אירועים והרצאות בפקולטה למדעי המחשב ע"ש הנרי ומרילין טאוב

Autonomous navigation technologies have matured and improved to the point that, in the past few years, self-driving cars have been able to safely drive an impressive number of kilometers. It should be noted though that, in all cases, the driver seat was never empty: a human driver was behind the wheel, ready to take over whenever the situation dictated it. This is an interesting paradox since the point of a self-driving car is to remove the most unreliable part of the car, namely the human driver. So, the question naturally arises: will the driver seat ever be empty? Besides legal liability issues, the answer to that question may lie in our ability to improve the autonomous navigation technologies to the point that the human can safely be removed from the loop altogether. However, things are not that simple... The primary purpose of this seminar is to raise awareness in the audience of the strong impact that the presence of moving objects in the environment of a robotic system has on its decision-making process when it comes to autonomous navigation. The difficulties that a robotic system has to solve as soon as it tries to navigate in a dynamic environments will be explored. The concept of Inevitable Collisions States (i.e. states for which no matter what the system does next, a collision eventually occurs) will be used to expose all the constraints that arise in a dynamic environment. It will be shown that even when complete information is available (i.e. full knowledge of the environment and its future evolution), safe navigation can sometimes be impossible to guarantee. In real situations where such a complete information is simply not available, things get worse. The seminar will also explore how the models used to represent the environment and its future evolution (e.g. conservative vs. probabilistic) can affect the safety of the navigational decisions that are made. Possible solutions to the safe navigation problem will be presented.

Speaker Biography:

Thierry Fraichard is an INRIA Chargé de Recherche (Research Scientist) in the INRIA Grenoble Rhône-Alpes Research Center. He received his PhD in Computer Science from the Institut National Polytechnique de Grenoble (INPG) in April 1992 for his dissertation on "Motion Planning for a Nonholonomic Mobile in a Dynamic Workspace". In March 2006, INPG awarded him the Habilitation à Diriger des Recherches (Accreditation to Supervise Research) for his work entitled "Contributions to Motion Planning". Thierry Fraichard's research focuses on motion autonomy for mobile robots and vehicles with a special emphasis on motion safety, motion planning (for nonholonomic systems, in dynamic workspaces, and in the presence of uncertainty), prediction of the future motion of moving objects, navigation amidst human beings, and the design of control architectures for autonomous vehicles.