Robot motion synthesis using ground reaction forces pattern

Abstract

Gait generation in its realization stage in majority of bipeds involves inverse dynamics model with the need of pre-determined joint trajectories and the zero moment point method for final postural stability adjustments. This requires generation of such joint trajectories, where the zero moment point criterion is fulfilled and the human-like posture is kept. Such task is complex and still misses the universal solution. Preserving natural (human-like) and stable body posture under acting disturbances is an issue too. High computational costs, problem with disturbances rejection and controller complexity are the major disadvantages associated with such approaches. This article presents a method for motion control with forward dynamics model and ground reaction forces as the reference values. The ground reaction forces measured during real human walking are utilized here. A MATLAB/Simulink framework is used for testing the proposed synthesis of bipedal locomotion. The ground reaction forces recorded during the walk with different footwear are applied as the references. The obtained motion patterns and postures are compared with those of the human. Obtained results were satisfactory and justified the proposed approach.