Computing and Mathematical Sciences Colloquium

Animals and machines interact with their environment mainly through physical contact. Yet the discontinuous nature of contact dynamics complicates planning and control, especially when combined with uncertainty. We have recently made progress in terms of optimizing complex trajectories that involve many contact events. These events do not need to be specified in advance, but instead are discovered by the optimizer fully automatically. Key to our success is the development of new models of contact dynamics, which enable continuation methods that in turn help the optimizer avoid a combinatorial search over contact configurations. We can presently synthesize complex humanoid trajectories in tasks such as getting up from the floor, walking, turning in about a minute of CPU time without informative initialization. When using warmstarts in the context of model-predictive control, one step of trajectory refinement can be done in about 20 milliseconds. This makes real-time model-predictive control applicable to humanoid robots, at least in simulation. In addition to the simulation results, I will describe ongoing work on executing the synthetic trajectories on physical robots, and methods for dealing with the inevitable modeling and estimation errors.