Bionic Jumping of Humanoid Robot via Online Centroid Trajectory Optimization and High Dynamic Motion Controller
Xiangji Wang,Wei Guo,Zhicheng He,Rongchao Li,Fusheng Zha,Lining Sun
DOI: https://doi.org/10.1007/s42235-024-00586-4
IF: 2.995
2024-09-05
Journal of Bionic Engineering
Abstract:The dynamic motion capability of humanoid robots is a key indicator for evaluating their performance. Jumping, as a typical dynamic motion, is of great significance for enhancing the robot's flexibility and terrain adaptability in unstructured environments. However, achieving high-dynamic jumping control of humanoid robots has become a challenge due to the high degree of freedom and strongly coupled dynamic characteristics. The idea for this paper originated from the human response process to jumping commands, aiming to achieve online trajectory optimization and jumping motion control of humanoid robots. Firstly, we employ nonlinear optimization in combination with the Single Rigid Body Model (SRBM) to generate a robot's Center of Mass (CoM) trajectory that complies with physical constraints and minimizes the angular momentum of the CoM. Then, a Model Predictive Controller (MPC) is designed to track and control the CoM trajectory, obtaining the required contact forces at the robot's feet. Finally, a Whole-Body Controller (WBC) is used to generate full-body joint motion trajectories and driving torques, based on the prioritized sequence of tasks designed for the jumping process. The control framework proposed in this paper considers the dynamic characteristics of the robot's jumping process, with a focus on improving the real-time performance of trajectory optimization and the robustness of controller. Simulation and experimental results demonstrate that our robot successfully executed high jump motions, long jump motions and continuous jump motions under complex working conditions.
robotics,engineering, multidisciplinary,materials science, biomaterials