Cooperative dynamic pathfinding for multiple autonomous underwater vehicles using D*

PDF

Authors
  1. Myers, V.
Corporate Authors
Defence R&D Canada - Atlantic, Dartmouth NS (CAN)
Abstract
A method for finding least-cost paths through an area by collaborating agents is presented. The method relies on extending existing dynamic path planning algorithms, namely D*, to the multiagent case. The search tree is centrally updated and then used by the agents to cooperatively decide upon which areas to survey, the results of which are then used to update the current best path. Several numerical examples are given where it is shown that, on average, a competitive ratio—the ratio between the cost of the path found by dynamic algorithm and one operating with full knowledge of the area — of nearly one can be achieved while requiring a survey of less than half the total area. It is also shown that the total mission time scales linearly with the number of vehicles, meaning little to no effort wasted on vehicle coordination. In the context of mine countermeasures (MCM) operations with autonomous underwater vehicles (AUVs), the results indicate that a channel through a potentially mined area representing an amount of risk nearly equal to the optimal path can be found in significantly less time than the time required to survey the entire area.

Il y a un résumé en français ici.

Keywords
Autonomy;AUV;Automatic Target Recognition;MCM
Report Number
DRDC-ATLANTIC-TM-2011-044 — Technical Report
Date of publication
01 Jan 2011
Number of Pages
40
DSTKIM No
CA046312
CANDIS No
806698
Format(s):
Electronic Document(PDF)

Permanent link

Document 1 of 1

Date modified: