Biomedical Engineering Reference
In-Depth Information
Table 2. Unsuccessful outcomes for the simulator. Code expresses the level of importance of the unsuc-
cessful outcome (the highest, the worst)
Code
U nsuccessful outcome
The goal was not reached but the robot did appropriate gate passing.
1
2
The goal was reached but the robot did not pass through gates AB.
Passed through the gate and stopped (stuck) near the gate.
3
Became trapped and could not escape before pass the gate.
4
in computer simulations the most promising
techniques for behaviour coordination, within
ER. They are the layered evolution coordination
and the AIS-based behaviour coordination. The
simulated robot was allocated to explore during a
fixed amount of time its environment (maximum
300 iterations). If it was successful in its task
completion, the time taken was recorded. If it
was unsuccessful the causes were recorded. Four
causes were set for experiments and are shown
in Table 2. All significant tests were at the 95%
confidence level using a t-test.
Experimental settings were similar than the
ones presented in the previous ER experimental
section, with Khepera
©
-based simulated robots.
The robot has the ability to sense obstacles (e.g.
walls and objects) and to sense the proximity to the
target (light source). Similar experiment settings
are proposed also in Whitbrook (2005), Vargas
et al. (2003), and Ishiguro et al. (1996). The task
consisted on passing through a gate (AB) in a
small simulated arena (Figure 10).
In the first aim, the robot is required to navi-
gate reactively in a safe way around the arena
and stop near the light source (goal) once it had
passed through the gate AB once. It was restricted
to pass through gaps XA or BY at any time due
their stretch separation with walls. It was given
a fixed number of iterations to complete the task,
starting from P0, P1, or P2 (initial positions) 10
times (independent tests). Note that a long-term
adaptation would require that the robot goes
as many times as possible to the goal, passing
through the gate in any direction. This was not
carried out here for short. As it may be seen, the
Figure 10. Experimental settings for comparison
between layered evolution and AIS based behav-
iour coordination
A Comparative Analysis
A conceptual comparison among different be-
haviour coordination approaches is always nec-
essary to analyze their scopes, possibilities and
applicability ranges. However, in engineering
areas it is also need an exhaustive comparison
among techniques within a specific domain of ap-
plication. Although this section is far from being
such exhaustive analysis, it pretends to evaluate
Search WWH ::
Custom Search