Automated Vehicles Sharing the Road: Surveying Detection and Localization of Pedalcyclists
Jevin Barnett, Nicholas Gizinski, Eduardo Mondragón-Parra, Joshua Siegel, Daniel Morris, Timothy J. Gates, Eva Kaßens-Noor, Peter T. Savolainen
Abstract
Automated Vehicles (AVs) must comply with traffic laws, including those requiring motorists to maintain safe distances when passing pedalcyclists. We review relevant U.S. legislation, statistics of traffic accidents in the U.S. resulting in pedalcyclists fatalities, and present “what if” scenarios for AV and pedalcyclist interactions to illustrate safety-preserving algorithms’ necessary ethical considerations. We identify a need for pervasive cyclist identification to improve situational awareness and propose as a solution the concept of using AV-mounted Bluetooth receivers for identifying and localizing a pedalcyclist's “beaconing” Bluetooth device, e.g., a preexisting, low-cost performance monitoring system. This method estimates the distance-to-bicycle from the receivers’ Received Signal Strength Indicator (RSSI). Experimental results prove its feasibility for coarse pedalcyclist localization. Such a system may augment contemporary vision-based detection systems, thereby improving detection resilience and safety, particularly in dark or poor weather conditions.