Karuppasamy, Abishek Kumar (The University of Texas at Dallas)

Assessing Vehicle Behavior for Road Safety

Scheduled for presentation during the Regular Session "Driver Assistance Systems II" (FrBT1), Friday, September 27, 2024, 14:50−15:10, Salon 1

2024 IEEE 27th International Conference on Intelligent Transportation Systems (ITSC), September 24- 27, 2024, Edmonton, Canada

This information is tentative and subject to change. Compiled on December 26, 2024

Abstract

In this paper, we address the problem of vehicle behavior assessment for road safety. We propose a solution that leverages the driver’s smartphone, utilizing data from the phone’s GPS to provide real-time assessments and the vehicle’s On-Board Diagnostics (OBD) system to predict the vehicle’s state. The solution’s decision-making process uses a Recurrent Neural Network (RNN) for extracting behavior features of the vehicle and a hybrid RNN-Transformer based approach to predict the vehicle state. We introduce a ’Danger-Level’ metric to assess the potential road risk posed by the vehicle. Our experimental results show that the agent was able to compute the ’Danger-Level’ of the vehicle with an accuracy of 80%. Additionally, the agent computes the future state of the vehicle with a minimal error of 0.207.