Cloud-Based Centralized-Decentralized Multi-Agent Optimization with Communication DelaysReport as inadecuate


Cloud-Based Centralized-Decentralized Multi-Agent Optimization with Communication Delays


Cloud-Based Centralized-Decentralized Multi-Agent Optimization with Communication Delays - Download this document for free, or read online. Document in PDF available to download.

We present and analyze a hybrid computationalarchitecture for performing multi-agent optimization. The optimizationproblems under consideration have convex objectiveand constraint functions with mild smoothness conditionsimposed on them. For such problems, we provide a primaldualalgorithm implemented in the hybrid architecture, whichconsists of a decentralized network of agents into which anupdated dual vector is occasionally injected, and we establishits convergence properties. In this setting, a central cloudcomputer is responsible for aggregating information, computingdual variable updates, and distributing these updates to theagents. The agents update their primal state variables and alsocommunicate among themselves with each agent sharing andreceiving state information with some number of its neighbors.Throughout, communications with the cloud are not assumedto be synchronous or instantaneous, and communication delaysare explicitly accounted for in the modeling and analysis of thesystem. Experimental results for a team of robots are presentedto support the theoretical developments made.



Georgia Robotics and InTelligent Systems Laboratory GRITS Publications - Georgia Robotics and InTelligent Systems Laboratory GRITS -



Author: Hale, Matthew T. - Nedić, Angelia - Egerstedt, Magnus B. - -

Source: https://smartech.gatech.edu/



DOWNLOAD PDF




Related documents