In the works: Ultrafast camera for autonomous vehicles

SINGAPORE — Scientists from Nanyang Technological University (NTU) have developed a new ultra-fast camera that can help self-driving vehicles react more quickly to traffic conditions.

Unlike optical cameras, which cannot pick up details in the dark and are “blinded” by bright light, this new camera can capture the slightest movements and objects in real time as it is sensitive to motion.

Developed by Assistant Professor Chen Shoushun from the university’s School of Electrical and Electronic Engineering, the camera — called Celex — is now in its fourth and final prototype phase.

Asst Prof Chen and his team, who have been working on it since 2009, are looking to commercialise the new camera technology through their start-up, Hillhouse Technology Pte Ltd, which is incubated by the university’s innovation and enterprise company, NTUitive. He expects the camera to be ready for the market by end of this year and said that they are already in talks with some autonomous vehicle companies.

Explaining how the camera works, he said that it mimicks how the human eye or animal eye works — to be more specific, the frog eye, which is very sensitive to motion. Instead of capturing the whole scene like a photograph, the camera only captures objects in motion, reducing data output and increasing its processing speed, thus enabling it to instantly process visual data.

Celex can process more than 100 frames per second, compared to the usual 30 frames per second for self-driving vehicles to “see” and analyse their environment.

Without the time lag needed to analyse and process the video feed, the continuous tracking feature and instant analysis of a scene help self-driving vehicles or drones to avoid unexpected collisions.

Celex also has the potential to be used for security and surveillance, as well as for robotics such as in robots that are programmed to do cleaning work.