PedLoc: UWB-based pedestrian localization for autonomous vehicles

Abstract

Perception is a key technology for autonomous vehicles. Current sensor-based approaches have limitations detecting pedestrians in non-line-of-sight environments. We propose PedLoc, a UWB-based pedestrian localization system that can accurately estimate the pedestrians’ position even in non-line-of-sight environments. PedLoc addresses the mobility, deployment, and obstacle challenges of existing UWB localization schemes in vehicular environments. We presents a overhearing based PedLoc protocol with a novel peak detection algorithm for vehicular environments. We have evaluated PedLoc through intensive experimentation on real world roads. Experimental results demonstrate that PedLoc outperforms existing solutions, successfully tracking pedestrians in complex road environments.