Parametric study for continuous quasi-passive walking of a musculoskeletal humanoid robot with anatomy trains

Bipedal walking is not merely a form of locomotion but a complex movement composed of various motor tasks, such as upper body posture control, weight support, leg lifting, and leg swinging. Anatomy trains (ATs) refer to the myofascial connections spanning throughout the body, which dramatically switch the dynamics of the whole-body joints. The ability to appropriately control the dynamics of the whole-body joints using ATs is considered to be a crucial factor in achieving the motor tasks required for bipedal walking. However, the design method of the design parameters, such as the magnitude of forces exerted by ATs and their contraction and relaxation timing, have not yet been fully elucidated. Hence, it is essential to theoretically clarify the design method of the human musculoskeletal system equipped with ATs. In previous studies, to establish a foundation for such theoretical analysis, we designed a musculoskeletal humanoid robot equipped with the SBLs and its mathematical model, which can achieve quasi-passive dynamic walking on a slope. We found a condition where more than 40 steps of continuous walking is achieved through trial and error with a feedback controller based on the upper body posture. In this study, we selected two representative design parameters and numerically investigated the relationship between their values and the step count.