Assisting Vision-Impaired Pedestrians to Cross Streets: An Innovative Accoustic Ranging Approach

This project aims to leverage innovative techniques to develop an intelligent system that assists blind pedestrians to decide when it is safe to cross streets, especially at the uncontrolled crossing locations, where neither traffic lights nor STOP signs are available. While great social resources have been designated to install the Accessible Pedestrian Signals (APS) at intersections to provide guidance instructions to people with visual impairments, it is economic infeasible to equip all intersections/walkways across the nation with such infrastructures. Currently, blind pedestrians mainly rely on themselves to decide if an uncontrolled street is clear to cross. The proposed system consists of two parts: vehicle-side acoustic source and pedestrian-side detection App. The source is designed to emit acoustic signals ranging from 16KHz to 20KHz that are out of human hearing. With the signals captured by the pedestrian’s smartphone, the App then estimates the moving speed, direction, and pedestrian-car distance for each vehicle nearby, so as to decide if any of them would cause collision to the pedestrian. The research team is going to leverage advanced signal processing and localization techniques as well as algorithm design to ensure the detection accuracy and efficiency. As deliverables to this project, the team will prototype the entire system, conduct in-field experiments, publish the results in top conferences, and prepare proposals for external funding opportunities. The success of this pilot project will largely enhance blind and vision impaired pedestrians’ experience on daily commute and bring significant social impact on transportation equality.