Combining direct and model-based perceptual information through schema theory

Coupling sensor information directly to action has been a successful design approach for robot systems that operate in a real-world environment. A problem with using sensor information directly is that what is “perceived” sometimes does not capture the correct aspects of the operating environment for decision making. We term this type of problem the false-positive problem, and present an approach that addresses this phenomenon: combine model-based perceptual information with direct sensory measurements. A robot control system was implemented and experiments were conducted to show how this approach overcomes some examples of false-positive problems