IEEE Robotics & Automation Magazine - June 2018 - 54

(b)

rating of four or above. We noticed
that the three PLC programmers without other programming experience
rated the questions as being difficult.
Some limitations of the language
with respect to viscosity (resistance to
change) and diffuseness (number of
symbols for one meaning) were identified. These are addressed in [15].

From Research to Industry
In the project's second phase, the infor(a)
(c)
mation collected in the first phase was
used to design the human-robot interFigure 2. Proofs of concept using the Baxter robot for multimodal interactions: (a) for
action of a cobot implemented at the
face recognition and (b) and (c) for gestures. (Photo courtesy of Robovision.)
Audi production plant. The data
include the social cues identified
during the experiments as well as the technical solutions
for multimodal human-robot interaction (gesture and
face recognition).

Figure 3. An example of a manually glued reinforcement plate.
The applied glue stripes are not uniformly distributed along the
plate. (Photo courtesy of the Audi Brussels factory.)

the Baxter robot. The gesture recognition is performed by
the SoftKinetic camera DS325. Features are extracted from
the three-dimensional (3-D) point cloud and mapped to
static gestures.
We conducted an evaluation of the language to investigate
whether different stakeholders [including engineers, robot
and programmable logic controller (PLC) programmers, and
all employees of the industrial partners in the ClaXon project]
would be able to understand the language and make adaptations to existing specifications. The experiment started with a
short introduction to the language, followed by seven multiple-choice questions. On a scale of one to seven, participants
had to rate how easy it was to answer these questions, with
seven being the easiest. Each experiment ended with a short
discussion about the language.
Nine of the 17 participants made no mistakes, four made
one, and the others gave two wrong answers. An analysis of
the results showed that all 17 participants could understand
simple specifications, and all indicated some guidelines for
more complex specifications to maximize comprehensibility.
For all the questions, 11 or more participants gave an ease
54

IEEE ROBOTICS & AUTOMATION MAGAZINE

June 2018

Use Case
The use case at Audi Brussels focuses on applying glue to
reinforcement plates (see Figure 3). These are used to support
the car's roof racks. The worker's job consists of picking the
reinforcement plates from a container and stacking them on a
small table. By using a glue pistol, the operator applies two
stripes of glue on top of the metal plates. The worker then
attaches the appropriate reinforcement plates to the side panel
of a two- or four-door car from two parallel assembly lines.
The drawback of this whole process lies in the gluing.
Since this is done manually, the quality is not optimal, i.e., the
glue is not uniformly distributed along the plate. It is also
dependent on the skills of the person performing the gluing
task and other factors, such as the temperature and time of
day. This was affecting the quality of the produced cars. In
this case, the use of a high-precision cobot enables a better
product quality while taking over the human worker's repetitive and messy task. This use case is also used as an example
to demonstrate the technologies developed throughout the
ClaXon project.
Cobot Walt
The cobot that was integrated in the Audi Brussels production plant was named Walt. Figure 4 shows the developed
MRK-Systeme robot, along with its interaction with factory
workers. Walt works in close proximity to humans and does
not need a safety fence. An actuated robotic head was developed by Robovision to express emotions for human-robot
communication. Relevant social cues identified during the
social experiments were implemented, such as eye movements and head nodding and shaking.
Gesture-Based Interaction
Because of the noisy environment in the factory, gestures were
used to communicate with the robot. The operator instructs

IEEE Robotics & Automation Magazine - June 2018

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - June 2018