goldiron

Driverless Road Vehicles: Now or Never?

For many decades now, driverless vehicles have been a dream (or a nightmare for some) which seemed unattainable in current traffic situations. However, robotics research has progressed tremendously since the 1960’s when the first robotics vehicles were able to move centimeter by centimeter up to now where, thanks mostly to the various military programs throughout the world, we can demonstrate driverless vehicles in close to real environments running at reasonable speeds.

In parallel (and sometimes jointly) real advances in sensors, signal processing, control, localization and communication have allowed to introduce in today’s cars functions which can take over the control of the vehicle, at least partially or in emergency situations. However, it still seems unrealistic to many to remove completely the driver from the control loop, if only for regulatory or liability reasons.

This can easily be overcome if we simplify slightly the complexity of the driving task and remove the unpredictable elements of driving. This is the approach which was taken when robots were introduced in the manufacturing environment: instead of trying to copy exactly the manual operations (such as bin picking which was one of the great challenges of robotic researchers for many years), the environment was changed to simplify the task of the robots.

As now is being demonstrated in the first driverless road vehicles (driverless metros have been in operation for decades), these vehicles can be introduced in specific environments such as dedicated roads in airports (e.g. the Schiphol ParkShuttles which were in operation in the late 1990’s or the new ULTra vehicles now put in operation at Heathrow airport for a commercial service before the end of 2009). In such environments, the variability of the driving task is greatly reduced, just by removing the vehicles with human drivers (whose behavior can be quite unpredictable). Furthermore, communications between all the vehicles and the infrastructure allow for much simpler overall control.

Research is now called for identifying the conditions which can allow the implementation and the expansion of these first driverless applications, how to design, operate and maintain the right infrastructures and how dual-mode vehicles (which would have access to the existing network in manual or computer assisted mode) could have access to these new infrastructures. This could lead to a new worldwide infrastructure, quite similarly when the freeway infrastructure was introduced 50 years ago!