Development and multiple mode control of modular and reconfigurable robot

There is a strong desire for robots to manipulate in uncontrolled environments. In uncontrolled environments, the robot has to adapt to the world consisting of only partially known or unknown objects and tasks, and real-time constraints. The capability of robots working in active or passive modes and switching between them helps enabling the robots to work in unstructured environments. Joint torque sensing is essential for implementing multiple mode control of robots. Though there have been a number of means of joint torque sensing, the existing joint sensing techniques have diverse limitations, such as in installation, reliability, cost, and noise immunity. This dissertation work develops a new joint torque sensing method for a modular and reconfigurable robot (MRR) with harmonic drive joints and provides solutions to multiple mode control of MRR based on the proposed sensing technique. This research consists of two main parts. In the first part, a novel mathematical model for compliance of harmonic drives has been proposed. The proposed model captures not only the nonlinear stiffness but also the hysteresis phenomenon of harmonic drive transmission. Based on the developed harmonic drive compliance model, a joint torque estimation method using position measurements is developed. Torque estimation using position measurements provides an advantage of noise immunity to the estimated joint torque. Using the compliance of harmonic drives instead of an additional elastic component does not change the joint dynamics. Building upon the new torque estimation technique, a multiple working mode control algorithm for MRR is developed and experimentally validated. The objective of the second part is to make the wrist suitable for dexterous manipulation in unstructured environments, such as door opening. A robust adaptive controller is developed for tracking control of the wrist in active mode; and a new interactive force compensation technique is proposed based on force sensor measurement, enabling passive working mode of the compact wrist without using mechanical solutions, which not only saves weight and volume, but also avoids losing tracking of the joints’ position when switching from passive mode to active mode. Experiments on a prototype wrist have demonstrated the effectiveness of the proposed method.

Development and multiple mode control of modular and reconfigurable robot

There is a strong desire for robots to manipulate in uncontrolled environments. In uncontrolled environments, the robot has to adapt to the world consisting of only partially known or unknown objects and tasks, and real-time constraints. The capability of robots working in active or passive modes and switching between them helps enabling the robots to work in unstructured environments. Joint torque sensing is essential for implementing multiple mode control of robots. Though there have been a number of means of joint torque sensing, the existing joint sensing techniques have diverse limitations, such as in installation, reliability, cost, and noise immunity. This dissertation work develops a new joint torque sensing method for a modular and reconfigurable robot (MRR) with harmonic drive joints and provides solutions to multiple mode control of MRR based on the proposed sensing technique. This research consists of two main parts. In the first part, a novel mathematical model for compliance of harmonic drives has been proposed. The proposed model captures not only the nonlinear stiffness but also the hysteresis phenomenon of harmonic drive transmission. Based on the developed harmonic drive compliance model, a joint torque estimation method using position measurements is developed. Torque estimation using position measurements provides an advantage of noise immunity to the estimated joint torque. Using the compliance of harmonic drives instead of an additional elastic component does not change the joint dynamics. Building upon the new torque estimation technique, a multiple working mode control algorithm for MRR is developed and experimentally validated. The objective of the second part is to make the wrist suitable for dexterous manipulation in unstructured environments, such as door opening. A robust adaptive controller is developed for tracking control of the wrist in active mode; and a new interactive force compensation technique is proposed based on force sensor measurement, enabling passive working mode of the compact wrist without using mechanical solutions, which not only saves weight and volume, but also avoids losing tracking of the joints’ position when switching from passive mode to active mode. Experiments on a prototype wrist have demonstrated the effectiveness of the proposed method.

Muscle-fiber array inspired, multiple-mode, pneumatic artificial muscles through planar design and one-step rolling fabrication

Although the advances in artificial muscles enable creating soft robots with biological dexterity and self-adaption in unstructured environments, producing scalable artificial muscles with multiple-mode actuations is still elusive. Inspired by muscle-fiber arrays in muscular hydrostats, we present a class of versatile artificial muscles, called MAIPAMs (Muscle-fiber Array Inspired Pneumatic Artificial Muscles), capable of multiple-mode actuations (such as parallel elongation-bending-spiraling actuations, parallel 10 bending actuations, and cascaded elongation-bending-spiraling actuations). Our MAIPAMs mainly consist of active 3D elastomer-balloon arrays reinforced by a passive elastomer membrane, which is achieved through a planar design and one-step rolling fabrication approach. We introduce the prototypical designs of MAIPAMs and demonstrate their muscle-mimic structures and versatility, as well as their scalable ability to integrate flexible while un-stretchable layers for contraction and twisting actuations and compliant electrodes for self-sensing. We further demonstrate that this class of artificial muscles shows promising potentials for versatile robotic applications, such as carrying a camera for recording videos, gripping and manipulating objects, and climbing a pipe-line.