Foot placement control underlies stable locomotion across species.

Animals navigate their environment stably without inefficient course corrections despite unavoidable errors. In humans, this stability is achieved by modulating the placement of the foot on each step such that recent errors are corrected. However, it is unknown whether animals with diverse nervous systems and body mechanics use such foot placement control; foot trajectories of many-legged animals are considered to be stereotypical velocity-driven patterns, as opposed to error-driven. Here, we put forth a unified "feedforward-feedback" control structure for stable locomotion that combines velocity-driven and body state error-driven foot placement. We provide empirical support for this control structure across flies, mice, and humans by mining their natural locomotor variability, finding that a competing control structure with purely velocity-driven foot placement is not supported by the data. This work finds shared behavioral signatures of foot placement control in flies, mice, and humans. We find that key characteristics of these signatures, such as their urgency and centralization, vary with neuromechanical embodiment across species. For example, more inherently stable multilegged animals exhibit less urgent control with a lower control magnitude and a slower correction timescale compared to humans. Furthermore, many-legged animals display modular, direction-, and leg-specific control signatures, whereas humans exhibit common signatures across both legs. Overall, our findings provide insight into stable locomotion across species, revealing how species with diverse neuromechanics achieve a shared functional goal: foot placement control.