Daniel Koditschek, University of Pennsylvania, USA

Title:  Transitions in the Locomotion of Robots and the Robotics of Locomotion

Abstract: This talk will address in the context of locomotion the interlinked problems of building experimental robots and experiments  with the  robots that get built. Contemporary robotics seems to be reaching a point of transition in both regards. Persistent steady state locomotion can be formally specified in the language of dynamical systems theory and has been achieved experimentally by a growing  variety of legged machines over a diversity of substrates in the presence of substantial disturbances.  It seems time to aim for a theory and practice of transitional locomotion behavior. At the same time, commercial legged robots can now be purchased that that arguably exceed the persistent steady state locomotion capability of any university laboratory platform past, present, and, most likely,  future.  It seems time to aim for a theory and practice of multi-use locomotion platform design and implementation. These two sorts of transitions - in the class of machines academics build; and in the class of behaviors they achieve - intersect in the common problem of  "task specification." This talk will motivate the intersecting need and explore the prospects for a transitional task specification formalism. talk will motivate the intersecting need and explore the prospects for a transitional task specification formalism.