Analyzing the Effect of Soft Arm Design on Obstacle Navigation through Collision

Abstract

The study of soft robot arms is often motivated by safe operation in contact or collision, such as an arm reaching into a tube, squeezing under a barrier or wrapping around an object. Existing work focuses on soft grasping, while collision-allowed navigation is neglected. Soft arms are not guaranteed to successfully push past obstacles simply because the arms are soft, but design rules for arms in collision have not been developed. This paper presents an initial empirical examination of planar arms in collision and specifically studies the relationship between key arm design variables - length, number of segments, taper - and the arm's ability to push past a semi-circular obstacle. Fourteen variants of a planar arm design were built and tested to determine the pressure required to push past the obstacle at a set of known locations. The pressure required was measured for actuation of one and multiple segments. The results were used to develop an empirical collision success model, which can be used with collision-allowed planning algorithms. Soft arms that balance flexibility, or range of motion, and applied force successfully push past obstacles in the largest range of locations. Flexibility and force are opposing traits driven by arm width; wider arm segments produce more force but have a smaller angular stroke and are harder to passively bend. This work serves as a foundation for future studies with a broader set of obstacle types and locations, which could be used to develop a robust collision model.