Assessing the Value of Future and Present Options in Real-Time Planning

摘要: Highly dynamic environments with uncertainty make inadequate long or rigid plans, because they are frequently dismissed by the arrival new unexpected situations. In these environments, most approaches eliminate planning altogether, and evaluate just current situation. We interested in on-line planning, where execution interleaved, short plans continuously re-evaluated. Now, plan evaluation itself could be an important issue. have proposed our recent work to taking into account quantity quality of future options, not single best option. this paper we present a detailed real-time performance, changing importance given situation, option, set options respective evaluations, context simulated soccer Robocup competition. Our results show that well-tuned combination mentioned factors outperform any them alone.