Fundamental Paper on Formation Control

designpadAI and Robotics

Dec 1, 2013 (3 years and 9 months ago)

59 views

Fundamental Paper on Formation Control


For CS 7631: Multi Robot Systems



Diogenes Molina


School of Interactive Computing


Georgia Tech



March 15, 2012


Radhika
,
Nagpal
, “Programmable Self
-
Assembly
Using Biologically
-
Inspired
Multiagent

Control,”
in
Proceedings of the 1
st

International Joint
Conference on Autonomous Agents and
Multiagent

Systems
, pp. 418
-
425, New York, 2002.


Problem statement and motivation


Brief overview of previous work


Insights from developmental biology


Proposed
multiagent

control methodology for
programmable self
-
assembly


Origami
-
inspired operations for global shape specification


Cell biology
-
inspired primitives


From global shapes to local primitives


Simulation results


Paper critique


Questions


How to design/program self
-
assembling complex
structures from locally interacting homogeneous
agents?

?????


Emerging technologies make approaches of this
type attractive


But. mapping global objectives to decentralized
local rules is hard


Limited agent resources


Reliability


Hierarchical and centralized approaches dominate


Can be unreliable and not fault
-
tolerant


Expensive and complex agents can be costly


Approaches based on cellular automata and
artificial life are difficult to generalize


There is no framework for generating local rules
automatically


Learning and evolutionary approaches



Biological systems are a clear proof that this can be
done effectively!


The inner workings of these systems can provide
inspiration


Morphogenesis and developmental biology as a
source of mechanisms and general principles


Implement some of these mechanisms as a
programming language


Epithelial cell behavior


Sheet and Agent Model


OSL: origami
-
inspired language for global shape
specification


Biologically
-
inspired local primitives


Gradients


Neighborhood query


Polarity inversion


Cell
-
to
-
cell contact


Flexible folding


Axiom implementation via local primitives


From OSL to agent program


Many shapes cannot be constructed using the
axioms


The initialization of the sheet is providing a
certain amount of “global” information


The simulation environment seems too artificial


Not clear about how the folding process is
executed


Some global synchronization must be occurring,
but not discussed


The symmetry of the sheet might not be satisfied
in some cases, but could have a huge impact

http://www.bio.davidson.edu/people/kabernd/BerndCV/Lab/EpithelialInfoWeb/index.html

http://en.wikipedia.org/wiki/Epithelium