Combining decision theory and hierarchical planning for a time-dependent robotic application

The authors present the Rational Reasoning System (RRS), which integrates hierarchical planning with decision theory in order to make rational decisions in time-dependent situations. RRS controls a robot in a simulated hazardous environment and allows it to address time-dependent emergencies of various urgencies. RRS uses decision theory to constrain and focus the hierarchical construction of plans, introducing computationally practical meta-reasoning methods that factor costs of planning into an overall optimal reasoning and acting behavior. The authors illustrate RRS's capabilities in a simulated nuclear power plant. The authors propose that decision theory should be a basis for any truly autonomous rational robotic system.<
>