The difficulty associated with placing intelligent agents in environments is in part one of designing the means by which they sense their world. If the world is complex, or if the information available in the world is in an inappropriate form, then a mismatch between the information in the environment and the information required by the agent can exist. Practical means of dealing with this mismatch are available but designing the system requires careful modelling. By modelling perception as the module within the agent responsible for sensing and making sense of the environment several advantages are realised. The explicit representation of perception allows consideration of the issues, affords software engineering advantages with respect to the specification and design of systems and is compatible with many accepted definitions of agency.
Lecture notes in computer science: advances in artificial intelligence: proceedings of the Pacific Rim International Conference on Artificial Intelligence (PRICAI 2000 Workshop Reader), Melbourne, Victoria, Australia, 28 August - 01 September 2000 / G. Goos, J. Hartmanis, and J. van Leeuwen (eds.),
Vol. 2112, pp. 239-248