A Search-based Control Architecture for Wheel-quadruped Robot Obstacle Negotiation

This paper constructed a two-level hierarchical control architecture for legged locomotion when wheel groups are locked, which enables the robot to negotiate rough terrain. A novel wheel-quadruped hybrid mobile robot with parallel driving mechanism based on Stewart 6-DOF platform, the 'BIT-NAZA' robot, is presented. The high-level planner selects a sequence of candidate footholds and achieves COG (center of gravity) trajectory planning, in which the graph search, optimizing approach, and best-first search are combined together. The low-level controller acquires the swing foot trajectories and COG attitude-adjusted criterion, and tracks these desired loci to steer robot obtain preferable stability margin during walking process. Otherwise, a simulated experiment is conducted to verify the controller sufficiency for traversing 3D challenging terrain with forbidden areas including ditches and stair edges.