Session WedAT1 Pegaso A Wednesday, October 10, 2012 ... - Lirmm

Session WedAT7 Vega Wednesday, October 10, 2012, 08:30–09:30
Multiple Mobile Robot Planning I
Chair Ronald Arkin, Georgia Tech.
Co-Chair
08:30–08:45 WedAT7.1
Combining Classification and Regression for
WiFi Localization of Heterogeneous Robot
Teams in Unknown Environments
Benjamin Balaguer, Gorkem Erinc, and Stefano Carpin
School of Engineering, University of California, Merced, U.S.A.
• Solves the problem of robot localization
with wireless signals using data-driven
machine learning classification and
regression techniques.
• Implementation of six classification
algorithms, compared and evaluated on
two different datasets.
• Novel regression algorithm builds upon the
best classification algorithms.
• End-to-end algorithm exploits robots’
odometry with Monte Carlo Localization.
• Algorithm works in completely unknown
environments, builds maps efficiently, and
localizes in real-time.
Localization traces (SLAM, WiFi
MCL, and Ground Truth) for an
indoor exploration scenario
09:00–09:15 WedAT7.3
A Bio-Inspired Developmental Approach to
Swarm Robots Self-Organization
Yan Meng
Department of Electrical and Computer Engineering, Stevens Institute of
Technology, USA
Hongliang Guo
Almende Organizing Networks, Netherlands
• Inspired by biological morphogensis, a
developmental approach, i.e., network motifs
based gene regulatory network model (NM-
GRN), is proposed for self-organization of
swarm robots to autonomously generate
dynamic patterns to adapt to uncertain
environments
• First, a general GRN model is proposed with
predefined network motifs as building blocks,
then covariance matrix adaptation evolution
strategy is applied to evolve the structure
and parameters of the general GRN model
to build up the NM-GRN
• Experimental results demonstrate the
efficiency and robustness of the NM-GRN
model.
60
50
40
30
20
10
0
-10
-10 0 10 20 30 40 50 60
60
50
40
30
20
10
0
-10
-10 0 10 20 30 40 50 60
-10
-10 0 10 20 30 40 50 60
Swarm robots traverse uncertain
environments
60
50
40
30
20
10
0
60
50
40
30
20
10
0
-10
-10 0 10 20 30 40 50 60
08:45–09:00 WedAT7.2
Distributed Coordination of a Formation of
Heterogeneous Agents with Individual Regrets
and Asynchronous Communications
Nicolas Carlési
LIRMM, Univ. Montpellier 2, France
Pascal Bianchi
Institut Télécom / Télécom Paris-Tech, CNRS – LTCI, France
• Objective: a distributed algorithm able
to coordinate heterogeneous agents to
perform various missions.
• Proposed approach: each agent
minimizes a regret function which takes
into account natural motion constraints
and individual objectives in order to find
its control variables.
• Simulations: comparison of the agents’
behavior for different communication
scenarios.
2012 IEEE/RSJ International Conference on Intelligent Robots and Systems
–125–
The trajectories of the agents
09:15–09:30 WedAT7.4
Real-time Optimization of Trajectories that
Guarantee the Rendezvous of Mobile Robots
Sven Gowal and Alcherio Martinoli
DISAL, EPFL, Switzerland
• The decentralized rendezvous of
differential-wheeled robots is
investigated.
• The individual trajectories are
optimized according to a userdefined
cost function using receding
horizon control.
• Mathematical guarantees on the
convergence of the robots to a
common rendezvous location are
given.
Trajectories of 4 real Khepera III
robots performing the rendezvous