Analysis and design of optimally fault tolerant robots

Robot manipulators can be used to navigate and perform tasks in unstructured and hazardous environments where human safety is a primary concern. For example, they are used for nuclear waste disposal, space exploration, nuclear power industry, military surveillance, etc. A number of such robot manipulators are being used but the concern is that these robots should be able to complete their critical tasks in the event of failures that they encounter working in such environments. One of the most common failures of field robots is an actuator failure. This type of failure affects the joints of the robots inducing failures like locked-joint failures and free-swinging joint failures. To design a fault tolerant system the robot has to rely on the incorporation of redundancy into its system. This redundancy takes several forms: sensor redundancy, analytical redundancy, and kinematic redundancy. This work focuses on using kinematic redundancy to deal with the issue of multiple locked-joint failures in the robotic systems.;The goal of this work was to analyze and design fault-tolerant manipulators. The robots designed are able to finish their required task in spite of a failure in one or more of its joints. In order to design optimally fault tolerant manipulators, it is necessary to quantify fault tolerance. The approach taken here was to define fault tolerance in terms of a suitable objective function based on the robot's manipulator Jacobian. In the case of the relative manipulability index, local fault tolerance is characterized by the null space of the manipulator Jacobian. Since the null space can be used to identify locally fault tolerant manipulator configurations, one goal of this work was to develop procedures for designing fault tolerant manipulators based on obtaining a suitable null space for the manipulator Jacobian.;In this work, optimally fault tolerant serial manipulators are designed that are fault tolerant to two locked-joint failures simultaneously. Furthermore, the symmetry of the manipulators is studied using positional and orientational Jacobians; and examples are presented for condition number and dynamic manipulability index to study the behavior of different fault tolerance measures. Lastly, a methodology for designing an optimally fault tolerant 4-DOF spherical wrist type mechanism was presented. It was shown that the orientational Jacobian must have a certain form for the manipulator to have the best possible relative manipulability index value. An optimal configuration along with the corresponding DH parameters was presented. Furthermore, it was pointed out that isotropic configurations of a 4-DOF spherical wrist type mechanism are fault tolerant and optimal in the sense that they have the largest possible manipulability index prior to a failure. An example of an orientational Jacobian was presented for a 6-DOF spherical wrist that is equally fault tolerant for any two joint failures.