DESIGN OF A ROBOT TEAM FOR SEARCH IN CENTRAL
EUROPEAN FORESTS
Santi Esteva, Antoni Martorano
Institut d’Inform`atica i Aplicacions, Girona University, C/ Santal´o S/N, Girona, Spain
Albert Figueras
Institut d’Inform`atica i Aplicacions, Girona University, C/ Santal´o S/N, Girona, Spain
Keywords:
Autonomous robots, Surveillance, Robot exploration, Robot team design.
Abstract:
Security incidents are a new topic for the autonomous robot task. Robots, sensors and intelligence make a
great combination to achieve specific goals in this area. This paper proposes a methodology to design the size
and the exploration form for a group of robots in the typical Central European forest. The objective is to find
victims hidden into the fallen leaves. A statistics of the trouble situations is the starting point to design the
team to displace at the objective. The robots are autonomous but is necessary one human supervisor to solve
some critical situations. The results show the partition and the area covered by a simple group of equal robots.
1 INTRODUCTION
In the last years, security incidents, like terrorism,
have raised new research areas focused in surveil-
lance issues, the automation of these tasks is an im-
portant research area using robotic platforms and sen-
sors. The use of a team of robots and sensors is an
approach that has many advantages and difficulties
over a single robot platform. In this paper it is pro-
posed a framework to develop a multi-robot surveil-
lance platform (Caballero D., 2010), (Caballero D.,
2009). Surveillance is the act to close watch kept over
someone. Robotic surveillance is the topic to refer at
close watch on the relevant events. The automation
of the surveillance task are the approaches oriented
to the formalization of the problem (Everett H. R.,
1999), (Everett H. R., 2003).
Related works to segment the space to explore for
a multi-robot teams, several authors uses Voronoi di-
agrams of the space exploration to assign rectangu-
lar polygons strategies to segment the space explo-
ration (Matcovschi M. H., 2008). Other works uses
the agent entities and discuss the ability of the agent
to explore determined area (Rybski P. E., 2000). And
other determine an algorithm to achieve optimized re-
gion partition and coverage according the energy and
some constraints (Hao Wu, 2008).
This work analyzes the optimum partition of the
space exploration according probabilistic information
of the abnormal situations for the robot, which is nec-
essary a human operator support. Optimizing the effi-
ciency of the whole team, robots and support and also
taking information from the biological communities
(Martorano, 2010) is the objective of the presented
work.
2 THE TERMITES
The Termites colonies are groups of social insects
composed by a social organization to share tasks in a
specialized form. The Workers, the Soldiers and Re-
productives elements are a self-organized system in a
decentralized form for a common collective behavior.
The Caste system of the Termite cologne depends
on the environmental scenarios and depredators, from
a 2% of soldiers in Africa zones to 12% of soldiers in
a Formosa Island.
Genetic information characterizes the Caste, size
and composition. Also stress situations can adapt the
system to the new changes, as temperature, depreda-
tors, etc. The biological process into the live of the
Termite insect evolves the colonies adapting at the
evolution phase of the group. The differences of the
Caste proportion are justified to improve the colony
survival at different conditions; the analysis of this in-
325
Esteva S., Martorano A. and Figueras A..
DESIGN OF A ROBOT TEAM FOR SEARCH IN CENTRAL EUROPEAN FORESTS.
DOI: 10.5220/0003570603250328
In Proceedings of the 8th International Conference on Informatics in Control, Automation and Robotics (ICINCO-2011), pages 325-328
ISBN: 978-989-8425-75-1
Copyright
c
2011 SCITEPRESS (Science and Technology Publications, Lda.)
Figure 1: Proportion of soldiers depending on the geo-
graphic area.
Figure 2: Graph of efficiency as the number of workers-
soldiers.
formation induces to propose different size and Caste
of a team of robots for each scenario.
Some experiments for one defined scenario with
different Castes show the result of the improvement
for the colony survival. The improvements are done
when the size of the colony is small and the Caste
ratio is at the lower rates. From these results is trans-
formed to other representation the improvement at in-
terchanging the role between one Worker to a Soldier.
Figure 3: Gain performance by the number of workers-
soldiers.
In this representation indicates the minimum size
and composition of the team to gather these best im-
provements.
3 TEAM OF ROBOTS
The robot team can be similar to the Termite colony
taking account several similarities or equivalences in
order to transport the termite results to the robot team:
The different role of the termite elements can be
viewed as robots specialization. Depredators in Ter-
mites are the blocking situations in the robot area.
Nest of termite is the operation centre on the robots
field.
There are also several aspects with big differences,
as size of the colony, objectives, movement, etc. The
main result of this comparison is: there exist one
Caste relation and size that will be optimal for a one
defined task and area.
4 THE ROBOT
The team of robots are composed by several units of a
specialized robot to autonomously navigate in forest
areas, coordinated by an operation center, where a hu-
man operator controlled all the system, the positions,
the finding targets, the stopped or blocked robots.
Figure 4: Unit of our autonomous robot.
The robot characteristics are:
Size: 51cm x 58 cm x 32 cm.
Weight: 5 Kg.
Batteries: 12V 14 Ah.
Autonomy: 2 hours.
The robot is equipped by:
- DGPS with 15 cm. global position accuracy.
- Laser Range Finder with 240 and 4m range obstacle
detection.
- Inertial Measurement Unit, which combines 3 gyros
and 3 accelerometers.
- Thermal sensor with 1m. range measure.
- Methane sensor.
- Digital camera.
ICINCO 2011 - 8th International Conference on Informatics in Control, Automation and Robotics
326
- WI-FI communication.
- Four-Wheeled Driver 60W.
5 EXPLORATION
The robot ability to explore unknown areas is demon-
strated, using D-GPS in outdoor scenarios and
equipped by four strong wheeled drive, and the robot
can easily navigate by typical Central European for-
est, due by their characteristics.
Some handicaps can be found in this kind of ter-
rains, holes, climbs, obstacles, water, etc. and cannot
be over passed by the robot.
Four typical blocking situations for the robot:
Figure 5: Lateral climbing.
Figure 6: Obstacle not detected.
A 2D Laser range sensor is useful to detect and
predict some of these troubles, the others can cause
problems to the robot and will be necessary the use of
other kind of support to solve these handicaps.
To define the support for the robots according the
number of the explorers will be defined according the
obstacle proportions for the exploring area.
Figure 7: Hole and slipping situation.
Figure 8: Changes in the field.
6 EXPERIMENTATION
One forest is necessary to be explored to find a victim,
dead or alive. The area is divided in circles or squares
where the robots are working, jointly with one human
operator to restore the robot at blocking situations.
Each area where the robots are moving is organized
to be covered in 2 hours of autonomy batteries for a
group of N robots. The velocity of the robot is 1 m/s
and during this time is covered 1 square meters of ex-
ploration. The objective is to configure the surfaces
to cover by a group of N robots, maintained by only 1
human operator. The probability of the blocking sit-
uation is 1 over 400, this means during 400 sec. one
abnormal situation appears. And the operator needs
the distance in time to arrange the situation.
With all this premises the analysis over square ar-
eas or circular is the following:
Figure 9: Square or circular areas.
The gravity center of a circular sector is 2/3R, and
DESIGN OF A ROBOT TEAM FOR SEARCH IN CENTRAL EUROPEAN FORESTS
327
over Diagonals of the square is L/
2, and in Verti-
cal or Horizontal lines is L/2, the mean result 0.6 L.
Assigning this distance for same surface covered re-
sult R=L/
π, and combined with the gravity center,
resulting the distance for circular areas 73% less than
square areas. This means the time to recover the robot
improves using areas limited by circles.
Once the shape of the areas is defined, is also nec-
essary to define the number of the robots working in
and the size of the area.
The number of robots is defined by the working
limit of the human operator, as much robots much
work to restore the operability of the robot front a
blocking situations.
Time between abnormal situation appears each
400 sec/ N , and during this time the operator uses
2/3 R in time to restore the robot:
400 / N = 2/3 R
And using the other formula to solve the problem, the
time to cover this area is 2 hours:
R2/N = 23600
Resulting a N = 8 robots and R = 135 m
Figure 10: The areas can be rearranged at hexagonal forms
maintaining the results.
7 CONCLUSIONS
The ability of the robot for this subject is the accurate
measure of the Methane or Thermal information and
their interpretation, the persistence and the working
journey. The humans have ability on the movements,
but not at the inspection and data interpretation, and
also are affected by the fatigue.
This article try to import some information from
the evolution process over the adaptation of the ter-
mites at different mediums in millions of years, and
incorporate any information as improvements for a
team of robots designed to exploring a forest area.
The characteristics of the distribution, and the
caste particularities of the termites colonies are the
principal focus where this work is centered, and the
results show one possible characterization of the team
of robots specialized on search in a forest area.
REFERENCES
Caballero D., Esteva S., e. a. (2009). Designing The Size
Of a Robot Team For Search And Rescue. ISSN1454-
8658.
Caballero D., Esteva S., e. a. (2010). Fleet size of robots
for rescue missions. Proceedings of IEEE Interna-
tional Conference on Automation, Quality and Test-
ing, Robotics.
Everett H. R., e. a. (1999). From Laboratory to Ware-
house: Security Robots Meet the Real World. The In-
ternational Journal of Robotics Research 1999, DOI:
10.1177/02783649922066448.
Everett H. R., e. a. (2003). Robotic Security Systems. IEEE
Instrumentation and Measurement Magazine.
Hao Wu, Guohui Tian, X. L. (2008). Research on opti-
mized partition algorithm for multi robots collabora-
tion coverage. Intelligent Control and Automation,
2008. WCICA 2008. 7th World Congress on.
Martorano, A. (2010). Required to incorporate SWARM in-
telligence into the equipment design of robot rescue
teams. Master Tesis, Mster MIIACS, University of
Girona, Director: Santi Esteva.
Matcovschi M. H., P. O. Stability and Set-Invariance Test-
ing for Interval Systems. Journal of Control Engineer-
ing and Applied Informatics, Vol.10, No.3 , pp.18-26,
2008, ISSN: 1454-8658.
Rybski P.E., e. a. (2000). A Team of Robotic Agents
for Surveillance. AGENTS ’00 Proceedings of
the fourth international conference on Autonomous
agents, ACM New York, NY, USA 2000., isbn: 1-
58113-230-1 edition.
ICINCO 2011 - 8th International Conference on Informatics in Control, Automation and Robotics
328