Development of a collision warning system using Explainable AI for Indian Urban Traffic

Arun K. Tangirala , Lelitha Devi Vanajakshi

This project aims to develop an explainable AI / ML -based methodology for driver and vehicle safety to provide safety alert to drivers in Indian urban traffic conditions using LIDAR (Light Detection and Ranging) measurements and onboard camera feeds. In the paradigm under consideration, the AI / ML module can provide a warning and possible solution (on the steering direction and speed) to the human in the loop. This is in the spirit of explainable AI as against a fully automated decision-making system driven by AI that would take over the vehicle bypassing the driver. This project aims to develop affordable technology that requires minimal instrumentation onboard and especially focus on Indian traffic conditions that are characterized by heterogeneity and lane indiscipline.