The paper presents a novel approach to the decentralized task assignment for multiple cooperative unmanned air systems, in a multiple target-multiple task environment. The vehicles (or agents) may have complete or partial a priori information about the targets that populate the scenario. Each vehicle autonomously computes the cost for servicing each task available at each target using a path planning algorithm, taking into account obstacles, pop-up threats, and weights the total path cost including potential risk areas. Vehicles assign an initial ranking to each task, and then exchange their ranking information with the others. Each agent then updates the ranking of its tasks using a non linear dynamic programming algorithm that is proven to be stable and to converge to an equilibrium point. The ranking dynamics is initially formulated as a continuous time system, and then time-discretized depending on available data, and transmission rate among the network. Stability of the network and independence of steady state values from the data rate are proved analytically. Current studies are directed towards the effect of communication delays. The validity and performance of the proposed method are verifed via extensive numerical simulation, and compared with alternate techniques such as an optimal MILP based integrated task assignment and path planning solver. Copyright © 2007 International Federation of Automatic Control All Rights Reserved.
Cooperative Task Assignment using Dynamic Ranking
INNOCENTI, MARIO;POLLINI, LORENZO
2008-01-01
Abstract
The paper presents a novel approach to the decentralized task assignment for multiple cooperative unmanned air systems, in a multiple target-multiple task environment. The vehicles (or agents) may have complete or partial a priori information about the targets that populate the scenario. Each vehicle autonomously computes the cost for servicing each task available at each target using a path planning algorithm, taking into account obstacles, pop-up threats, and weights the total path cost including potential risk areas. Vehicles assign an initial ranking to each task, and then exchange their ranking information with the others. Each agent then updates the ranking of its tasks using a non linear dynamic programming algorithm that is proven to be stable and to converge to an equilibrium point. The ranking dynamics is initially formulated as a continuous time system, and then time-discretized depending on available data, and transmission rate among the network. Stability of the network and independence of steady state values from the data rate are proved analytically. Current studies are directed towards the effect of communication delays. The validity and performance of the proposed method are verifed via extensive numerical simulation, and compared with alternate techniques such as an optimal MILP based integrated task assignment and path planning solver. Copyright © 2007 International Federation of Automatic Control All Rights Reserved.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.