A Wearable Sensor Network for Cyclists Safety in Mixed Traffic, a Pilot Study

The coexistence of autonomous vehicles and humans in public traffic represents a challenging space sharing conflict. In fact, coordination in road traffic is usually made possible by the capacity of humans to communicate their intentions to the fellow road users. With the advent of autonomous vehicles, new ways to communicate intentions should be investigated. To increase the safety of cyclists in mixed traffic is important to early recognize their intentions. This work aims to lay the foundations for augmenting bicycle frames and cycling garments with sensors. To this end, a custom wearable sensor network comprised of three inertial measurement units and a pedaling cadence sensor is proposed and used in a field study with eight participants. The aim of this pilot study is to early-identify the intention of cyclists to perform a left turn, that in right-handed traffic, represents a potentially dangerous maneuver. Collected data are analyzed and annotated, with the support of video, to classify the actions performed by a cyclist when approaching a left turn. Three events where identified: (i) turning of the head, (ii) interruption of pedaling, (iii) left hand out. Results show how the proposed methodology represents a feasible solution to early-identify the intention of a cyclist to turn left.