Co-design of imaging systems and traffic paint

PI: Aswin Sankaranarayanan

Co-PI(s): N/A

University: Carnegie Mellon University

Industry partner: PPG Industries

Over the last many decades, there has been little innovation in the design of the paint used in traffic markings. Today, traffic paint is primarily designed to convey basic information about the layout of the street and the lanes within, while ensuring basic visibility in adverse driving conditions along with resilience to the harsh environment induced by weather and the vehicles themselves. This project aims to enhance the functionality of traffic paint using sophisticated imaging systems that have a perception beyond that of the human eye. We aim to manipulate the spectral properties of paint and combine it with a spectrally sensitive camera, to encode information on the street. This allows us to significantly enhance the amount of information that the environment/infrastructure can present to an assisted or autonomous system. At the same time, the design of the paint will use metamerism to ensure little or no perceptual difference in the human eye. Together, this technology will provide a richer control of the environment as seen by a machine perception system, thereby reducing computational overhead at inference, and leading to a safer driving experience. Such a technology would be an invaluable accelerator for the broader autonomous driving ecosystem in Western PA.