Abstract
During the last decades, mobile robots have been a classic subject for robotics researchers, covering a large number of topics ranging from image processing, SLAM (Simultaneous Localisation and Mapping), all the way to control technologies and swarm techniques. However, these techniques did not see many applications in road transport until the Prometheus Project in Europe (1986-1994) and the AHS (Automated Hignway Systems) in Japan and in the USA and for a long time it was considered by many to be impossible to implement safety critical function in large production road vehicles.
However, some robotics techniques are now finally arriving in production vehicles with systems involving sensing, decision making and control of the vehicle. The first such systems concerned the “longitudinal control” of the vehicle with a radar (or lidar) sensing the distance (and sometime their lateral position) to the vehicles ahead and controlling the acceleration and braking of the “ego vehicle” to maintain a safe distance. Now vision systems can assist the driver to keep his or her vehicle on the lane (“lateral control”). So, how far are we from a fully autonomous vehicle?
The DARPA challenges in 2004 and 2005 have shown us that automated vehicles are feasible in “simple” environments and in particular when we do not have to consider other moving vehicles. The following challenge from DARPA in 2008 (Urban Challenge) tried to address this problem with some promising results. However, we are still far from operating a fully autonomous vehicle in daily traffic, especially in urban environments where the scene complexity is very large. In order to operate fully autonomous vehicles in a realistic way, we therefore have to take the same approach as in the manufacturing industry when the first robots were introduced: simplify the environment.
This is the approach which is being taken now with the cybercars. These vehicles are designed for a fully automated urban transport of passengers or goods and they operate on a road network for on-demand, door to door transport. At the moment, these roads are more or less protected from intrusions by people or other types of vehicles. The cybercars that have been designed and tested in the early 2000’s are now being put in operation in various cities throughout Europe (www.citymobil-project.org). Although their environment has been somewhat simplified, they must use advanced robotics technologies to avoid obstacles and plan their trajectories even when they have to cross the path of other similar vehicles (with which they communicate). One of the challenging problems is the definition of common rules so that interoperability of different vehicles on a same infrastructure is guaranteed.
Biography
Michel Parent is currently the program manager at INRIA of the R&D team on advanced road transport (IMARA research group). This group focuses on research and development of information and communication technologies for road transport and in particular on fully automated vehicles (the cybercars).
Before his current position which he holds since 1991, Michel Parent has spent half of his time in research and academia at such places as Stanford University and MIT in the USA and INRIA in France, and the other half in the robotics industry. He is the author of several books on robotics, vision and intelligent vehicles, and numerous publications and patents. He was the coordinator of the European Project CyberCars between 2001 and 2004 and the follow-up project CyberCars2 (2006-2009). He was involved in many other French European projects on ITS.
Michel Parent has an engineering degree from the French Aeronautics School (ENSAE), a Masters degree in Operation Research and a Ph.D. in Computer Science, both from Case Western Reserve University, USA.
|