| "Vision, Instruction, and Action", David Chapman, MIT A.I. Lab Technical Report 1204, Aug. (Chatila and Laumond 85] "Position Referencing and Consistent World Modeling for Mobile Robots", R. Chatila and J. Laumond, IEEE International Conference on Robotics and Automation, March, 138-145. |
....not build world models We answer this question affirmatively below, but first we need to understand the following. It turns out that there are more problems with world models than just uncertainty. Creature researchers who work in simulated worlds (e.g. Pollack and Ringuette 90] is an example; [Chapman 90] is an exception) are usually not even aware of the problem of uncertainty. But worse, they are not aware of the current technological impossibility of using sensors and perception systems, to build models of the world which reflect an objective reality. Such an objective reality is 1 Admittedly, ....
....readings, but they have the property that they are both positively and negatively repeatable with respect to any particular place in the world over a wide variance in the robot s actual position and trajectory through that place. Such representations of the world we called deictic by [Agre 88] and [Chapman 90]. 3.1.3 Analyzing Behaviors The discussion above can lead us to a rather simpler way of analyzing the behavior of a creature than if we adopted a stance of having objective world models. In particular this method will let us deal with uncertainty entirely at analysis time (analogously to dealing ....
[Article contains additional citation context not shown here]
"Vision, Instruction, and Action", David Chapman, MIT A.I. Lab Technical Report 1204, Aug. (Chatila and Laumond 85] "Position Referencing and Consistent World Modeling for Mobile Robots", R. Chatila and J. Laumond, IEEE International Conference on Robotics and Automation, March, 138-145.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC