Efficiency of Searching Knowledge Bases for Robotic Control

Autonomous intelligent robotic systems that can survive in dynamic changing environments have yet to be realized. The robot must process incoming information, make decisions, and learn from previous actions concurrently and in real time. Reliance on one all-encompassing system with one overburdened processor, no matter how fast, makes this goal unreachable. One of the primary reasons the goal is unreachable is that the knowledge base that the robot must access to make its decisions grows so large over time that simply searching the knowledge base for a decision becomes too time consuming for acceptable robotic behavior. One solution is to distribute the knowledge base among several processors and search in parallel. An experiment was contrived to test parallel searching algorithms which must search large databases of information, and provide some insight on performance of the underlying programming environment.