Information Technology Reference
In-Depth Information
1.00
1.00
0.75
0.75
0.50
0.50
0.25
0.25
0.00
0.00
-0.25
-0.25
-0.50
-0.50
-0.75
-0.75
-1.00
-1.00
0
50
100
150
0
50
100
150
10
3
)
10
3
)
Time steps (
Time steps (
×
×
(a)
s
isolated
(b)
s
assembled
Fig. 4.
Useful speed evolution for each robot
(or with low probabilities) to grasp activated or assembled blocks. In Fig. 4(a) we
see that during evolution the majority of the robots grasp the activated blocks.
Regarding the assembled blocks, we can observe in Fig. 4(b) that there are two
clear behaviors, with a portion of robots that grasp them and another portion
that ignore them. As a consequence, ASiCo achieves a stable solution where all
the robots grasp activated blocks but where only a portion of them is able to
grasp assembled blocks too. In addition, the algorithm provides the particular
distribution of robots required on each species to solve the task dynamically.
This result differs from what we expected, but, in fact solves the problem quite
eciently allowing for all the robots to be able to search the area and assemble
blocks and reserving the assembled block transportation capabilities (which is
more infrequent, requires more energy and implies less speed) to just some of
the robots, which specialize in this task although when idle help out in finding
and assembling blocks.
4 Conclusions
In this paper we have applied the Asynchronous Situated Co-evolution (ASiCo)
algorithm to a collective gathering and construction task where homogeneous
teams are suboptimal in order to show its capabilities for achieving emergent
specialization if required by the task. The results confirm that the algorithm ob-
tains two clear species in a scenario where the spatial separation between robots
is not an evolutionary advantage, as in previous works. With this specialization,
coordination is successfully achieved in an autonomous way.
Search WWH ::
Custom Search