Crowd Motion Detection and Prediction for Transportation Efficiency in Shared Spaces

In the shared space scenario where pedestrian crowds and autonomous vehicles coexist, the transportation efficiency of the shared space can be improved by predicting the intention of the crowd and adjusting the driving strategy of the autonomous vehicles. This study proposes a framework that consists of the detection of individual pedestrians in a crowd via both on-vehicle and infrastructure sensors, the prediction of the crowd motion given the vehicle driving strategy, and the evaluation of the transportation efficiency in shared spaces. Methods for pedestrian detection and scenario prediction are introduced. Several aspects for improving transportation efficiency in shared spaces are discussed. Preliminary results of pedestrian detection on individual sensors and a simulation case study for estimating the desired time for an autonomous vehicle to pass the a shared space scenario demonstrate the potential of the proposed framework.