MUMBAI, India, June 30 -- Intellectual Property India has published a patent application (202641056293 A) filed by Rajalakshmi Engineering College on May 04, 2026, for Behavioral Crash Detection System For Two-Wheelers Using Rider Stability Index And Context- Aware Tin.
Inventors include Dr. M. Thangamani; V. Govarthan; B. Dinesh; R. Rakesh; and S. Santhosh.
The application for the patent was published on June 26, 2026, under issue no. 26/2026.
Abstract: A behavioral crash detection system (101) for a two-wheeler rider is disclosed. The system comprises a multi-sensor set including an inertial measurement unit (104). an edge processor (110) executing a TinyML model, a sensor fusion engine (111), a Rider Stability Index engine (112), a context-aware crash verification engine (113), a GPS module (108), and a GSM communication module (109). Sensor data are processed locally to compute a Rider Stability Index representing rider stability over time. Crash verification is performed using impact detection, motion-pattern evaluation, temporal RSI collapse, and post-event position confirmation to reduce false positives. Upon confirmation, an emergency escalation controller (114) provides a rider cancellation window and, in the absence of cancellation, transmits a location- enabled emergency alert without requiring internet connectivity. In optional embodiments, the system supports pre-crash risk warning and privacy-preserving federated model updates. A novel behavioral crash detection mechanism for two-wheelers through the use of a Rider Stability Index (RSI) along with context-aware TinyML inference at the edge for real-time crash detection is presented. This mechanism involves the use of sensors integrated within the two-wheeler, such as accelerometers, gyroscopes, and positioning units, which enable capturing rider behavior by measuring motion dynamics and contextual variables. The calculation of the RSI takes into account various factors, such as tilt angles, acceleration variation, braking power, and trajectory difference in assessing rider stability. An edge-based machine learning algorithm that can run on resource-constrained devices allows for quick and independent detection without relying on cloud services. Context awareness enables distinguishing between regular disturbances experienced while riding and actual crashes. In case a crash occurs, the system automatically sends out emergency alerts along with geolocation details.
Disclaimer: Curated by HT Syndication.
