Kinematics-searched framework for quadruped traversal in a parallel robot

Purpose: In this research, the authors established a hierarchical motion planner for quadruped locomotion, which enables a parallel wheel-quadruped robot, the “BIT-NAZA” robot, to traverse rough three-dimensional (3-D) terrain. Design/methodology/approach: Presented is a novel wheel-quadruped mobile robot with parallel driving mechanisms and based on the Stewart six degrees of freedom (6-DOF) platform. The task for traversing rough terrain is decomposed into two prospects: one is the configuration selection in terms of a local foothold cost map, in which the kinematic feasibility of parallel mechanism and terrain features are satisfied in heuristic search planning, and the other one is a whole-body controller to complete smooth and continuous motion transitions. Findings: A fan-shaped foot search region focuses on footholds with a strong possibility of becoming foot placement, simplifying computation complexity. A receding horizon avoids kinematic deadlock during the search process and improves robot adaptation. Research limitations/implications: Both simulation and experimental results validated the proposed scenario available and appropriate for quadruped locomotion to traverse challenging 3-D terrains. Originality/value: This paper analyzes kinematic workspace for a parallel robot with 6-DOF Stewart mechanism on both body and foot. A fan-shaped foot search region enhances computation efficiency. Receding horizon broadens the preview search to decrease the possibility of deadlock minima resulting from terrain variation.