Models and analysis of locomotion and gripping in soft robots.

摘要

Recent designs of soft robots and nano robots feature locomotion mechanisms that entail orchestrating changes to intrinsic curvature or length to enable the robot's limbs to either stick, adhere, or slip on the robot's workspace. The resulting locomotion mechanism has several features in common with peristaltic locomotion that can be found in the animal world. One of the purposes of this dissertation is to examine the feasibility of, and design guidelines for, a locomotion mechanism that exploits the control of intrinsic curvature on a rough surface featuring stick, slip, and adhesion interaction. Our work complements the ever-increasing body of work on soft robots that is primarily focused on the design and fabrication of these systems. Modeling and analyzing these robots is challenging because of the difficulties in faithfully modeling the flexible nature of their components.;The study of locomotion presented in this dissertation is composed of two parts. First, we consider the simplest possible model for a soft robot. The resulting model is a lumped parameter system featuring a pair of mass particles and a spring with a variable natural length. By appropriately varying the natural length as a function of time ℓ0(t), we show how locomotion can be achieved and examine the energy efficiency for a variety of choices of ℓ0(t). We then take the lessons gained from this model and use them to understand the locomotion of a block that is propelled on a rough surface with the aid of a flexible arm. Our analysis of the rod-based model for this system focuses on the development of a structurally stable mechanism to move the block. This analysis exploits recent results on stability of adhered rods that we supplement with a new discretized stability criterion.;Beyond locomotion, soft robots have the ability to gently grip and maneuver objects with open-loop kinematic control. Guided by several recent designs and implementations of soft robot hands, we exploit our earlier works on locomotion and analyze a rod-based model for the fingers in the hand of a soft robot. We show precisely how gripping is achieved and how the performance can be affected by varying the system's parameters. The designs of interest feature pneumatic control of the soft robot and we model this actuation as a varying intrinsic curvature profile of the rod. Our work provides a framework for the theoretical analysis of the soft robot and the resulting analysis can also be used to develop some design guidelines.