A Mobile Robot with Modular and Reconfigurable Mobility and Manipulation Open Access
Downloadable ContentDownload PDF
This dissertation explores modular and reconfigurable mobility and manipulation in field robotic applications. The motivations of this work stem from the need to develop mobile robotic systems capable of adapting their shape to changes in the terrain layout and to the nature of the assigned task. In recent years, significant progress has been accomplished in the field of adaptable modular robotics. However, two main limitations primarily dominate the state-of-the-art. The first is the docking strength, which represents the key enabling component of re-configurability in field applications. The second is the existence of a manipulator arm capable of accomplishing meaningful tasks.In this dissertation, a new platform for robotic mobility and manipulation called STORM (for Self-configurable and Transformable Omni-directional Robotic Modules) is proposed, and is characterized by three main attributes. The first is a tri-state active reversible and non-back-drivable docking interface which provides coupling rigidity, and employs a dual-rod slider rocker mechanism (DRSR) to deliver three independent modes of operation. The second is a locomotion module with hybrid wheeled-tracked multidirectional mobility which enables alignment between the mating elements prior to docking, and the third is a manipulation module carrying an arm and an end-effector.This dissertation presents a synthesis of STORM's architecture including the kinematic and dynamic properties of the proposed docking interface and the DRSR. This synthesis is supplemented by a force analysis that validates the structural strength of the interface, and further corroborates the results with measurements on a proof-of-concept prototype. Subsequently, a discussion on the locomotion module and its unique dynamic properties is presented, along with an algorithm for body roll and drive handling active control to stabilize the module during wheeled mobility. A second learning-basedalgorithm, called COIN (for Circles Of Initialization), is proposed for the motion control of the manipulation module, with broader extrapolations to the motion generation of redundant arms with n-joints. The dissertation concludes with an insight into dynamicmotion synchronization for a sample three-module formation of STORM, and with a forward looking statement into prospective investigations.