Qualitative Reasoning about Perception and Belief.
Álvaro del Val, Pedrito Maynard-Reid, Yoav Shoham · 1997
We present a qualitative model for reasoning about perceptions, sensors, and belief, and a logic to reason about this model. Basic to our model is a distinction between precision and accuracy, for both of which we provide qualitative definitions. In our logic this distinction gives rise to two modal operators---P for actual perception, and C p for perceptual capability, which is captured as a set of possible percepts. Adding to these operators the standard B operator to model belief, we end up with a logic combining standard Kripke-style semantics with the almost-standard `neighborhood semantics.' We define various agent types in the logic, from agents who believe all and only what the sensors tell them, to much more skeptical agents. We define each agent both axiomatically and model-theoretically, and provide soundness and completeness results relating the two types of definitions. 1 Introduction A great deal of attention in AI has been devoted to modeling states of information, whe...