Atlas uses its whole body - legs, arms, torso - to perform a sequence of dynamic maneuvers that form a gymnastic routine.
We created the maneuvers using new techniques that streamline the development process.
First, an optimization algorithm transforms high-level descriptions of each maneuver into dynamically-feasible reference motions.
Then Atlas tracks the motions using a model predictive controller that smoothly blends from one maneuver to the next.
Using this approach, we developed the routine significantly faster than previous Atlas routines, with a performance success rate of about 80%.