ADOPT: An environmentally-friendly system for alerting drivers to occluded pedestrians traffic

Abstract

The emergence of sensing technologies and vehicular communications has brought significant opportunities for enhancing pedestrian safety on city streets. However, existing solutions rely on costly technologies such as computer vision and trajectory prediction to detect crossing pedestrians, while they have limits in detecting pedestrians who are occluded by parked cars. Despite the presence of collaborative perception by surrounding vehicles and infrastructure, there is a notable absence of incorporating existing parked cars themselves due to their insufficiency in detecting pedestrians and communicating with other cars while they are turned off. Furthermore, accommodating pedestrians on streets has been linked to an additional cost to the environment. This cost is due to the fluctuations in the speed of the car to avoid collisions with pedestrians, which increases fuel consumption and CO2. We first propose to enlist the help of cars parked along the sidewalk to detect and protect crossing pedestrians. In support of this goal, we propose ADOPT: an Environmentally-friendly system for Alerting Drivers to Occluded Pedestrian Traffic. ADOPT lays the theoretical foundations of a system to use parked cars to detect and protect occluded pedestrians. We cope with the resource constraints in parked cars by utilizing short-range and low-power radio frequency sensors to detect pedestrians who also transmit radio signals from energy-harvesting wearables. To estimate fuel consumption and CO2 of cars, we found that the existing estimating approaches for user-specific requirements are not suitable for our goal. We overcome this limitation by using a simple version of the energy demand model knowing the most suitable powertrain efficiency. Thus, in this dissertation, we evaluate the vehicle energy demand model by testing several powertrain efficiencies. This allows us to accomplish our next task in this dissertation. Next, we propose speed reduction schemes based on studying possible scenarios for midblock crossing. In these scenarios, the approaching car receives, in advance, caution messages about crossing pedestrians from ADOPT system. We show that these schemes reduce the fuel consumption and CO2 emissions of approaching cars. With this, we show how ADOPT is an environmentally friendly system without compromising the safety of midblock pedestrians by utilizing parked cars along the street.