A mission planning expert system for an autonomous underwater vehicle

A progress report on the use of a Lisp machine and the KEE expert system shell to realize a mission planning expert system for a small autonomous underwater vehicle (AUV) is presented. The methodology adopted provides for incremental development through the modularity facilitated by object-oriented and rule-based programming as provided by KEE. In addition, the use of Lisp as the language for the AUV onboard mission control computer greatly simplifies code development, since recoding of simulation software is largely avoided. A key feature of this expert system is its ability to choose an appropriate search strategy for a particular mission, both for premission path planning and for real-time mission execution. In particular, for missions in which time is not critical, optimal methods such as A* search can be performed using all available information about the vehicle and its environment. On the other hand, when critical completion deadlines are required for either offboard or onboard path planning and obstacle avoidance, an appropriate suboptimal method capable of giving all acceptance answer within the mission time constraints is adopted. The performance of the mission planning expert system has been evaluated using a detailed graphical simulation of the AUV including a full nonlinear dynamic model