Pedestrian and Cyclist Detection Devices for Transit Buses

This SBIR proposal presents a robust and cost-effective pedestrian and cyclist detection and collision warning system for transit buses. It uses inexpensive but robust sensors and incorporates state-of-the-art object detection techniques to enable effective pedestrian and cyclist detection near transit buses in different environmental conditions. The proposed collision warning system exploits measurements and estimates from different sensors about the presence, relative locations, and dynamics of pedestrians and cyclists near the bus, and assesses the risk of collision by combining this information with the movement of the bus and environmental variables. Upon detection of an alert, a warning is relayed to the bus operator in an intuitive fashion for an immediate response. The collision warning system is modular in design and is built on Robotic Operating System, an open source meta-processing operating system which offers integration with other open source machine vision and analytics software such as OpenCV. The proposed system leverages recent innovations and developments in the areas of sensors, sensor exploitation, robotics, and commercial vehicles.