Abstract: In fully mechanized working face in thin coal seams, fixed rail-type inspection robots are difficult to adapt to curvature of working face track and floor undulations, cable-suspended inspection robots are difficult to adapt to environments with large roof undulations, while ground-moving inspection robots have low obstacle-crossing speed and insufficient flexibility. To address these problems, a quadruped inspection robot walking mechanism suitable for the complex environment of a fully mechanized working face in thin coal seam was proposed. The walking mechanism was driven by an explosion-proof servo motor and enabled switching between high- and low-leg-lifting modes through a linkage-leg mechanism and a gear shifting mechanism, thereby balancing walking efficiency and obstacle-crossing capability. Taking the installation position of the crankshaft as the design variable and the leg lifting height of the linkage-leg mechanism and the driving torque as optimization objectives, a multi-objective optimization model was established and solved using the NSGA-II algorithm, and the Pareto optimal solution set under both high- and low-leg-lifting modes were obtained, which significantly improved the obstacle-crossing capability of the walking mechanism while considering driving requirements and force transmission performance. A robot simulation scenario for a fully mechanized working face in thin coal seam was established in ADAMS and dynamic simulations under four operating conditions, including flat-ground walking, slope walking, flat-ground obstacle crossing, and slope obstacle crossing, were carried out. The results showed that the maximum walking speed of the robot reached 15.8 m/min, the robot walked stably on a 10° slope, and it crossed obstacles with a height of 100 mm. Under the four operating conditions, the robot posture remained controllable without instability, meeting the requirements for motion performance and stability for follow-up inspection in fully mechanized working faces in thin coal seam.