At the current stage of the R&D project, the automated cars, based on models from Volkswagen and Citroen, already can follow a pre-programmed route and avoid collisions with sudden obstacles without intervention from the driver. Presently, they compute their trajectories based on the input from their sensors whereas the cameras play the dominant role: The vehicles follow the lane markings as “seen” by the camera; once these markings are faded or become entirely invisible, Mailyn and Martti get lost.
Of course, the VTT researchers assure, this is only the first step; by 2020 the vehicles and their sensing equipment will have a state of evolution that allows them to drive on gravel roads or even on roads covered with snow which is said to be not uncommon in the country crossed by the polar circle.
To get their bearings, the vehicles are equipped with a long-range radar and laser scanners to generate the big picture, plus a short-range radar that provides high-resolution scans of the vicinity. And of course tha cameras: A stereo camera that delivers high-quality images and a thermal (infrared) camera that detects humans in the trajectory and on the side of the road. The sensor kit is complemented by the usual array of inertia sensors that determine directional changes and accelerations as well as GPS/Glonass receivers for positioning. All sensor signals are fed to a central computer that compares them, performs plausibility checks, determines position and direction to go and issues commands to the actuators in the powertrain, steering system and brakes.
And now the communication between the cars and between the car and roadside transmitting stations is adding to the picture. "The communications channel of the automated cars is open, but the messages are not yet fully compliant with the standards. Come autumn, the cars will exchange information in a standard format, also allowing others to talk with them," says researcher Matti Kutila from VTT.