Design of a distributed control architecture for the SAMURAI deep submergence manipulator

The design, development, and testing of a distributed control architecture for a dexterous deep submergence manipulator is described. Many manipulators designed for the undersea environment use a centralized control mechanism, be it a system of valves for a hydraulic manipulator or a central electronics housing for an electrical manipulator. Without the constraints of deep submergence, many modern manipulator systems have adopted distributed architectures to facilitate higher bandwidths and greater modularity for maintenance and reconfiguration. The Subsea Arctic Manipulator for Underwater Retrieval and Autonomous Interventions (SAMURAI) manipulator, developed under NASA and NSF funding, uses a distributed network of Local Processing Units (LPUs) that share the loading of the control system for the manipulator and are co-located in miniature surface-pressure housings at each of the actuator sections. The design of this system is detailed, showing the challenges of such a design, as well as the advantages and limitations of this system as compared to a centralized architecture. In addition, the paper also describes a prototype development architecture for the SAMURAI manipulator that uses commercial off the shelf motion controllers and a commercial data bus in a similar distributed architecture. This system provides reduced but still powerful functionality as compared to the custom architecture, with shorter development times and at significantly lower costs.