Hierarchical Dynamic Control Framework for a Hexapod Robot in Complex Terrain

ABSTRACT Legged robots are capable of performing locomotion tasks in unstructured terrains. However, research on dynamic control methods for hexapod robots in complex environments remains limited. This paper proposes a hierarchical dynamic control framework tailored for hexapod robots navigating complex terrains. A novel Adaptive Slope Saturation Filter is proposed, which synergistically combines a first‐order low‐pass filter and a nonlinear saturation function to achieve adaptive filtering of attitude angles and smooth limitation of output angles. A Model Predictive Control model predictive control–based feedforward method is designed to compute foot‐ground forces for the hexapod's dynamic model, while multitask prioritized control is realized through null‐space projection. Whole‐Body Control enables relaxed optimization of joint torques. Furthermore, Foot‐end Cartesian space force Compensation based on attitude adjustment enhances terrain adaptability. Implemented on the Robot for Embodied Neural Systems, Hexapod 3 hexapod robot, the proposed framework demonstrates natural adaptability across diverse complex terrains, enabling robust sensorless blind navigation.