A humanoid robot toying a spinning top: Analysis and design

There exists certain unusual and highly demanding human motor tasks that require bimanual high-end dexterous skills, such as throwing a curve ball, skating, RipStik caster boarding, to name a few, wherein precise fast nonlinear dynamics and advanced robust control play a major role. Similarly, the spinning top (ST), an amazingly popular toy across cultures from millennia, has recently motivated formal research due to its apparent complex perplexing physics that maintains it upright without falling even when its rotational axis is inclined, a motion regime we call “dancing”. In this paper, a humanoid robot (HR) that replicates the throwing and dancing motion regimes of a 6D ST toy (STT) is proposed, using full nonlinear models. A finite-time continuous robust, yet model-and chattering-free controller is synthesized for the HR to track the aggressive but smooth trajectories from a real-time motion capture system of an expert human player. Simulations show the successful throwing and dancing of STT. Moreover, analysis suggests a trade-off on theoretical and practical aspects subject to current HR technological limitations.