Task Allocation with Manipulative Dynamic Auctioneering for Multi Robot Systems

Abstract:

This research proposes to improve standard auction-
eer systems in Multi Robot Task Allocation (MRTA) with a novel
auctioneering strategy called ‘Manipulative Dynamic Auction
System’ (MDAS), which is inspired by the ‘Leaky Integrate and
Fire’ neuron model of the human brain. This model is developed
and simulated as an extension ofMRTeAm, a ROS based software
framework built to test MRTA auctioneering strategies. The
performance of MDAS is compared against a simpler version of
Dynamic Auctioneering named ‘Simple Dynamic Auctioneering
System’, as well as a standard Stationary Auctioneering System
called the ‘Ordered Single Item Auction’, using a range of
experiments. It is observed that MDAS is faster and more efficient
than the Simple Dynamic Auctioneering System. Also, it is more
sophisticated in its allocation of tasks to robots, when compared
to Stationary Auctioneering Systems, due to its consideration
to the behavior of robots during the auction process. Potential
for future research lies in building a Hybrid Auctioneering
system using a combination of both Stationary and Dynamic
Auctioneering Strategies for task allocation.

Download PDF