Title: Constraint-Based Robot Motion Planning in Complex Environments Speaker: Russ Gayle, University of North Carolina at Chapel Hill Date/Time: Tuesday, February 28, 2006, 10:30 – 11:30 am Location: Building 980, Room 95 (Sandia NM) Brief Abstract: Much of the work on robot motion planning has focused on robots composed of many rigid links. These works tend to characterize just the geometric aspects of the robot, neglecting accurate interactions with the environment. However, it is important to take into account both the mechanical and physical constraints of these robots as well. This includes how the robot can move as well as how it interacts with the environment. Additionally, it is desirable to allow for robots that are highly articulated or can bend and flex, or otherwise exhibit non-rigid motion. This has many applications, including medical simulation, cable or industrial planning, virtual prototyping, emergency response, or even character animation. Building a system to support this poses several challenges. From a robotics standpoint, the configuration space of a deforable robot can be intractably large. Accurate motion requires robust simulation models for both rigid and deformable bodies. Efficient interaction with complex environments and structures requires sophisticated models for planning, collision detection, response, and dynamics simulation. Additionally, the problem of planning with both physical and geometric constraints has not been greatly studied. This talk will describe the on-going work used in developing a constraint-based planner for these tasks. This includes recent work on a using a technique for efficient an adaptive, error-bounded forward dynamics algorithm for articulated bodies with bounded errors as well as some graphics processor optimizations.CSRI POC: Andy Wilson (505) 844-1089 |