|
Control >> Finding object/goal >> Goal search
Goal search
In this section, we present a second series of experiments,
in which a group of ten s-bots has to find a goal item-the
prey-placed at varying distances from the nest.
Similar to the task of the previously presented experiment,
the s-bots have to explore the environment by forming chains.
The difference is that a goal object is added to the environment. When
the goal is encountered by a chain, a connection has to be established,
in this way forming a path that connects nest and goal, and that allows
other s-bots to navigate between the two locations.
Experimental setup
We use the same controller as in the previous section.
Results
Figure 1 shows the success rates for the three strategies for
finding and connecting to a prey object which is placed at two
different distances from the nest. We define an experiment to be
successful if the group of s-bots is able to locate the prey
object within 1000 s.
The moving strategy in general performs better than the other
ones because robots aggregated into a chain contribute to the
exploration process by collectively moving around the nest. In
opposition to the other two strategies, a lower probability to
disaggregate from a chain results in a higher success rate because low
values of P(chain->expl) decrease the frequency of situations in
which a chain may not move, in this way increasing the time during
which a chain can explore the environment. The most successful values
of the probability to aggregate into a chain are in the range
[0.005,0.01]. Higher values lead to the formation of more than one
chain, and lower values, while leading to a single chain, result in a
slower chain formation process.
References
Control >> Finding object/goal >> Goal search
| 
