Adaptive legged manipulation: Versatile disturbance predictive control for quadruped robots with robotic arms

Qingfeng Yao, Cong Wang, Jilong Wang, Linghan Meng, Shuyu Yang, Qifeng Zhang, Donglin Wang

Abstract: Highlights•The low-level controller can eliminate the influence of the robotic arm with an appropriate disturbance parameter.•A high-level reinforcement learning estimator is used to estimate the disturbance coefficient with a transferable forward model.•Experiments with different forms and types of robotic arms demonstrate the transferability of our algorithm.