Modeling, Optimal Design, Shape Estimation and Workspace Analysis of Soft Robotic Manipulators
Open Access
- Author:
- Trivedi, Deepak
- Graduate Program:
- Mechanical Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- April 21, 2010
- Committee Members:
- Christopher Rahn, Dissertation Advisor/Co-Advisor
Christopher Rahn, Committee Chair/Co-Chair
Eric M Mockensturm, Committee Member
George A Lesieutre, Committee Member
Mary I Frecker, Committee Member - Keywords:
- continuum mechanics
design optimization cosserat rod
soft robotic manipulators
pneumatic air muscles - Abstract:
- Traditional robot manipulators have rigid links and can manipulate objects only using their specialized end effectors. They encounter difficulties operating in unstructured and highly congested environments. Several animal organs such as elephant trunks, mammal and lizard tongues, and octopus arms address this problem by not having any rigid components. These muscular hydrostats are composed of natural muscles and connective tissue. Animal muscle is soft material with large strain, moderate stress, high efficiency, fast response time, high power/weight ratio and long lifetime, capabilities that conventional actuators do not possess. Researchers have been inspired by biology to design, build and test soft robotic manipulators based on electro-active polymers and pneumatic muscles. The unusual compliance and redundant degrees of freedom of these manipulators are essential for applications requiring delicate manipulation in cluttered and/or unstructured environments. With no hard parts, these robots can squeeze through tiny spaces and manipulate objects of widely-varying sizes. A key challenge in the design and control of soft robotic manipulators is the development of accurate models that predict the shape of the arm given the loading and actuation inputs. Existing models make several assumptions about the material properties, loading conditions and kinematics of these manipulators and are not sufficiently accurate under in real world situations. The first contribution of this thesis is the development of geometrically exact models that describe the dynamics of soft robotic manipulators that can be used in design, sensing, stability analysis, and control. The manipulator is modeled using Cosserat rod theory and takes into account the effect of finite shear, curvature and extension, and material nonlinearities. The model is validated on the OctArm V manipulator. Parametric studies are done with the model to gain an understanding of the mechanics of soft robotic manipulators, providing insight into the optimal design of pneumatic and hydraulic soft manipulators. Theoretically, soft robots have infinite degrees of freedom (dof), but the number of sensors and actuators are limited. This complicates shape estimation and control of soft robotic manipulators. The second contribution of this thesis is the development of three novel methods of shape estimation for soft robotic manipulators based on the geometrically exact model. The first method uses load cells mounted at the base of the manipulator and the second method makes use of cable encoders running through the length of the manipulator. The third method uses inclinometers mounted at the end of each section of the manipulator. Using simulation and experiments these methods are compared for the accuracy of endpoint position estimation for unloaded and loaded OctArm VI. OctArm-type soft robotic manipulators are complex and difficult to design and fabricate. The third major contribution of this thesis is a simpler, cost effective design for a pneumatic air muscle based soft robotic manipulator in which the actuators for the distal section extend from the base to the tip of the arm, thereby simplifying the pneumatic design and eliminating the need for complex manifolds. We compare the workspace and dexterity of this new continuous tube design with the OctArm manipulator and conclude that although the two designs have comparable workspace area, the OctArm workspace has better dexterity characteristics.