Abstract
A closed-chain legged robot with a reconfigurable trunk is designed, and its continuous and collision-free obstacle-crossing strategy is presented to enhance terrain adaptability and intelligence, ultimately enabling autonomous obstacle crossing. The design integrates four coupled Watt-I type linkages that form a quadruped unit. Motors link four quadruped units to the robot's trunk, enabling the overall reconfigurability of the unit. The kinematic model is analyzed, and the dimensions of the leg components are optimized. Finally, the foot trajectory analysis is performed. Obstacles are categorized according to their scale based on the analysis of foot-endpoint trajectories. For small-scale obstacles, gait is planned through preplanned posture mapping relationships, while a stride length planning algorithm is designed to plan the obstacle-crossing gait for large-scale obstacles. The collision-free obstacle-crossing strategy is achieved by controlling the heights of the maximum and minimum effective crossing points, the feasibility of which is confirmed via simulations and experimental studies.
