Sven Koenig (University of Southern California)
Tuesday, 24.6.2008, 11:30
Path planning for robots in unknown terrain has been studied in both theoretical robotics and theoretical computer science. However, empirical robotics researchers have often developed their own planning methods. These planning methods have been demonstrated on mobile robots that solve complex real-world tasks and perform well in practice. I will discuss some of these planning methods (namely, Planning with the Freespace Assumption, Greedy Mapping and Greedy Localization) and show how to use tools from graph theory to analyze the resulting plan quality. These results provide guidance for empirical robotics researchers when to use which method and how to improve on them.
This is joint work with C. Tovey, Y. Smirnov and our students.