<i>Camponotus japonicus</i> locomotion on vertically undulating mesoscale rough terrain

Mesoscale rough terrain constitutes the primary interactive environment for animal locomotion, yet how roughness variations at this scale influence animal movement remains poorly understood. This study investigates the locomotion of Camponotus japonicus on finely segmented, vertically undulating mesoscale rough terrain through integrated kinematic analysis and gait space clustering. We designed six terrains with different roughness levels, recorded ant locomotion using high-speed cameras, and then applied markerless pose estimation to extract time-series data of limb keypoints. Our results show that with increasing terrain roughness, ants exhibit reduced walking speed via shorter stride length and longer stride period. Footfall analysis reveals a pattern of lateral foot placement away from the body midline. Both phase analysis and gait space clustering indicate a gradual transition from the stereotyped tripod gait toward more variable tetrapod-like gaits. Speed-matched analysis indicates that, in addition to walking speed, terrain roughness also influences leg coordination. Together, these findings indicate that mesoscale rough terrain is associated with a reorganization of ants' spatiotemporal gait patterns. The discovery of these general, mesoscale environmental locomotive adjustment patterns provides crucial bio-inspired insights for multi-legged bionic robots to overcome locomotive adaptability bottlenecks in complex, unstructured environments.