DMCA
ThB05.4 Decentralized Algorithms for Vehicle Routing in a Stochastic Time-Varying Environment
BibTeX
@MISC{Frazzoli_thb05.4decentralized,
author = {Emilio Frazzoli and Francesco Bullo},
title = {ThB05.4 Decentralized Algorithms for Vehicle Routing in a Stochastic Time-Varying Environment},
year = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Abstract — In this paper we present decentralized algorithms for motion coordination of a group of autonomous vehicles, aimed at minimizing the expected waiting time to service stochastically-generated targets. The vehicles move within a convex environment with bounded velocity, and target generation is modeled by a spatio-temporal Poisson process. The general problem is known as the m-vehicle Dynamic Traveling Repairperson Problem (m-DTRP); the best previously known control algorithms rely on centralized a-priori task assignment and locational optimization, and are of limited applicability in scenarios involving ad-hoc networks of autonomous vehicles. In this paper, we present a new class of algorithms for the m-DTRP problem that: (i) are spatially distributed, scalable to large networks, and adaptive to network changes, (ii) are provably locally optimal in the light load case, and (iii) achieve the same performance as the best known centralized algorithms in the heavy-load, single-vehicle case. Simulation results are presented and discussed. I.
Keyphrases
stochastic time-varying environment decentralized algorithm vehicle routing autonomous vehicle bounded velocity m-dtrp problem convex environment stochastically-generated target network change general problem simulation result light load case known centralized algorithm large network new class a-priori task assignment waiting time target generation motion coordination locational optimization m-vehicle dynamic traveling repairperson problem spatio-temporal poisson process single-vehicle case ad-hoc network limited applicability
