Model-Based Control of Handed Shearing Auxetics (HSA) Robots

While serial continuum soft robots have been intensively investigated in recent years, parallel soft robots are less studied despite exhibiting exciting properties such as an improved stiffness-to-weight ratio. One recent development in this field are HSA robots, which consist of multiple HSA rods that are connected at their distal end through a rigid platform. Twisting of the proximal end of an HSA causes the rod to elongate and enables complex motion primitives in 3D space. Recent work has investigated the mechanical characterization, simulation, and kinematic modeling of HSA robots, but control has yet to be tackled. In this work, we make a first step towards achieving task-space control by designing model-based regulators for planar motions—our approach considers essential characteristics of HSA robots, such as underactuation, shear strains, and varying stiffness. In prior work, we derived a dynamic model for planar HSA in Euler-Lagrange form and experimentally verified it. We notice that the resulting planar dynamics are underactuated and that the actuation forces are non-affine with respect to the control inputs, which are the motor angles. The latter is a peculiarity of these systems, rarely observed in other robots. Based on the model knowledge, we propose in this chapter two control strategies for planar HSA robots capable of regulating the end-effector towards a desired position in task-space. The first strategy performs steady-state planning to identify an admittable configuration and steady-state control input matching the desired end-effector position and then subsequently applies a P-satI-D feedback controller on the collocated form of the system dynamics. The second strategy directly regulates the end-effector position using a Cartesian impedance controller. This allows us to unite the soft robot's embodied intelligence with computational intelligence to guarantee compliance and interaction safety. In summary, we state our contributions as (i) a provably stable model-based control strategy for guiding the end-effector of the robot towards a desired position in Cartesian space with a configuration-space controller that combines an integral-saturated PID with a potential shaping feedforward term, (ii) a Cartesian impedance controller that allows combining the passive compliance of the HSA robot with active compliance in the control strategy, and (iii) extensive experimental verification of both control strategies.