A Common-Sense Planning Strategy for Ambient Intelligence

Systems for Ambient Intelligence contexts are expected to exhibit an autonomous and intelligent behavior, by understanding and reacting to the activities that take place in such contexts. These activities, specially those labeled as trivial or simple tasks, are carried out in an effortless manner by most people. In contrast to what it might be expected, computers have struggled to deal with these activities, while easily performing some others, such as high profile calculations, that are hard for humans. Imagine a situation where, while holding an object, the holder walks to a contiguous room. We effortlessly infer that the object is changing its location along with its holder. However, such inferences are not well addressed by computers due to their lack of common-sense knowledge and reasoning capability. Providing systems with these capabilities implies collecting a great deal of knowledge about everyday life and implementing inference mechanisms to derive new information from it. The work proposed here advocates a common-sense approach as a solution to the shortage of current systems for Ambient Intelligence.