Multi-sensory perception is one of the enabling technologies required for driverless cars. Inria and the CEA are exploring probabilistic anti-collision algorithms with the Perfect project, an IRT Nanoelec PULSE program initiative. The system, which is being tested on a Renault Zoé, is built on a “smart” box and embedded sensors—lots of them. The car is equipped with a GPS and inertial measurement unit to determine direction and position; LIDAR (Light Detection and Ranging) sensors to reflect light off of objects to determine their position, much like a bat uses echolocation to “see”; and cameras coupled with artificial intelligence to analyze the objects picked up by the LIDAR sensors.
The Renault Zoé is being test-driven around the MINATEC campus and in and around Grenoble. It is communicating with the infrastructures (cameras, traffic lights, instrumented barriers, etc.) at IRT Nanoelec’s Bridge Technologies labs. It is this two-way communication between vehicle and connected environment that will enhance both safety (by avoiding hidden obstacles) and user services (like finding an available parking space).