基于高斯过程模型预测控制的人形机器人运动控制  

魏嘉梁 孔令云 邹星赞 尹俊仁 赵明伟  

西京学院，陕西省西安市，710100；  

摘要：人形机器人的动态平衡控制面临模型简化与实时性矛盾的关键挑战。传统方法如零力矩点（ZMP）控制存在鲁棒性差、弹簧倒立摆（SLIP）模型存在滞后性，而传统非线性模型预测控制（NMPC）在模型失配时性能下降。本文针对这些问题，提出一种融合单刚体模型（SRBD）、高斯过程（GP）与模型预测控制（MPC）的改进框架，旨在提升复杂环境下的运动稳定性和实时性。  

关键词：人形机器人；模型预测控制；单刚体模型；高斯过程  

参考文献

[1].Kajita S, Kanehiro F, Kaneko K, et al. Biped walking pattern generation by using preview control of zero-moment point[C]//2003 IEEE international conference on robotics and automation (Cat. No. 03CH37422). IEEE, 2003, 2: 1620-1626.

[2].Christian Ott、Alin Albu-Schäffer, et al. (2005).A hybrid system approach to model predictive control of legged robots. IFAC World Congress.

[3].Koolen T. Capturability-based analysis and control of legged locomotion[J]. Delft University of Technology, thesis for the degree of Master of Science, 2011.

[4].Bledt G, Kim S. Implementing regularized predictive control for simultaneous real-time footstep and ground reaction force optimization[C]//2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2019: 6316-6323.

[5].Sleiman J P, Farshidian F, Minniti M V, et al. A unified mpc framework for whole-body dynamic locomotion and manipulation[J]. IEEE Robotics and Automation Letters, 2021, 6(3): 4688-4695.