R. Peter Bonasso
This paper describes a situated reasoning architecture, originally used with ground mobile robots, which is shown to easily integrate control theoretic algorithms, navigation heuristics and human supervision for semi-autonomous robot control in underwater field environments. The control architecture produces reaction plans that exploit low-level competences as operators. The low-level competences include both obstacle avoidance heuristics and control-theoretic algorithms for generating and following a velocity/acceleration trajectory. Experiments with an undersea remotely-piloted robot in a test tank at the Deep Submergence Laboratory at Woods Hole, MA are described. The robot performed both pilot-aided and autonomous exploration tasks robustly during normal changes in the task environment. The architecture was implemented in the GAPPS/REX situated automata programming language. The guaranteed constant cycle time of the synchronous REX circuits allowed for rapid tuning of the parameters of the control-theoretic and heuristic algorithms to obtain smooth, safe motion.