Robust Time Varying Formation Control With Adaptive Submodularity
Abstract
An adaptive formation controller is developed to position nodes within a mobile Network Control System (NCS) composed of heterogeneous agents. Each node is represented with distinct capabilities and constraints with regard to communications, sensing, and mobility. Metrics used to quantify network robustness are developed for weighted graphs. Formation control is implemented to position nodes relative to virtual leaders. A utility function that encapsulates the sensing, communications, robustness, and dynamics of the NCS is designed and shown to be submodular. Submodular function maximization is then used to adaptively recompute the optimal formation in simulation. Submodularity is a property of set functions, which guarantees near-optimal performance if a greedy algorithm is used to iteratively select node locations. This effectually reduces the NP-hard combinatorial optimization problem to a polynomial time process. The greedy algorithm is used to adaptively recompute the optimal formation in simulation. This controller reduces the complexity of employing large numbers of autonomous agents in support of competing objectives.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 2018
- Accession Number
- AD1060097
Entities
People
- Noah Wachlin
Organizations
- Naval Postgraduate School