Nonparametric Modeling and Control of High Performance Maneuvers
Abstract
We developed several nonparametric modeling techniques as well as nonlinear controller design techniques. We successfully implementing these techniques on a robot arm. We used a locally weighted modeling approach as the basis of our nonparametric modeling technique. Locally weighted modeling avoids negative interference by retaining the original training data, so the approach is adaptive to changing data distributions. We used sophisticated representations of high dimensional sub-manifolds to enable dynamic programming in higher dimensional spaces than are currently possible.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 20, 1997
- Accession Number
- ADA329721
Entities
People
- Christopher G. Atkeson
Organizations
- Georgia Tech