This paper described a novel hybrid framework that integrates ASP and POMDPs to enable a mobile robot to represent and reason with domain knowledge, automatically adapt sensing and information processing to the task at hand, merge logical facts with probabilistic beliefs, and use the information extracted from sensory cues and high-level human inputs to revise the knowledge base. The framework is evaluated in the context of visual target localization-results show that the robot is able to operate reliably and efficiently.
展开▼