IEEE Robotics & Automation Magazine - June 2011 - 15

*
Further collaborating with
Autonomous Systems LaboEsatroll, we made use of their
ratory [1], [2].
l Map-The
map of the
cost-efficient forklift for logistics
environment is a graphand launched Paquito (Figure 3),
like structure with nodes
an industrial forklift running
representing positions the
with a payload of 1,200 kg,
robot has to reach in order
moving autonomously and
to perform a certain task.
safely through factory buildFurthermore it contains
ings. Since 2009, the ANTthe information about
driven Paquitos have been
all the features in the enoffering their services to
vironment, which are
different Italian companies
either segments (walls,
working 24 h a day, 7 days
furniture, or other struca week on production
tures) or points (table's
chains.
legs, other furniture,
A new wind was
Figure 2 Gilberto guide.
or, when needed, reflecblowing in 2010 with
tors). This permits to
the collaboration with
calculate which feature is visible from
KOKEISL Industrial Systems AG, an
the current position of the robot and
international vendor and system proto use it for localization.
vider of automated systems for the bulk
solids industry and AGVs (Figure 4). l Localization-The ANT localization implements an extended KalOur joint work resulted in the KOKEISL
man filter [3]. The filter uses the
picking robot (KPR), a modularly built
wheel encoder to calculate the odovehicle powered by ANT. The robot is a
metrical position and integrates the
multitalented one, collecting bulk mateextracted features presented above
rials from different batches from a few
in a weighted manner based on
kilograms to two tons. Working in
their uncertainty. The use of these
harmony with the law of gravity, the
features makes the position estiKPR gathers materials vertically. Using
mation highly robust in dynamic
automatic shovels, it can dose and asenvironments.
semble formulas and small quantities
accurately to the nearest gram. The l Global planning-The global planning is based on the aforemenKPR can autonomously withdraw matetioned graph structure, where the
rials from almost any container in the
nodes are locations of interest and
plant and looks after the transport
edges denote traversability between
between the warehouse, the dispenslocations. The planner employs a
ing terminals, and the mixing stations
depth-first search and generates a
with a positioning precision of 610 mm,
length-optimal path.
thanks to ANT.
ANT-Our Core Technology
The recurrent theme in all these
developments is ANT, which has over
time become our main focus and most
successful product. ANT is an environment-based navigation system that
can be installed on literally any type of
mobile system requiring localization
and navigation skills.
ANT uses several components to
implement the navigation tasks, such
as map, localization, global planning,
and motion control. These components are extensions of the technologies formerly developed at the

Figure 3 Esatroll's Paquito.

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - June 2011