2007 RSS RSS 2007

A Discrete Geometric Optimal Control Framework for Systems with Symmetries

Abstract

This paper studies the optimal motion control of mechanical systems through a discrete geometric approach. At the core of our formulation is a discrete Lagrange-d’Alembert-Pontryagin variational principle, from which are derived discrete equations of motion that serve as constraints in our optimization framework. We apply this discrete mechanical approach to holonomic systems with symmetries and, as a result, geometric structure and motion invariants are preserved. We illustrate our method by computing optimal trajectories for a simple model of an air vehicle flying through a digital terrain elevation map, and point out some of the numerical benefits that ensue. Download: Bibtex: @INPROCEEDINGS{ Kobilarov-RSS-07, AUTHOR = {M. Kobilarov and M. Desbrun and J. Marsden and G. Sukhatme}, TITLE = {A Discrete Geometric Optimal Control Framework for Systems with Symmetries}, BOOKTITLE = {Proceedings of Robotics: Science and Systems}, YEAR = {2007}, ADDRESS = {Atlanta, GA, USA}, MONTH = {June}, DOI = {10.15607/RSS.2007.III.021} }

🌉 Interdisciplinary Bridge — Artificial Intelligence and Reinforcement Learning
📈 Trend Setter — Robotics
🧭 Keyword Pioneer — geometric mechanics
🐝 Cross-Pollinator — Artificial Intelligence, Computer Vision, Deep Learning, Machine Learning, Mathematics & Optimization, Reinforcement Learning, Robotics
🐣 Hot Topic Early Bird — motion planning