Legged robots remain impractical in non-inertial environments, where the ground and surrounding objects accelerate relative to the inertial Earth frame. According to the paper, the main obstacle is the presence of unknown, time-varying, persistent perturbations that current control methods cannot handle well.

The proposed approach is an adaptive rejection scheme designed for unknown hybrid sinusoidal disturbances. This allows the controller to estimate and cancel disturbances that change over time, rather than relying on fixed models. The work specifically targets humanoid walking under unknown ground motion, a scenario that would be common in real-world applications such as moving platforms or earthquake-like conditions.

Because only one source was used, there are no conflicting findings to compare. The abstract does not provide experimental results, so the claims are limited to the proposed method and its intended application.