Abstract: Even though there are now a number of legged robots capable of a wide variety of high performance dynamic behaviors, their autonomous operation in ways that are both scalable in scope and sustainable in time has not yet been possible. This is, to some extent, a result of the incompatible formalisms in which continuous control and discrete planning problems have traditionally been addressed. In this talk, I will briefly introduce Constrained Intuitionistic Linear Logic (CILL), a relatively new language that promises to constitute a tractable logical formalism in which hybrid planning problems can both be expressed and reasoned about. This language, however, also needs to be supported by appropriate abstractions of primitive behaviors as well as provably correct ways in which they can be combined. Consequently, I will also summarize our group’s ongoing research in the development and characterization of behavioral primitives as well as scalable compositional methods for combined reactive deployment of multiple behaviors. Finally, as a roadmap of our group’s future research, I will describe several problems that need to be addressed before the practical utility of CILL in this context can be established.
- This event has passed.