Stochastic optimal enhancement of distributed formation control using Kalman smoothers. Issue 2 (March 2014)

• Record Type: Journal Article
• Authors: Anderson, Ross P.; Milutinović, Dejan
• Is Part Of: Robotica. Volume 32:Issue 2(2014)
• Page Start: 305
• Page End: 324
• Publication Date: 2014-03
• Subjects: Robots -- Periodicals; 629.89205
• DOI: 10.1017/S0263574714000022
• Languages: English
• ISSNs: 0263-5747

Abstract

<abstract abstract-type="normal"> <title>SUMMARY</title> <p>Beginning with a deterministic distributed feedback control for nonholonomic vehicle formations, we develop a stochastic optimal control approach for agents to enhance their non-optimal controls with additive correction terms based on the Hamilton–Jacobi–Bellman equation, making them optimal and robust to uncertainties. In order to avoid discretization of the high-dimensional cost-to-go function, we exploit the stochasticity of the distributed nature of the problem to develop an equivalent Kalman smoothing problem in a continuous state space using a path integral representation. Our approach is illustrated by numerical examples in which agents achieve a formation with their neighbors using only local observations.</p> </abstract>