Adaptive Control and Adjustment Technology for Robot Motion Attitude in Complex Terrain Environment

To solve the deficiencies of traditional control methods in dynamic balance, environment perception and impact control, the study proposes model-based predictive control to design the foot-end position mapping and virtual slope attitude adjustment algorithms, and solves the slope terrain control problem through dynamic coordinate conversion and inertia compensation optimization. The optimal predictive control is introduced and a three-prism model is constructed to integrate the center-of-mass stability planning and torso angular momentum adjustment to achieve the dynamic balance of blind stair climbing. Fusion of visual perception using point cloud segmentation modeling with elliptical gait planning. The results show that the depth camera can accurately detect the point cloud plane of the stairs, the maximum deviation of the center of mass trajectory from the expected value is only 0.01 s, the actual and expected pitch angle reaches 0.35 radians and stable tracking at 10 seconds, and the traverse roll angle fluctuation is controlled within ±0.04 radians. The capture points are all within the reachable domain and rise orderly along the gait curve. The method significantly improves the autonomous motion and adaptation ability of the quadruped robot in complex terrain.