| "Adaptive Gaze Control", Paul A. Viola, MIT SM Thesis, 1990. |
....is the cornerstone of modern neural network research, inconvenient for use in embodied or situated systems. In fact, most of the examples in the new wave of neural networks have not been situated or embodied. There are a few counterexamples (e.g. Sejnowksi and Rosenberg 87] Atkeson 89] and [Viola 90] but in the main they are not based on back propagation. The most successful recent learning techniques for situated, embodied, mobile robots, have not been based on parallel algorithms at all rather they use a reinforcement learning algorithm such as Q learning ( Watkins 89] as for example, ....
....three of these types of learning. The last one has not yet been successfully tackled 32 . Learning representations of the world was already discussed above concerning the work of [Mataric 90, 91] The next step will be to generalize active constructive representations to more classes of use. Viola 90] demonstrated calibration of a complex head eye system modeling the primate vestibulo ocular system. In this system there is one fast channel between a gyroscope and a high performance pan tilt head holding the camera, and a slower channel using vision which produces correction signals for the ....
"Adaptive Gaze Control", Paul A. Viola, MIT SM Thesis, 1990.
No context found.
"Adaptive Gaze Control", Paul A. Viola, MIT SM Thesis, 1990.
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