• Sorted by Date • Classified by Publication Type • Classified by Topic • Grouped by Student (current) • Grouped by Former Students •

Ilan Lupu. ** Optimal Construction of Control Graphs in Multi-Robot Systems**. Master's Thesis, Bar Ilan University,2015.

Control graphs are used in multi-robot systems to maintain information about which robot senses another robot, and at what position. On the basis of such graphs, it is possible to compute a shared coordinate system, localize relative to others, and maintain stable formations. While existing work shows how to utilize control graphs for these tasks, it makes two critical assumptions. First, it assumes edge weights of control graphs are single deterministic scalars. However in reality there are many stochastic factors (e.g, latency, resource costs, or position errors), that affect optimality of control graphs. Second, it assumes that a single robot is given, to serve as the global anchor for the robotsâ€™ relative positioning and location estimates. However, optimal selection of this robot is an open problem. In this work, we generalize control graphs to distinguish different stochastic sensing factors that may be represented by control graphs, beyond existing work, and discuss risk-based policies for their treatment. We show that existing work in coordinate frame alignment and formation maintenance may be recast as graph-theoretic algorithms inducing control graphs for more general representation of the sensing capabilities of robots. We then formulate the problem of optimal selection of an anchor, and present a centralized algorithm for solving it. We evaluate use of these algorithm on physical and simulated robots equipped with depth and image sensors (RGB-D cameras), and show they very significantly improve on existing work.

@MastersThesis{lupu-msc, author = {Ilan Lupu}, title = {Optimal Construction of Control Graphs in Multi-Robot Systems}, school = {{B}ar {I}lan {U}niversity}, year = {2015}, OPTkey = {}, OPTtype = {}, OPTaddress = {}, OPTmonth = {}, OPTnote = {Available at \url{http://www.cs.biu.ac.il/~galk/Publications/b2hd-lupu-msc.html}}, OPTannote = {}, wwwnote = {}, abstract = {Control graphs are used in multi-robot systems to maintain information about which robot senses another robot, and at what position. On the basis of such graphs, it is possible to compute a shared coordinate system, localize relative to others, and maintain stable formations. While existing work shows how to utilize control graphs for these tasks, it makes two critical assumptions. First, it assumes edge weights of control graphs are single deterministic scalars. However in reality there are many stochastic factors (e.g, latency, resource costs, or position errors), that affect optimality of control graphs. Second, it assumes that a single robot is given, to serve as the global anchor for the robotsâ€™ relative positioning and location estimates. However, optimal selection of this robot is an open problem. In this work, we generalize control graphs to distinguish different stochastic sensing factors that may be represented by control graphs, beyond existing work, and discuss risk-based policies for their treatment. We show that existing work in coordinate frame alignment and formation maintenance may be recast as graph-theoretic algorithms inducing control graphs for more general representation of the sensing capabilities of robots. We then formulate the problem of optimal selection of an anchor, and present a centralized algorithm for solving it. We evaluate use of these algorithm on physical and simulated robots equipped with depth and image sensors (RGB-D cameras), and show they very significantly improve on existing work. }, }

Generated by bib2html.pl (written by Patrick Riley ) on Mon Nov 16, 2020 22:25:46