Abstract: It is very common to observe, especially in Computer
Science studies that students have difficulties to correctly understand
how some mechanisms based on Artificial Intelligence work. In
addition, the scope and limitations of most of these mechanisms
are usually presented by professors only in a theoretical way,
which does not help students to understand them adequately. In this
work, we focus on the problems found when teaching Evolutionary
Algorithms (EAs), which imitate the principles of natural evolution,
as a method to solve parameter optimization problems. Although
this kind of algorithms can be very powerful to solve relatively
complex problems, students often have difficulties to understand
how they work, and how to apply them to solve problems in
real cases. In this paper, we present two interactive graphical
applications which have been specially designed with the aim of
making Evolutionary Algorithms easy to be understood by students.
Specifically, we present: (i) TSPS, an application able to solve the
”Traveling Salesman Problem”, and (ii) FotEvol, an application able
to reconstruct a given image by using Evolution Strategies. The
main objective is that students learn how these techniques can be
implemented, and the great possibilities they offer.
Abstract: This paper presents the novel Rao-Blackwellised
particle filter (RBPF) for mobile robot simultaneous localization and
mapping (SLAM) using monocular vision. The particle filter is
combined with unscented Kalman filter (UKF) to extending the path
posterior by sampling new poses that integrate the current observation
which drastically reduces the uncertainty about the robot pose. The
landmark position estimation and update is also implemented through
UKF. Furthermore, the number of resampling steps is determined
adaptively, which seriously reduces the particle depletion problem,
and introducing the evolution strategies (ES) for avoiding particle
impoverishment. The 3D natural point landmarks are structured with
matching Scale Invariant Feature Transform (SIFT) feature pairs. The
matching for multi-dimension SIFT features is implemented with a
KD-Tree in the time cost of O(log2
N). Experiment results on real robot
in our indoor environment show the advantages of our methods over
previous approaches.