or SDNN ultra-low pattern appears together in the HRV value.
[0200] At this time, if the environmental data is within the normal range and no free fall signal, impact signal, or posture change appears in the motion data, the scenario classification module (254) classifies the abnormality as an individual physiological abnormality rather than an environmentally induced abnormality or a change caused by a fall.
[0201] Hyperventilation or panic attack scenarios are determined based on whether the RR value increases, the heart rate increases, and the SpO2 value remains within the normal range.
[0202] The scenario classification module (254) determines that it is hyperventilation or a panic attack when the RR value increases to 25 beats / min or more, the heart rate value rises to 110 bpm or more, and the SpO2 value is in the normal range or appears to be 98% or more.
[0203] At this time, if the CO2 concentration value, VOC sensor output value, gas sensor output value, temperature value, and humidity value are within the normal range, the scenario classification module (254) classifies the condition as an individual physiological abnormality rather than environmentally induced hypoxia or chemical exposure.
[0204] Exposure scenarios to hazardous gases or chemicals are determined based on simultaneous changes in VOC sensor output values, gas sensor output values, CO2 concentration values, and biological responses.
[0205] The scenario classification module (254) determines that there is an abnormal air quality condition when the VOC sensor output value changes by more than 200% compared to the baseline, the gas sensor output value changes rapidly, or the CO2 concentration value rises above 1,000 ppm, and also determines that there is exposure to harmful gases or chemicals when the RR value increases by more than 22 beats / minute or the heart rate value exceeds 100 bpm.
[0206] This is intended to classify conditions in which rapid changes in environmental data occur together with respiratory or heart rate responses as chemical exposure scenarios.
[0207] In this way, the scenario classification module (254) does not determine the biological data, environmental data, and motion data individually, but determines the emergency scenario based on the combination of sensors and the order of change that appear in the same time interval.
[0208] Therefore, even if the same increase in heart rate occurs, if an increase in body temperature and WBGT occurs together, it is determined to be heatstroke or thermal abnormality; if irregular changes in IBI and abnormalities in HRV values occur while environmental data is within the normal range, it is determined to be arrhythmia; and if a decrease in SpO2 value and an increase in CO2 concentration occur together, it can be determined to be hypoxia or suffocation.
[0209] Accordingly, the intelligent assistant and emergency situation recognition and judgment system performs a complex judgment responding to the cause of the emergency and the emergency scenario, rather than simply determining whether the risk level has been exceeded.
[0211] Furthermore, the standard state learning unit (295) forms worker standard state information and field standard state information using biometric data and environmental data newly input into the corresponding wearing section within the on-device AI station (200), and then updates the worker standard state information and field standard state information by reflecting the data of the section determined to be normal or cautionary state as additional learning data.
[0212] The individual standard value learning unit (295) receives biometric data and environmental data collected during the initial period of wear by the worker as standard state learning data, and calculates the average and standard deviation of each data item included in the standard state learning data to form worker standard state information and field standard state information for the corresponding wear period.
[0213] Here, the initial wearing period refers to a time period including a stable state before or immediately after the start of work after the worker wears the mono headset-type smart wearable device (100).
[0214] Specifically, the reference state learning unit (295) uses biometric data such as heart rate, body temperature, SpO2, HRV, and RR values collected during the initial wearing period, and environmental data such as temperature, humidity, atmospheric pressure, VOC sensor output, gas sensor output, and CO2 concentration as reference state learning data.
[0215] In one embodiment, the individual reference value learning unit (295) sets the data for the first 3 to 10 minutes after the worker wears the mono headset type smart wearable device (100) as the data for the initial wearing period, and excludes the period in which the worker's sudden movement, walking, or impact is detected using the motion data of the inertial measurement unit (170) from the reference state learning target.
[0216] Accordingly, the standard state learning unit (295) forms worker standard state information using heart rate value, body temperature value, SpO2 value, HRV value and RR value of the section in which the worker is in a relatively stable state, and forms field standard state information using temperature value, humidity value, atmospheric pressure value, VOC sensor output value, gas sensor output value and CO2 concentration value of the same section.
[0217] Additionally, the standard state learning unit (295) calculates the average value and standard deviation for each data item, sets the average value as the state center value, sets the range within 1 standard deviation from the state center value as the normal standard range, the range from 1 to 2 standard deviations as the caution standard range, and the range exceeding 2 standard deviations as the candidate range for warning or emergency judgment.
[0218] For example, if the average heart rate value in the initial period of wearing by the worker is calculated to be 78 bpm and the standard deviation is 8 bpm, the state determination unit (250) determines the normal, caution, warning, and emergency stages based on the degree to which the heart rate value measured thereafter deviates from the state center value and state range included in the worker's standard state information.
[0219] According to this method, the reference state learning unit (295) does not use a fixed personal profile even in an environment where the equipment is not issued to the same worker every time, and forms a reference state model using newly input data in the corresponding wearing section.
[0220] Therefore, the problem of incorrectly applying standards based on the previous wearer or previous usage history to the current worker is reduced, and the environmental conditions of different work locations, such as a confined area, a high-temperature area, or a general work area, even within the same industrial site are reflected as the field standard condition information of the corresponding wear section, thereby increasing the judgment accuracy of the condition judgment unit (250).
[0221] At this time, the state determination unit (250) compares the measured biometric data and environmental data with the worker standard state information and field standard state information formed by the standard state learning unit (295), and determines the worker's state as one of four stages: normal, caution, warning, and emergency, using the degree to which each measured value deviates from the standard state information.
[0222] Additionally, the standard state learning unit (295) updates the worker standard state information and the field standard state information by reflecting the biometric data and environmental data of the section determined to be normal or cautionary by the state determination unit (250) as additional learning data even after the start of work, and conversely, the data of the section determined to be warning or emergency by the state determination unit (250), the section where an impact signal or sudden change in posture is detected by the inertial measurement unit (170), and the section where an emergency keyword is recognized by the emergency voice recognition unit (260) are excluded from the update target of the standard state information.
[0223] Accordingly, the on-device AI station (200) does not judge all workers equally based solely on a fixed threshold, but applies on-device AI-based judgment criteria that reflect the individual worker's biological characteristics and the environmental characteristics of the site.
[0224] At this time, the standard state learning unit (295) stores the formed and updated worker standard state information and field standard state information in the internal storage unit (241), and subsequently allows them to be referenced in the judgment process of the state judgment unit (250), cross-fusion judgment module (253), and scenario classification module (254).
[0225] Therefore, the on-device AI-based intelligent assistant and emergency situation recognition and judgment system forms and updates the reference state model of the relevant wear section internally at the site and reflects this in the determination of emergency causes and emergency scenarios.
[0227] Furthermore, the QR interpretation and work guidance unit (280) interprets the QR code obtained by the QR recognition camera unit (220) to generate voice guidance, and when a worker request voice signal is transmitted through the voice communication unit (180), it generates voice guidance by reflecting the work guidance request content included in the worker request voice signal.
[0228] The above QR code may be a code attached to equipment, a code displayed on a process chart or work chart, a code for worker identification, etc., and the QR interpretation and work guidance unit (280) extracts at least one of equipment identification information, process identification information and worker identification information from the above QR code.
[0229] Specifically, the QR interpretation and work guidance unit (280) matches the extracted identification information with corresponding information among the work manual, production schedule, and work status stored in the internal storage unit (241).
[0230] If equipment identification information is extracted, the inspection procedures, operating conditions, maintenance history, or precautions for handling hazardous materials of the relevant equipment are retrieved; if process identification information is extracted, the work sequence, input materials, inspection items, or next work step of the relevant process are retrieved; and if worker identification information is extracted, the assigned task, work authority, or current work status of the relevant worker is retrieved.
[0231] Additionally, when a worker request voice signal is additionally transmitted, the QR interpretation and work guidance unit (280) extracts work guidance request content regarding the inspection order, work procedure, production schedule, work status, precautions, or next work step from the worker request voice signal.
[0232] In this case, the QR interpretation and work guidance unit (280) selects information corresponding to the work guidance request content from among the information called in response to identification information extracted from the QR code, and generates voice guidance based on the selected information.
[0233] Additionally, the QR interpretation and work guidance unit (280) generates selected information in response to the called information or work guidance request content as voice guidance that the worker can listen to.
[0234] At this time, the QR interpretation and work guidance unit (280) does not output the document-type work manual or work status data stored in the internal storage unit (241) as is, but organizes key procedures, precautions, or verification items that the worker must immediately follow at the site in the form of voice guidance, so that the worker can receive work guidance without operating a separate portable terminal even while wearing gloves or holding a tool.
[0235] The generated voice guidance is output to a single speaker (120) through the voice communication unit (180) of a mono headset-type smart wearable device (100).
[0236] Accordingly, the worker recognizes the QR code of the field equipment or work schedule through the QR recognition camera unit (220), and then listens to the work instructions corresponding to the equipment or process through the single-sided speaker (120).
[0237] Additionally, if necessary, the worker requests additional work guidance by voice, and the QR interpretation and work guidance unit (280) outputs voice guidance reflecting the above work guidance request to the single-sided speaker (120).
[0238] At this time, according to the present invention, since the single-sided speaker (120) provides sound to only one ear of the worker, the worker can hear the sound of equipment operation, warning sound, or the voice of a nearby worker through the other ear while listening to the work instructions.
[0239] Additionally, the voice guidance output of the QR interpretation and work guidance unit (280) can be subject to priority control by the event management unit (290). Normally, work guidance corresponding to the QR code and additional work guidance corresponding to the worker request voice signal are output through the single-sided speaker (120). When an emergency judgment by the status judgment unit (250) or emergency keyword recognition by the emergency voice recognition unit (260) occurs, the event management unit (290) stops the work guidance output of the QR interpretation and work guidance unit (280) or switches it to a lower priority.
[0240] At this time, the intelligent assistant and emergency situation recognition and judgment system of the present invention prioritizes emergency response operations even while QR-based work support and work guidance based on worker request voice signals are being provided.
[0242] Meanwhile, biometric data, environmental data, motion data, and voice signals generated from a mono headset-type smart wearable device (100) are transmitted to an edge computing server (240) through the internal network (230) of an on-device AI station (200).
[0243] At this time, as described above, the bio-data is data corresponding to the heart rate value, body temperature value, SpO2 value, PPG raw signal, HRV value, or RR value generated by the bio-sensor unit (150), the environmental data is data corresponding to the temperature value, humidity value, atmospheric pressure value, VOC sensor output value, gas sensor output value, or CO2 concentration value generated by the environmental sensor unit (160), and the motion data is data corresponding to the 3-axis acceleration value, 3-axis gyroscope value, free fall signal, impact signal, or attitude change generated by the inertial measurement unit (170).
[0244] The edge computing server (240) stores the original data of the biometric data, environmental data, motion data, and voice signal in the internal storage unit (241), and also stores the inference results and log data generated during the processing of the state determination unit (250), emergency voice recognition unit (260), voice communication processing unit (270), QR interpretation and work guidance unit (280), and event management unit (290) in the internal storage unit (241).
[0245] Here, the original data refers to data received directly from each sensor or microphone (130) or data arranged together with time information, and the inference result includes the emergency cause determination result of the state determination unit (250), the emergency scenario determination result of the scenario classification module (254), the emergency keyword recognition result of the emergency voice recognition unit (260), and the priority control result of the event management unit (290).
[0246] At this time, the log data includes the time of data reception, the time of judgment, the time of event occurrence, the history of voice guidance output, the history of administrator connection, and the history of emergency response processing.
[0247] The edge computing server (240) does not transmit the original data, inference results, and log data to an external cloud, but stores and processes them within the on-device AI station (200).
[0248] Accordingly, data that may include a worker's physical condition or on-site utterances, such as biometric data and voice signals, is not transmitted to an external server, and status determination, emergency keyword recognition, work guidance generation, and event priority control can be performed within the equipment inside the industrial site.
[0249] In addition, the on-device AI station (200) continues to process biometric data, environmental data, motion data, and voice signals using the internal network (230) and the edge computing server (240) even when the external network connection is interrupted or delayed, thereby maintaining emergency cause determination, emergency scenario determination, emergency keyword recognition, and work guidance priority control within the site, even in underground areas, confined spaces, plants, factory interiors, or industrial sites where external network access is restricted.
[0250] In this way, the intelligent assistant and emergency situation recognition and judgment system stores worker-related data generated from a mono headset-type smart wearable device (100) and inference results and log data generated from an on-device AI station (200) in an internal storage unit (241), and secures data security and continuity of emergency response through an on-site internal processing structure that does not transmit to an external cloud.
[0252] Below, a control method using the aforementioned intelligent assistant and emergency situation recognition and judgment system is described.
[0253] Since the control method of the present invention is performed on the premise of the intelligent assistant and emergency situation recognition and judgment system described above, the names, symbols, connection relationships, meanings of data, and basic operations of each component mentioned in the control method are understood to be identical to the contents of the system described above.
[0254] Therefore, below, descriptions that overlap with the structure of each component are omitted or briefly described, and the processing flow of the data acquisition step (S10), data transmission step (S20), reference state learning step (S30), state determination step (S40), emergency voice recognition step (S50), work guidance output step (S60), and event control step (S70) is explained in detail.
[0255] In the above data acquisition step (S10), the biosensor unit (150), the environment sensor unit (160), and the inertial measurement unit (170) of the mono headset type smart wearable device (100) acquire the biosensor data of the worker, the environment data around the worker, and the motion data of the worker at the same time through time synchronization.
[0256] This data acquisition step (S10) does not measure only the physical condition of the worker or only the field environment separately, but rather groups the biological condition, environmental condition, and motion condition occurring at the actual location of the worker into a single time standard, and accordingly, it is possible to distinguish whether a rise in body temperature occurred together with a high temperature and humidity environment, whether a decrease in oxygen saturation value occurred together with an increase in CO2 concentration value, or whether an impact signal occurred together with a change in posture.
[0257] In the above data transmission step (S20), the data collection and transmission unit (210) of the on-device AI station (200) transmits the biometric data, environmental data, and motion data acquired in the data acquisition step (S10) to the edge computing server (240) through the internal network (230).
[0258] In this data transmission step (S20), the data required for determining the worker's status is not first transmitted to an external cloud, but is instead transmitted to a processing path within the on-device AI station (200).
[0259] Therefore, even in industrial sites with unstable external network connections, the flow of data collection and status assessment is not interrupted, and the data required for emergency situation assessment is rapidly processed on-site.
[0260] In the above reference state learning step (S30), the reference state learning unit (295) of the edge computing server (240) forms and updates worker reference state information and field reference state information.
[0261] At this time, the reference state learning unit (295) uses biometric data and environmental data collected during the initial period of wear by the worker as reference state learning data, and calculates the average and standard deviation of each data item to form worker reference state information and field reference state information for the corresponding wear period.
[0262] Additionally, the standard state learning unit (295) updates the worker standard state information and the field standard state information by reflecting the biometric data and environmental data of the section determined to be normal or cautionary by the state determination unit (250) as additional learning data even after the standard state learning stage (S30). At this time, the data of the section determined to be a warning or emergency state, the section where an impact signal or sudden change in posture is detected, and the section where an emergency keyword is recognized are excluded from the update of the standard state information.
[0263] Through this standard state learning step (S30), the state determination unit (250) determines the state based on the newly formed and updated worker standard state information and the field standard state information in the corresponding wearing section, without applying the same fixed threshold value to all workers or calling up the long-term personal profile of a specific worker.
[0264] As a result, even in operating environments where equipment is issued to different workers each time, the problem of previous users' standards being incorrectly applied to the current worker can be reduced, and the reliability of normal, caution, warning, and emergency judgments can be increased by reflecting the worker's current vital signs and the current environmental conditions of the work location.
[0265] In the above state determination step (S40), the biomarker analysis module (251) calculates biomarker abnormality signs from biomarker data, the environmental indicator analysis module (252) calculates the environmental risk level from environmental data, and the cross-fusion judgment module (253) maps the biomarker abnormality signs, environmental risk level, and operation data by synchronizing them in time.
[0266] At this time, the scenario classification module (254) classifies the cause of the emergency into one of individual physiological abnormalities, environmentally induced abnormalities, and complex abnormalities based on the mapping results, and determines the emergency scenario.
[0267] In this state determination step (S40), an alarm is not generated based solely on whether a single sensor value is exceeded, and the cause of the emergency is distinguished by using a combination of biometric data, environmental data, and motion data.
[0268] Therefore, even if the same increase in heart rate occurs, if high temperature and humidity conditions are accompanied by an increase in body temperature, it is identified as heatstroke or thermal abnormality, whereas if irregular changes in heart rate intervals occur while environmental data is within the normal range, it can be classified as an individual physiological abnormality.
[0269] In the above emergency voice recognition step (S50), the emergency voice recognition unit (260) recognizes the emergency keyword offline from the voice signal received from the microphone (130).
[0270] In this emergency voice recognition step (S50), an emergency event is generated by only utterances such as "help," "danger," "rescue," and "emergency," even in a situation where the operator cannot operate the terminal with their hands.
[0271] In addition, since the voice signal is not transmitted to an external voice recognition server but is processed within the edge computing server (240), emergency keyword recognition will be possible even in sites with communication delays or external network failures.
[0272] In the above work guidance output step (S60), the QR interpretation and work guidance unit (280) generates voice guidance corresponding to a work manual, production schedule, or work status from the QR code obtained by the QR recognition camera unit (220), and outputs the generated voice guidance to the single speaker (120) of the mono headset type smart wearable device (100).
[0273] Additionally, when a worker request voice signal is transmitted through the voice communication unit (180), the QR interpretation and work guidance unit (280) generates a voice guidance by reflecting the work guidance request content included in the worker request voice signal, and outputs the generated voice guidance to the single-sided speaker (120).
[0274] In this work guidance output step (S60), the worker receives voice guidance on the inspection procedure of the equipment, work sequence by process, precautions, or work status simply by recognizing the QR code displayed on the equipment, process chart, or work chart.
[0275] Additionally, if necessary, the worker requests additional work guidance by voice, and the QR interpretation and work guidance unit (280) outputs voice guidance reflecting the above work guidance request.
[0276] Therefore, the worker can receive work support without operating a separate portable terminal while wearing gloves or using tools, and can perceive surrounding warning sounds or equipment operation sounds while listening to guidance through the single-sided speaker (120).
[0277] In the above event control step (S70), the event management unit (290) checks whether an emergency determination of the status determination step (S40) or an emergency keyword recognition of the emergency voice recognition step (S50) has occurred.
[0278] At this time, if no emergency judgment or emergency keyword recognition occurs, voice communication between the worker and the voice guidance of the work guidance output step (S60) is maintained.
[0279] On the other hand, if an emergency judgment or emergency keyword recognition occurs, the event management unit (290) stops the voice guidance output of the work guidance output step (S60) and switches the voice communication processing unit (270) to the administrator terminal priority connection channel.
[0280] Accordingly, work support functions are provided under normal circumstances, while in emergency situations, rescue requests, connection to an administrator, and emergency guidance take priority over work guidance.
[0281] Additionally, the original data, inference results, and speech recognition results from the data acquisition step (S10) to the event control step (S70) are not transmitted to an external cloud, but are stored and processed in the internal storage unit (241) of the edge computing server (240) or the internal processing area of the on-device AI station (200).
[0282] Accordingly, this control method reduces the transmission of workers' biometric data and voice signals to external servers, and continuously performs intelligent assistant and emergency situation recognition and judgment functions internally, even in industrial sites where external network connectivity is limited.
[0284] According to the present invention as described above, a worker at an industrial site checks the biological state, surrounding environment state, operation state and voice state together while wearing a mono headset-type smart wearable device (100), and processes emergency situation judgment and work guidance within the site through an on-device AI station (200).
[0285] Accordingly, workers can receive voice communication and work guidance while keeping both hands free, and managers can quickly identify workers' vital signs, environmental hazards, falls, or emergency fires at the site level.
[0286] In addition, since the present invention performs data storage, artificial intelligence inference, emergency scenario determination, and event priority control within the on-device AI station (200) without transmitting original data to an external cloud, it continuously provides emergency situation recognition and judgment functions even in underground spaces, enclosed spaces, plants, factories, construction sites, logistics warehouses, and equipment inspection sites where external network connections are unstable.
[0287] Furthermore, by integrating QR code-based work guidance and priority handling of emergencies within a single wearable system, it can enhance both work efficiency and worker safety in industrial settings, thus possessing high industrial applicability.
[0289] Description of the presented embodiments is provided so that a person skilled in the art may use or practice the present disclosure. Various modifications to these embodiments will be apparent to a person skilled in the art, and the general principles defined herein may be applied to other embodiments without departing from the scope of the present disclosure. Thus, the present disclosure is not limited to the embodiments presented herein, but should be interpreted in the broadest possible scope consistent with the principles and novel features presented herein. Explanation of the symbols
[0290] 100: Mono headset-type smart wearable device 110: Main body 120: Single-sided speaker 130: Microphone 140: Support member 141: Tension providing part 150: Biosensor unit 151: PPG sensor 152: Body temperature sensor 155: Functional layer 156: Adhesive layer 157: Conductive layer 158: Waterproof layer 160: Environmental sensor part 161: Temperature sensor 162: Humidity sensor 163: Barometric pressure sensor 164: VOC sensor 165: Gas sensor 166: CO2 sensor 170: Inertial measurement unit 171: 3-axis accelerometer 172: 3-axis gyroscope sensor 180: Voice communication unit 190: Data Communications Department 200: On-device AI Station 210: Data collection and transmission unit 220: QR recognition camera unit 230: Internal network 240: Edge computing server 241: Internal storage unit 250: Status determination unit 251: Biomarker Analysis Module 252: Environmental Indicator Analysis Module 253: Cross-fusion Judgment Module 254: Scenario Classification Module 260: Emergency voice recognition unit 270: Voice communication processing unit 280: QR Interpretation and Work Guidance Department 290: Event Management Department 295: Reference state learning unit S10: Data acquisition stage S20: Data transmission phase S30: Reference state learning phase S40: Status determination stage S50: Emergency voice recognition stage S60: Job guidance output stage S70: Event control stage
Claims
Claim 1 A mono headset-type smart wearable device worn by an industrial site worker comprises: a main body worn on the worker's head; a single-sided speaker provided on one side of the main body to provide sound to only one ear of the worker; a microphone provided on the main body; a support member provided on the other side of the main body and including a tension providing member inside; a biosensor unit positioned to be in close contact with the temporal artery region of the worker by the support member and measuring heart rate, body temperature, and oxygen saturation (SpO2), including a PPG sensor and a body temperature sensor; an environmental sensor unit inserted into a sensor receiving groove formed in the outer housing of the main body and integrally provided with the main body to measure the environment surrounding the worker, including a temperature sensor, a humidity sensor, a barometric pressure sensor, a VOC sensor, a gas sensor, and a CO2 sensor; an inertial measurement unit provided on the main body or the support member and measuring 3-axis acceleration and 3-axis gyroscope; and a voice communication unit connected to the single-sided speaker and the microphone. A monoheadset type smart wearable device comprising: a data communication unit that transmits data from the biosensor unit, the environment sensor unit, and the inertial measurement unit to an external device; wherein the tension providing unit includes an elastic piece, a plate spring, a coil spring, or an elastic wire inserted along the longitudinal direction of the support member, one end of which is supported on the main body side and the other end transmits force to the sensor support end on the biosensor unit side, so that when worn, the support member spread outwardly attempts to return inward by the restoring force of the tension providing unit, thereby pressing the skin contact surface of the biosensor unit toward the temporal artery region, and an opening or porous protective cover through which air passes is formed on the outer side of the sensor receiving groove so that air around the worker reaches the temperature sensor, humidity sensor, barometric pressure sensor, VOC sensor, gas sensor, and CO2 sensor, and a functional layer including an adhesive layer, a conductive layer, and a waterproof layer is formed on the skin contact surface side of the biosensor unit, wherein the waterproof layer is formed limited to the outer periphery so as not to cover the central measurement area. Claim 2 A monoheadset type smart wearable device according to claim 1, wherein the functional layer is composed of a multilayer structure in which an adhesive layer and a conductive layer are laminated on the skin contact surface side of the biosensor part, and a waterproof layer surrounds the outer periphery of the adhesive layer and the conductive layer, wherein the adhesive layer maintains the skin-adhering state of the biosensor part, the conductive layer assists in the transmission of biosignals between the biosensor part and the temporal artery region, and the waterproof layer inhibits sweat or external moisture from entering the measurement surface of the biosensor part. Claim 3 A mono headset type smart wearable device according to claim 1, wherein the voice communication unit and the data communication unit process voice signals and sensor data through separate communication paths using different wireless communication protocols or different channels. Claim 4 A monoheadset type smart wearable device according to claim 1, wherein the biosensor unit generates biodata corresponding to at least one of heart rate variability (HRV) and respiratory rate (RR) based on a raw PPG signal output from the PPG sensor, and the environmental data generated by the environmental sensor unit and the motion data generated by the inertial measurement unit are time-synchronized with the biodata and transmitted through the data communication unit. Claim 5 Mono headset-type smart wearable device worn by industrial site workers; and an on-device AI station that is informationally connected to the mono headset-type smart wearable device and is mounted on the worker's body, work clothes, or protective gear, or carried by the worker; wherein the mono headset-type smart wearable device comprises: a main body worn on the worker's head; a single-sided speaker provided on one side of the main body to provide sound to only one ear of the worker; a microphone provided on the main body; a support member provided on the other side of the main body and including a tension providing part inside; a biosensor unit positioned to be in close contact with the temporal artery area of the worker by the support member and measuring heart rate, body temperature, and oxygen saturation (SpO2), including a PPG sensor and a body temperature sensor; an environmental sensor unit integrally provided on the main body to measure the environment around the worker and including a temperature sensor, a humidity sensor, a barometric pressure sensor, a VOC sensor, a gas sensor, and a CO2 sensor; an inertial measurement unit provided on the main body or the support member and measuring 3-axis acceleration and 3-axis gyroscope; and a voice communication unit connected to the single-sided speaker and the microphone. and a data communication unit that transmits data from the biosensor unit, the environment sensor unit, and the inertial measurement unit to an external device; wherein a functional layer including an adhesive layer, a conductive layer, and a waterproof layer is formed on the skin contact surface side of the biosensor unit, and the on-device AI station comprises: an internal network forming a data transmission and reception path within the on-device AI station; an MCU-based data collection and transmission unit connected to the internal network and collecting sensor data from the data communication unit of the mono-headset type smart wearable device and transmitting it to the internal network; a QR recognition camera unit connected to the internal network and acquiring a QR code; and an edge computing server that is informationally connected to the data collection and transmission unit, the QR recognition camera unit, and the voice communication unit of the mono-headset type smart wearable device through the internal network.The edge computing server comprises: an internal storage unit that stores biometric data, environmental data, motion data, voice signals, inference results, and log data; a state determination unit that determines the cause of an emergency and an emergency scenario using biometric data from the biometric sensor unit, environmental data from the environmental sensor unit, and motion data from the inertial measurement unit; an emergency voice recognition unit that recognizes emergency keywords offline in voice signals received from the microphone of the mono-headset type smart wearable device through the voice communication unit and the internal network; a voice communication processing unit that controls real-time voice communication between workers in conjunction with the voice communication unit of the mono-headset type smart wearable device; and a QR interpretation and work guidance unit that interprets a QR code acquired by the QR recognition camera unit and generates voice guidance by reflecting the work guidance request content when there is a worker request voice signal transmitted through the voice communication unit. The on-device AI-based intelligent assistant and emergency situation recognition and judgment system comprises: an event management unit that controls the operation priority of the state determination unit, the emergency voice recognition unit, the voice communication processing unit, and the QR interpretation and work guidance unit; wherein the state determination unit comprises: a biometric indicator analysis module that calculates signs of biological abnormalities from biometric data from the biometric sensor unit; an environmental indicator analysis module that calculates an environmental risk level from environmental data from the environmental sensor unit; a cross-fusion judgment module that maps the signs of biological abnormalities, the environmental risk level, and the operation data by time synchronization, and classifies the cause of the emergency into one of personal physiological abnormalities, environmentally induced abnormalities, and complex abnormalities; and a scenario classification module that determines an emergency scenario based on the classification result. Claim 6 In paragraph 5, the above biological data further includes heart rate variability values and respiratory rate values generated by processing the raw PPG signal output from the PPG sensor, the above environmental data includes temperature values, humidity values, atmospheric pressure values, VOC sensor output values, gas sensor output values, and CO2 concentration values, and the above motion data includes free fall signals, impact signals, posture changes, and head tilt abnormalities, and the above scenario classification module determines cardiac arrest when a decrease in heart rate, a decrease in oxygen saturation, a decrease in heart rate variability, and loss of respiratory signals occur together based on a combination of the above biological abnormality signs, the above environmental risk level, and the above motion data that are time-synchronized and mapped by the above cross-fusion judgment module, and determines stroke when irregular changes in heart rate intervals and sudden changes in heart rate variability occur together with head tilt abnormalities, and determines syncope when a sudden change in heart rate occurs together with at least one motion data among a free fall signal, an impact signal, and maintaining a horizontal posture, and determines that a decrease in oxygen saturation is accompanied by an increase in CO2 concentration values, an increase in VOC sensor output values, or An on-device AI-based intelligent assistant and emergency situation recognition and judgment system characterized by determining hypoxia or suffocation when it occurs together with at least one of the abnormal gas sensor output value; determining heatstroke or thermal abnormality when body temperature rise and heart rate rise occur together with conditions of rising temperature and humidity values; determining a fall when a free fall signal, an impact signal, and a change in posture occur sequentially; determining arrhythmia when irregular changes in heart rate intervals occur together with abnormal heart rate variability; determining hyperventilation or panic attack when increased respiratory rate occurs together with increased heart rate and oxygen saturation within the normal range; and determining exposure to harmful gases or chemicals when at least one of VOC sensor output value rise, abnormal gas sensor output value, or CO2 concentration value rise occurs together with increased respiratory rate or increased heart rate. Claim 7 In claim 5, the edge computing server receives biometric data and environmental data during the initial wearing period of the worker as reference state learning data, calculates the mean and standard deviation of the reference state learning data to form worker reference state information and field reference state information for the corresponding wearing period, and subsequently includes a reference state learning unit that updates the worker reference state information and field reference state information by reflecting biometric data and environmental data of the period determined to be normal or cautionary by the state determination unit as additional learning data, and the state determination unit determines one of four stages: normal, caution, warning, and emergency based on the deviation between the measured biometric data and environmental data and the worker reference state information and field reference state information. Claim 8 In claim 5, the QR interpretation and work guidance unit extracts at least one of equipment identification information, process identification information, and worker identification information from the QR code, calls corresponding information among work manuals, production schedules, and work status stored in the internal storage of the edge computing server to generate voice guidance, and when a worker request voice signal is transmitted through the voice communication unit, extracts work guidance request content from the worker request voice signal, and generates voice guidance using information corresponding to the work guidance request content among the called information, and the generated voice guidance is output to the single speaker through the voice communication unit of the mono headset-type smart wearable device, characterized by an on-device AI-based intelligent assistant and emergency situation recognition and judgment system. Claim 9 In claim 5, the biometric data, environmental data, motion data, and voice signals generated from the mono headset-type smart wearable device are transmitted to the edge computing server through the internal network of the on-device AI station, and the edge computing server stores the original data, inference results, and log data of the received biometric data, environmental data, motion data, and voice signals in the internal storage unit and processes them within the on-device AI station, and is configured not to transmit them to an external cloud, characterized by an on-device AI-based intelligent assistant and emergency situation recognition and judgment system. Claim 10 In a method for industrial site safety management and work support using a system according to claim 7, a data acquisition step (S10) in which a bio-sensor unit, an environmental sensor unit, and an inertial measurement unit of the mono-headset type smart wearable device acquire the bio-data of the worker, environmental data around the worker, and motion data of the worker by time synchronization at the same time; a data transmission step (S20) in which a data collection and transmission unit of the on-device AI station transmits the acquired bio-data, environmental data, and motion data to the edge computing server through the internal network; a reference state learning step (S30) in which a reference state learning unit of the edge computing server forms worker reference state information and field reference state information for the corresponding wearing period using bio-data and environmental data collected during the initial wearing period of the worker as reference state learning data, and subsequently updates the worker reference state information and field reference state information by reflecting bio-data and environmental data of the period determined to be normal or cautionary state as additional learning data; a bio-indicator analysis module calculates bio-abnormal signs from the bio-data, and an environmental indicator analysis module calculates an environmental risk level from the environmental data, and A state determination step (S40) in which the cross-fusion judgment module classifies the cause of an emergency into one of a personal physiological abnormality, an environment-induced abnormality, or a complex abnormality by mapping the biological abnormality signs, the environmental risk level, and the operation data in time synchronization, and the scenario classification module determines the emergency scenario; and an emergency voice recognition step (S50) in which the emergency voice recognition unit recognizes an emergency keyword offline from a voice signal received from the microphone;A work guidance output step (S60) in which the above QR interpretation and work guidance unit generates voice guidance corresponding to a work manual, production schedule, or work status from a QR code obtained by the above QR recognition camera unit, and when a worker request voice signal is transmitted through the above voice communication unit, generates voice guidance by reflecting the work guidance request content included in the above worker request voice signal, and outputs the voice guidance to a single speaker of the above mono headset type smart wearable device; The control method of an on-device AI-based intelligent assistant and emergency situation recognition and judgment system comprises: an event control step (S70) in which, when the event management unit causes an emergency determination in the state determination step (S40) or an emergency keyword recognition in the emergency voice recognition step (S50) to occur, the voice guidance output in the work guidance output step (S60) is stopped and the voice communication processing unit is switched to an administrator terminal priority connection channel; wherein the original data, inference results, and voice recognition results in the data acquisition step (S10) to the event control step (S70) are not transmitted to an external cloud, but are stored and processed only within the internal storage unit of the edge computing server or within the internal network.
Citation Information
Patent Citations
Method and device for filtering physiological signal motion artifacts of industrial wearable equipment in real time
CN121242496A
Construction site safety management server and construction site safety management method using it
KR1020250046567A
System and method for confined space safety management
KR102958436B1
Wearable deep ground emergency rescue human body monitoring and diagnosing method and system
CN121148692A
Physiological signal sensing patch
KR1020090008786A
