Articulated Soft Robots: A Feasibility Driven Control Approach

Soft robots offer safer interaction during tasks, but suitable controllers that leverage their uniqueness are still under development. Differential Dynamic Programming (DDP) is an effective technique for handling highly dynamic tasks with high-dimensional robots. However, the research community has focused on applying DDP to fully articulated robots while also using numerical differentiation. We present an efficient DDP-based algorithm for trajectory optimization of articulated soft robots, fully actuated or not, that jointly optimizes the states, the inputs, and the stiffness profile. We experimentally validate the approach using a fully actuated 2-DoFs compliant arm.