Artificial Intelligence and Robotics

loutclankedAI and Robotics

Nov 13, 2013 (3 years and 1 month ago)

44 views



©

200
5

EU
-
MOP Consortium

Athens, June 2006

EU
-
MOP

Artificial Intelligence

and Robotics

Dennis Fritsch


Fraunhofer
-
Institute

for Manufacturing

Engineering and

Automation (IPA)



©

200
5

EU
-
MOP Consortium

Introduction


The EU
-
MOP robots should be
able to response to oil spills
autonomously.


Autonomously means that each
unit will have an energy system
and an oil skimming device as
well as that the EU
-
MOP units
will make its own decisions
how to response to the oil spill.


Thus, each unit needs
(artificial) intelligence, which
will be given to the units with
the help of sensors and control
systems.



©

200
5

EU
-
MOP Consortium

Control of the EU
-
MOP units


In order to achieve an highly
robust and flexible oil
response system the swarm
intelligence approach has been
selected as control paradigm
for the EU
-
MOP robots.


Thus, the EU
-
MOP swarm is a
homogeneous group of robots
without any hierarchies or
central control system.





©

200
5

EU
-
MOP Consortium

Control of the EU
-
MOP units


Example: Oil spill in a harbour, several very small
patches of oil have to be recovered:



©

200
5

EU
-
MOP Consortium

Control of the EU
-
MOP units


Strategy of each robot:



©

200
5

EU
-
MOP Consortium

Control of the EU
-
MOP units

4 following robots

1 robot moving random paths

pier

oil



©

200
5

EU
-
MOP Consortium

Control of the EU
-
MOP units

obstacle occurs



©

200
5

EU
-
MOP Consortium

Control of the EU
-
MOP units

swarm 1:

1 following robots

1 robot moving


random paths

swarm 2:

2 following robots

1 robot moving


random paths



©

200
5

EU
-
MOP Consortium

Control of the EU
-
MOP units

swarm 1 and swarm 2

merge, but the “leading”
robot has a malfunction.



©

200
5

EU
-
MOP Consortium

Control of the EU
-
MOP units

swarm will again be split
into two swarms with
each 1 “leading” robot
and 1 following robot.

Thus, the EU
-
MOP system will be

very flexible and fault
-
tolerant.



©

200
5

EU
-
MOP Consortium

Sensors for the EU
-
MOP robots


Nevertheless, the EU
-
MOP robots
will need several sensors, e.g.:

-
for the detection of oil or the measurement
of the thickness of the oil spill,

-
for absolute or relative positioning of itself,
its neighbours, all other units, the mother
ship, etc.,

-
for the detection of collisions with other
units, shipwrecks, debris etc.,

-
for winds, currents, etc.,

-
for the internal state of the unit
(malfunction, full of oil, energy low etc.)

-
as well as a communication system for
communication with other units, with the
mother ship and with a human operator.



©

200
5

EU
-
MOP Consortium

Sensors for the EU
-
MOP robots


Sensors might increase the
performance of the robots,
nevertheless, sensors also have
disadvantages, e.g.:

-
increased costs

-
increased weight and volume

-
increased consumption of power

-
increased amount of information that has
to be processed


Thus, the question is: What is the
best sensor configuration for the
EU
-
MOP robots? And related to
that: What are the best strategies
in order to response to the oil spill.


These questions will be answered
with the help of the simulation
technique.



©

200
5

EU
-
MOP Consortium

Simulation of the EU
-
MOP robots


Architecture of the simulation



©

200
5

EU
-
MOP Consortium

Simulation of the EU
-
MOP robots


Oil Fate model



©

200
5

EU
-
MOP Consortium

Simulation of the EU
-
MOP robots


Robot / Swarm Simulation



©

200
5

EU
-
MOP Consortium

Simulation of the EU
-
MOP robots


Visualisation

water

oil slick

harbour

coast



©

200
5

EU
-
MOP Consortium

Simulation of the EU
-
MOP robots


Visualisation

1 unit



©

200
5

EU
-
MOP Consortium

Simulation of the EU
-
MOP robots


Visualisation



©

200
5

EU
-
MOP Consortium

Simulation of the EU
-
MOP robots


Visualisation



©

200
5

EU
-
MOP Consortium

Simulation of the EU
-
MOP robots


Visualisation



©

200
5

EU
-
MOP Consortium

Simulation of the EU
-
MOP robots


Visualisation



©

200
5

EU
-
MOP Consortium

Simulation of the EU
-
MOP robots


Visualisation



©

200
5

EU
-
MOP Consortium

Simulation of the EU
-
MOP robots


Comparison of two types of swarms:

-
swarm A consists of units
without

oil sensor

-
swarm B consists of units
with

oil sensor.


type B:

will be able to move

intelligent

type A:

will not be able to

move intelligent



©

200
5

EU
-
MOP Consortium

Simulation of the EU
-
MOP robots


Comparison of two types of swarms:






Conclusion (for

this
scenario)

-
The larger the swarm the better the recovery time.

-
The better the sensor configuration the better the recovery
time.

-
Nevertheless, the swarm without oil sensors reaches for
large swarm sizes more or less the same recovery time as
the swarm with oil sensor.

Swarm size (N)

1

2

3

4

5

10

20

Recovery time
of swarm A

52627

20496

16843

14170

13207

5564

2905

Recovery time
of swarm B

37293

14416

11360

9132

8134

4026

2860



©

200
5

EU
-
MOP Consortium

Conclusion


The simulation will be able to determine

-
the recovery time,

-
the energy consumption,

-
the quantity of recovery oil

-
the quantity of oil that polluted the coast,

-
etc.


These date will be the basis

-
for an assessment of the EU
-
MOP units, and

-
for a cost
-
benefit
-
analysis.


Thus, this proceeding will ensure that the EU
-
MOP consortium will develop a highly effective,
flexible and robust oil response system.