A Review of Exploration of the Stable Operation Principle of Quadruped Robots in Complex Terrains

Abstract

 This review examines the principles enabling quadruped robots to operate stably in complex terrains. The review focuses on three key areas: mechanical design, control algorithms, and environmental perception. Inspired by quadrupedal animals, bionic design strategies such as implementing rigid and flexible joints and bio-inspired leg structures have been used in the mechanical design and have significantly improved terrain adaptability. While Model Predictive Control (MPC) handles constraints, Reinforcement Learning (RL) enables adaptive gait planning. LIDAR, visual sensors, and sensor fusion improve terrain recognition so that quadruped robots can recognize and quickly respond to changes in terrain or the environment. To this day, quadruped robots demonstrate practical value in industrial routing inspection, disaster rescue, complex-terrain exploration, and military fields. However, challenges remain in payload capacity, algorithm real-time performance, and sensor robustness, indicating that future research should emphasize integrated and interdisciplinary innovation.