DESC:The invention is an AI-driven safety monitoring system (100) designed to proactively detect the presence of dependents, such as children, elderly individuals, or persons with disabilities, in confined spaces like vehicles, rooms, or isolated areas. The system (100) integrates multiple sensor modules—including motion, sound, temperature, door status, and environmental sensors—along with an AI risk assessment engine (105) to identify potential danger scenarios where a dependent may be trapped and unable to act. Once a risk is detected, the system immediately sends real-time alerts via mobile communication modules (106) to caregivers and, if necessary, triggers an emergency response through pre-configured contacts or alarms.
Existing solutions either depend on the dependent's ability to make noise or wear a device, or rely on the caregiver’s memory and discipline, which can easily fail in stressful or distracting situations. This invention addresses these challenges by offering a comprehensive, non-intrusive, and intelligent system that operates independently of the dependent’s actions. It continuously monitors environmental and behavioral parameters to identify risks like silence, lack of motion, rising temperatures, or closed doors that could indicate a trapped dependent. By combining multi-sensor data with AI-driven analysis, the system generates accurate, proactive alerts that ensure rapid caregiver response and prevent accidents or fatalities. This innovation is scalable, affordable, and applicable in a wide range of settings, including homes, vehicles, eldercare facilities, and public transport, significantly improving safety for dependents while reducing caregiver burden.
The objectives are achieved by designing an intelligent multi-sensor safety system (100) that integrates environmental and behavioral monitoring with AI-driven risk assessment. The system comprises sensor modules that detect motion, sound, temperature, CO2 levels, and door status within a confined space. These modules continuously feed data to a central processing unit running trained AI models capable of interpreting sensor inputs to determine the presence of a dependent and assess potential danger scenarios. Once a risk threshold is crossed—such as lack of movement for a specified duration combined with rising temperatures or closed doors—the system triggers alerts. The communication module (106) leverages GSM, Wi-Fi, or other network protocols to deliver real-time notifications to caregivers’ mobile devices. In critical cases, the system automatically contacts pre-configured emergency numbers or initiates audio-visual alarms in the environment to expedite intervention. The AI component refines its detection capabilities over time, incorporating caregiver feedback and system logs to minimize false positives and improve reliability.

The structure/arrangement of the current invention:
The system comprises the following primary components:
? Occupant Detection Module (101): Utilizes motion detectors and infrared sensors to determine the presence and activity level of a dependent in a confined space.
? Environmental Sensor Module (102): Continuously monitors temperature, CO2 levels, and other environmental parameters indicative of unsafe conditions.
? Sound and Voice Module (103): Analyzes ambient sound levels to detect silence or distress noises that may indicate a problem.
? Door and Status Sensor (104): Monitors the opening/closing and locking status of doors or access points to infer potential entrapment scenarios.
? AI Risk Assessment Engine (105): Processes and correlates multi-sensor data using machine learning models to determine the likelihood of a dependent being trapped and in distress.
? Mobile Communication Unit (106): Sends alerts via SMS, push notifications, or automated calls to designated caregivers and emergency contacts when risk thresholds are exceeded.
? Caregiver Dashboard (107): Provides caregivers with a real-time interface to monitor dependent status, receive alerts, review historical logs, and manage system settings.
? Emergency Trigger Module (108): Automatically initiates predefined emergency protocols, including local audio alarms and external emergency contact calls, when critical danger is detected.
All modules operate in a unified, event-driven framework where sensor inputs are continuously processed and correlated. When specific conditions—such as no movement detected for 10 minutes plus a temperature exceeding 40°C—are met, the system transitions from passive monitoring to active alerting. The inclusion of GSM communication ensures alerts can be sent even in the absence of Wi-Fi or internet connectivity. Battery backup modules maintain operation during power failures, and an offline alert system emits audible alarms to draw local attention.
Operational Logic of the System: Upon system activation, each sensor module begins continuous data collection, with sensor thresholds configurable via the caregiver dashboard. Sensor readings are processed in real time by the AI risk assessment engine (105), which correlates multiple parameters to assess risk levels. If high-risk conditions are detected—such as silence plus a temperature spike and a closed-door signal—the system triggers the mobile communication unit (106) to send alerts. The caregiver receives notifications detailing the dependent’s location, risk status, and environmental conditions. If no response is logged, the system escalates to the emergency trigger module (108), initiating automated calls to emergency services or designated contacts.
Example Use Case:
The system can be effectively deployed in a wide range of real-world scenarios where dependents such as children, elderly individuals, or persons with disabilities are at risk of being unintentionally left unattended in confined spaces. One such use case involves a school transportation van. During drop-offs, an elderly caregiver or driver may accidentally overlook a sleeping child in the back seat. The AI-based system, installed within the vehicle, continuously monitors the presence of occupants using motion and sound sensors, as well as ambient environmental conditions such as temperature and CO2 levels.
If no movement or sound is detected for a predefined duration while the doors are closed and the internal temperature begins to rise, the AI risk assessment engine interprets this as a potential hazard. The system then activates the communication module to immediately alert the school authorities and the child’s parents via SMS and app notifications. If no response is received within a certain time frame, the emergency trigger module initiates an automated call to predefined emergency contacts and activates an in-vehicle alarm to prompt immediate human intervention.
This use case demonstrates the invention’s practical application in ensuring the safety of children in school transport systems. Similar deployments can be made in eldercare facilities, private vehicles, daycare centers, or even at home—any setting where vulnerable individuals may be left unattended and proactive detection of their safety is crucial.
The embodiments in Figure 2 provide a system for alerting caregivers about trapped dependents (hereinafter referred to as “system” or “system (200)”), configured to provide an alert on caregivers’ smartphone devices upon detection of trapped dependents. The system comprises a sensor module, a smartphone, and application software configured to be installed and run on the smartphone and the GSM server connecting the sensor with the application software on the smartphone.
In one of the exemplary embodiments of the present invention Figure 2, the system is configured to send alerts generated by the application software that is installed and run on a smartphone carried by the caregivers of the dependents. The dependents may constitute children of age group below 12 years. The alerts may be in the form of a text message or a phone call.
In an implementation of one of the exemplary embodiments of the present invention, structural and functional aspects of the system (200) are explained by referring to Figure 2. The system (200) comprises a sensor module (202) configured to sense a plurality of parameters related to the vehicle and the child. The plurality of parameters may include the state of motion of the vehicle, the presence or absence of the movement of the child, the child’s cry, the interior temperature of the vehicle, and the condition of the vehicle doors. The sensor module (202) is communicatively coupled to the application software installed on a smartphone (203) carried by the caregivers. The application software is configured to generate alerts in the form of a text message or a call, upon detection of adverse conditions such as a stationary state of the vehicle, the child sleeping, or crying, the interior temperature of the vehicle reaching a certain predefined limit of 750C or crosses it, the vehicle doors in locked condition; that occur in isolation or any of their combinations. GSM server (201) connects the sensor module (202) with the application software on the smartphone (203).
ADVANTAGES OF THE INVENTION:
? Operates independently of the dependent’s ability to signal distress, ensuring reliable detection in emergencies.
? Provides proactive, real-time alerts to multiple caregivers through multiple communication channels.
? Monitors a combination of factors—motion, sound, temperature, CO2 levels, and door status—to enhance detection accuracy.
? Is scalable and adaptable for diverse environments such as vehicles, homes, eldercare facilities, and public transport systems.
? Reduces accidental fatalities and improves caregiver response times.
? Incorporates feedback learning mechanisms to enhance system accuracy and reduce false positives.
? Includes battery backup and GSM communication to ensure functionality during power or internet outages.
,CLAIMS:We Claim,
1. A safety monitoring system 100 for detecting trapped dependents in a confined space and alerting caregivers in real time, the system comprising:
an occupant detection module 101 using passive infrared (PIR) sensor, ultrasonic proximity sensor, motion sensor and thermal imaging sensor configured to detect presence and/or activity (motion) of a dependent;
an environmental sensor 102 module configured to monitor ambient temperature, carbon dioxide (CO2) levels via a non-dispersive infrared (NDIR) sensor, and humidity via a capacitive;
a sound and voice analysis module 103 using a microphone array and digital signal processing (DSP) to analyze ambient audio for distress sounds such as crying or prolonged abnormal silence;

a door and status sensor 104 including a magnetic reed switch or Hall-effect sensor configured to detect door open/closed state and locked/unlocked status;
a GSM server communicatively coupling sensor data to caregiver devices and ensuring connectivity during internet outages;
an AI risk assessment engine 105 configured to correlate data from the occupant detection module, environmental sensor module, sound analysis module, and door status sensor to generate a risk signal using sensor-fused machine learning models trained on historical data after satisfying one or more predefined conditions indicative of a trapped dependent;
a mobile communication unit 106 for sending alerts via SMS, application notifications, or automated calls to one or more caregivers;
a caregiver dashboard 107 enabling real-time monitoring, alert management, and configuration of threshold parameters; and
an emergency trigger module 108 configured to initiate local alarms and/or contact emergency services if alerts remain unacknowledged for a predefined duration, and to escalate alerts to secondary contacts if primary caregivers are unresponsive.

2. The system of claim 1, wherein the AI risk assessment engine 105 executes a machine-learning model trained on historical sensor data to compute a risk score based on features derived from occupancy, environmental conditions, sound patterns, and door status.

3. The system of claim 1, wherein the mobile communication unit 106 comprises a GSM modem configured to send SMS messages and place automated voice calls, and a Wi-Fi/LTE modem configured to send application push notifications.

4. The system of claim 1, comprising a caregiver dashboard 107 provisioned on a caregiver’s smartphone or web portal, the dashboard configured to:
(a) display real-time sensor data and AI-computed risk scores;
(b) present historical logs of sensor readings and alert events; and
(c) allow configuration of one or more threshold parameters for the AI risk engine and caregiver contacts.

5. The system of claim 1, comprising a battery backup that supplies power to all the modules to ensure uninterrupted operation during power failures.

6. The system of claim 1, wherein the confined space is selected from the group consisting of a vehicle interior, a room within a residential building, an eldercare facility room, and a public transport compartment.

7. The system of claim 1, comprising an offline alert mechanism that emits an audible alarm in the confined space when network connectivity is unavailable and a high-risk condition is detected.

8. The system of claim 1, wherein the caregiver dashboard 107 is configured to allow caregivers to set geofencing parameters so that threshold sensitivity is adjusted based on the dependent’s geographic location.

9. The system of claim 1, wherein the AI risk assessment engine 105 refines its machine-learning model incrementally based on feedback received from caregivers through the caregiver dashboard 107 indicating whether a prior alert was false positive or genuine.

10. The system of claim 1, comprising a humidity sensor, wherein humidity data is also forwarded to the AI risk assessment engine 105 for multi-parameter correlation.

11. The system of claim 1, wherein the mobile communication unit 106 logs all outgoing communications and timestamps in a secure memory for post-event analysis.

12. A method for detecting trapped dependents and monitoring a dependent in a confined space and alerting caregivers upon detecting a hazardous condition, by utilizing the system as claimed in claim 1 for:
continuously sensing occupant presence and motion data with an occupant detection module;
concurrently monitoring ambient temperature and CO2 level data with an environmental sensor module;
concurrently analyzing ambient audio data with a sound and voice module to detect distress signals or extended silence;
concurrently detecting door open/closed state and locked/unlocked status with a door and status sensor;
transmitting all sensed data to an AI risk assessment engine;
computing a risk score based on correlated multi-sensor inputs and comparing the risk score to a predefined risk threshold;
sending an alert notification to one or more caregiver devices if the risk score meets or exceeds the predefined threshold in the form of an SMS, a push notification via a smartphone application, or an automated voice call to a caregiver;
initiating emergency protocols after predefined interval of unacknowledged alerts, including local alarms and contacting emergency services.

13. The method of claim 12, wherein computing the risk score comprises applying a machine-learning model trained on historical sensor data to feature vectors derived from the motion, temperature, CO2, audio, and door status measurements.