Real-time jumping trajectory generation for a one legged jumping robot

Abstract

As a recent trend in humanoid robot research, running motion appears to be a very important issue since faster and faster mobility is required. Being an essential fact, jumping phenomenon seems to be the previous step of succeeding running motion as jumping dynamics play the key role to achieve stable running motion cycles. Therefore, in this paper, authors group sought a method to generate real-time jumping patterns which may ensure the overall stability in every possible phases through a complete jumping cycle. To be able to reach this goal, we discretized ZMP equation in polar coordinates which enabled us to include angular momentum information without referring any predetermined function or set of values. Subsequently, vertical and forward one leg jumping simulations are run on a 3D dynamic simulator applying the aforementioned technique. Simulation results indicate that stable and repetitive jumping cycles are obtained both for vertical jumping and forward jumping cases. Finally, vertical jumping experiments are performed within the result of successful jumping cycles.