Demonstration Abstract: New Developments in Real-Time Heuristic Search

摘要: Our demonstration consists of a poster, videos and interactive simulations of real-time heuristic search algorithms for goal-directed navigation on a priori completely or partially unknown grids. It provides a brief introduction to real-time heuristic search by describing LSS-LRTA* and RTAA*. It then illustrates a performance issue of LSSLRTA* and RTAA* due to depressions in the h-value surface. It describes the real-time heuristic search algorithms aLSS-LRTA*, daLSS-LRTA*, aRTAA*, and daRTAA*—which address this issue—and summarizes their properties. Our demonstration also illustrates a performance issue of LSSLRTA* and RTAA* due to performing repeated A* searches around the current cells of the agent. It describes RTBA* and TBAA*, two real-time heuristic search algorithms that address this issue, and summarizes their properties.