Stochastic Model-Based Control of Multi-Robot Systems
Abstract
In this report we consider control of single- and multi-robot systems as an optimal control problem. Solution of this problem may be of enormous complexity because of a large-number of robots, a large number of redundant states, and environment uncertainties. Motivated by estimation methods based on statistical sampling employed for solving complex estimation problems, we explore the possibility of using stochastic process samples for computing the optimal control. This approach can ultimately provide small-size, low-cost and efficient computational hardware for solving complex multi-robot control problems and in which computations are driven by laws of statistical physics.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 30, 2009
- Accession Number
- ADA520667
Entities
People
- Dejan Milutinović
- Devendra P. Garg
Organizations
- Duke University