Apr 26 2010
General Motors (GM) Research Israel’s researchers are working with image processing expert Prof Shai Avidan and his team from the Faculty of Engineering at the Tel Aviv University on the development of enhanced algorithms that will empower mounted cameras on GM cars to sense threats so that the drivers can be warned to make decisions quickly.
General Motors' Artificial Eye for Driving
According to Avidan the main challenge is the development of such a system that is capable of recognizing persons and differentiates them from various objects in motion reliably and quickly. He explained that a model capable of reacting instantaneously to threats will be developed. Finally he expects that research on computer vision will render cars smarter and make roads safer.
A smart camera system that was developed by an Israeli startup firm MobilEye can be incorporated in cars to make them more intelligent. Avidan was a member of the technical team of MobilEye, which developed the system for detecting vehicles and also help in tracking them in real time. He is currently applying this technology for developing state-of-the-art smart cameras that are conversant with their environment. He aims to develop a camera that can differentiate pedestrians against various objects in motion so that warning of a likely accident is provided to the driver through developing algorithms that can identify and track objects.
He explains that this security tool will be useful for preventing the car door from opening automatically when a cyclist races towards the car, helps the driver to change the direction of the car when a child wanders on the street, and act as a double check for other vehicles in the car’s blind-spot.
He expects that cameras that can identify any moving object could be ideal for applications like autonomous vehicles. The related technology is likely to be utilized in computer gaming applications for tracking the movements of a player or for surveillance for the detection of intruders.
Avidan has shown that this technology is able to work on color, grey-scale, and infrared cameras. He revealed that cameras function like dumb machines unless one knows the method to extract information from them. He added that more interesting applications that will keep human lives comfortable and safe will emerge with growth in computer power and drop in camera prices.
The earlier radar-enabled detection systems were not very sensitive to detect human activities and were more expensive, while a smart camera powered by a powerful chip will be able to identify activities of humans and animals, and empower the car to react as per the situation, like locking the car doors or applying the car brakes.