A Density-Sensitive IoT System for Detection of Vehicle Submersion and Auto-Buoyancy

Abstract

 Water-related vehicle submersion accidents, such as those in rivers, canals and flooded roads, are becoming a rapidly growing cause of traffic fatalities in the world. An estimated 300,000 people drowned in the world in 2021 and drowning is the third leading cause of unintentional injury-related death in the world, accounting for 7% of all unintentional injury-related deaths. Many of these deaths have been attributed to being people trapped in vehicles who were unable to evacuate in time, a substantial proportion of these incidents.

Introduction/Importance of Study: Vehicular water submersion is a fatal emergency situation which causes loss of thousands of lives every year, and the existing vehicular IoT-based safety systems do not have a physical buoyancy response system to prevent the vehicle sinking.

Novelty Statement: A unique density-sensitive multi-sensor fusion algorithm and a flat inflatable TPU platform (shoes, arms, legs etc. under the vehicle chassis in 3 to 5 seconds using CO2 to create stable level buoyancy which eliminates false alarms due to rain <5 kg/m³ versus genuine submersion >800 kg/m³).

Material and Method: The proposed SIPSAS will use an MS5837 water density sensor, water level sensor, ADXL345 accelerometer, and MPU-6050 tilt sensor along with an Arduino UNO microcontroller. The sensor data are processed is processed by a Kalman filter and a three-stage false alarm reduction process. The TPU inflatable platform (4m x 2m x 0.3m) deploys beneath the chassis and inflates with CO2 cartridge. In the event of confirmed submersion, GPS/GSM modules send out Emergency SMS notifications to the rescue services.

Result and Discussion: Using Python-based sensor simulation in Google Colab, the proposed SIPSAS attained the accuracy of 97%, precision of 91%, and recall of 88% and specificity of 86%. The seven test scenarios—heavy rain, car wash, shallow puddle, and the three real submersion environments—were all properly classified without any false alarms. The average time to inflate the platforms was 3.8 seconds, resulting in a horizontal floating period of more than 10 minutes.

Concluding Remarks:  The SIPSAS offers a physics-based, automatic, and dependable submersion safety system for vehicles that is superior to current IoT-based approaches and addresses key limitations in current buoyancy response systems and suppression of false alarms.

The proposed system was tested with 118 synthetic sensor samples from 7 real-world scenarios (heavy rain, car wash, shallow puddle, flooded road, river, lake, sea water), created in Python with simulation done in Google Colab (Python 3.10). A confusion matrix analysis was used for validation, with an accuracy of 97%, precision of 91%, and recall of 88% and specificity of 86%. This represents a 36% improvement in accuracy from single water-contact sensor baselines. All 4 non-submersion test scenarios had zero false alarms.