Agents based on the subsumption architecture are specifically designed to function effectively in real-world environments. In particular, the goal of the subsumption architecture has been to develop robots that can navigate in a complex and dynamic environment such as cluttered office floors with people moving about. As such, the approach has been not to simplify the environment, but rather to simplify the architecture . Therefore, simplifying assumptions such as perfect perception and modified time scales that are made in simulated environments cannot be made in a real-world environment. In response to these difficulties, work on subsumption agents has focused on perception and real-time execution .
Go to a discussion of this environment for multiple architectures.