Fault tolerance for kinematically redundant robotic manipulators

Christopher L Lewis, Purdue University


Robotic manipulators have been identified as a major technology to be utilized in the cleanup and management of hazardous waste and for deep sea and space exploration. The corrosive nature of such environments reduces the expected life of sensors and actuators. The cost of repairing systems in such environments is often prohibitively large especially when radioactive elements are present. This work is concerned with developing methods by which kinematically redundant manipulators can continue to operate after joint failures. Specifically, failures are considered that result in a joint being locked in the position in which it fails. The basic premise of using kinematically redundant manipulators is that the extra joints will provide the necessary degrees-of-freedom to complete the task should one of the joints fail. In this work, the redundancy is not only used for continued operation after a failure but also to avoid configurations prior to a failure in which the manipulator would not be usable if a joint failed. Two distinct methods are proposed. First, local fault tolerance measures are defined for kinematically redundant manipulators. Using these measures, optimally fault tolerant configurations are identified. Then, control strategies are designed to maintain high values of these measures. In the second method, more global conditions are considered in order to identify regions of the manipulator's workspace which can be guaranteed to be reachable by the manipulator after an arbitrary joint failure.




Maciejewski, Purdue University.

Subject Area

Electrical engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server