An Auction-based Position Selection Algorithm for Efficient Reconfiguration in Modular Robots

Advisor Information

Raj Dasgupta

Location

UNO Criss Library, Room 107

Presentation Type

Oral Presentation

Start Date

7-3-2014 3:15 PM

End Date

7-3-2014 3:30 PM

Abstract

We consider the problem of reconfiguration in modular self-reconfigurable robots where modules, starting from arbitrary locations, are required to assume appropriate positions so that they can get into a new target configuration. This problem is non-trivial as the desired positions of different modules in the target configuration could conflict with each other resulting in occlusions and failed attempts to achieve the target configuration; modules should also select positions that reduce their energy expenditure for locomotion and communication. To address this problem, we propose an algorithm called the spot allocation (SA) algorithm that uses a utility based model on each module to rank positions, followed by an auction-based technique to allocate positions to modules. We prove analytically that our algorithm is deterministic, complete and optimal (in case of no conflicts between modules). We have also verified the operation of our algorithm in simulation within the Webots simulator and compared the algorithm’s performance with other allocation strategies. Our results show that our proposed algorithm is able to successfully reconfigure different numbers of modules to different target configurations and performs better than the compared strategies in terms of run time, utility and violations of energy (battery) constraints.

This document is currently not available here.

COinS
 
Mar 7th, 3:15 PM Mar 7th, 3:30 PM

An Auction-based Position Selection Algorithm for Efficient Reconfiguration in Modular Robots

UNO Criss Library, Room 107

We consider the problem of reconfiguration in modular self-reconfigurable robots where modules, starting from arbitrary locations, are required to assume appropriate positions so that they can get into a new target configuration. This problem is non-trivial as the desired positions of different modules in the target configuration could conflict with each other resulting in occlusions and failed attempts to achieve the target configuration; modules should also select positions that reduce their energy expenditure for locomotion and communication. To address this problem, we propose an algorithm called the spot allocation (SA) algorithm that uses a utility based model on each module to rank positions, followed by an auction-based technique to allocate positions to modules. We prove analytically that our algorithm is deterministic, complete and optimal (in case of no conflicts between modules). We have also verified the operation of our algorithm in simulation within the Webots simulator and compared the algorithm’s performance with other allocation strategies. Our results show that our proposed algorithm is able to successfully reconfigure different numbers of modules to different target configurations and performs better than the compared strategies in terms of run time, utility and violations of energy (battery) constraints.