Locomotion Design and Experimental Validation of a 10-Ton-Class Firefighting Hexapod Robot

Legged locomotion at the multi-ton scale remains largely unexplored due to challenges in actuation, structural robustness, stability, and power consumption. This paper presents the design and experimental walking implementation of a 10-ton-class firefighting hexapod robot on physical hardware. A hydraulically actuated hexapod with parallelogram leg mechanisms is developed, and a statically stable walking gait is designed based on kinematic constraints and phase coordination. Reinforcement learning is employed in a lightweight manner to modulate high-level gait parameters without introducing complex dynamic control. Simulation and physical experiments demonstrate stable and repeatable flat-ground walking, providing evidence that legged locomotion at the multi-ton scale is feasible.