Seminar: J.-P. Laumond, "Motion planning: What does it move for?", May 27 at 14:30
Title: Motion planning: What does it move for?
Time and location: Friday, May 27, 2011, 14:30, room A7
Abstract: Motion planning has a long and successful story (not as long as the 50 year history of Robotics, but almost!). Consequently, a strong and rich background is available to today’s researcher. This background is often presented as being either theoretical or practical. In this talk I would like to take another point of view. A lot of results in motion planning have been steered by the “toy” problems. In a toy problem, the robot is hyped up to be a justification for theoretical developments. Such developments have found very nice applications in fields outside robotics, such as PLM, automation, computer animation or bio-informatics, i.e. in virtual and virtualized worlds. But, the success in making robots move with “algorithmic motion planning” is less pronounced. One clear reason is the lack of connection with perception. Another reason comes from the “motivation” for a robot to move. Motion planning does not exist per se. A robot moves to do something. We should then rather talk about action planning. However, we will see that trying to connect action planning and motion planning is not a trivial task. This question represents a current challenge for robot motion modeling and planning algorithms.
Bio: Jean-Paul Laumond (IEEE Fellow) is Directeur de Recherche at LAAS-CNRS in Toulouse, France. With his group Gepetto (www.laas.fr/gepetto) he is exploring the computational foundations of anthropomorphic motion. He has been a Coordinator for two European Esprit projects, PROMotion (1992–1995) and MOLOG (1999–2002), both dedicated to robot motion planning technology. During 2001–2002, he created and managed Kineo CAM, a spin-off company from the LAAS-CNRS to develop and market motion planning technology. From 2005 to 2008 he was a co-director of JRL, a French-Japanese CNRS-AIST laboratory dedicated to humanoid robotics. He teaches robotics at the ENSTA and Ecole Normale Superieure in Paris. He published more than 150 papers in international journals and conferences in computer science, automatic control, robotics and neurosciences. He is a member of the IEEE RAS AdCom.
Contact: Giuseppe Oriolo (oriolo@dis.uniroma1.it)