A new solution method for nonlinear stochastic optimal control based on stochastic analysis and statistical mechanics and its application to learning control

2014 Fiscal Year Final Research Report

• Project/Area Number: 24760338
• Research Category: Grant-in-Aid for Young Scientists (B)
• Allocation Type: Multi-year Fund
• Research Field: Control engineering
• Research Institution: Hiroshima University
• Principal Investigator: SATOH Satoshi 広島大学, 工学(系)研究科(研究院), 助教 (60533643)
• Project Period (FY): 2012-04-01 – 2015-03-31
• Keywords: 非線形制御 / 確率制御 / 最適制御

Outline of Final Research Achievements

We have developed a new iterative solution method for nonlinear stochastic optimal control problems based on stochastic analysis and statistical mechanics. We have also clarified conditions for convergence of a sequence of iterative solutions. This method enables us to obtain an optimal feedback control input for a nonlinear stochastic optimal control problem for which a sufficient solution has not been provided so far.
Then, we have proposed a stochastic approximate solution to a deterministic optimal control problem. In this method, a sub-optimal feedback control input is obtained by iteration of laboratory experiments as the covariance of an artificially added noise tends to zero. Finally, we have performed experimental verification of the proposed methods with a robot manipulator.

Free Research Field

制御工学