Object Detection and Distance Estimation for Vehicular Platoon Verification Using a Monocular Camera

Andres Mendoza (Klipsch School of Electrical and Computer Engineering)
Alexander Choi (University of Arizona)
Ameer Nessaee (University of Arizona)
Michael Villasana (University of Arizona)

The concept of vehicular convoys is intended to address challenges regarding transportation safety, fuel efficiency, and carbon emissions. Autonomous vehicles are a strong candidate for platoons due to their vastly lower latency of agent response in comparison to human-driven alternatives. However, without human intervention, autonomous vehicles are at risk from adversarial attacks that may incite harmful or life-threatening consequences. Proof of Following (PoF) is a concept that will tackle this problem to securely locate, identify and verify a candidate vehicle. This paper focuses on the development of a secure computer vision-based protocol in order to increase resistance against adversaries. More specifically, we implement an algorithm that localizes a vehicle then estimates its distance from the verifier. Additionally, we use this measured distance to verify that a candidate vehicle is within a specified range of its claimed distance. The accuracy of the predicted estimation is evaluated using the KITTI dataset.