International Journal of Pure and Applied Mathematics

Volume 116 No. 20 2017, 437-441

ISSN: 1311-8080 (printed version); ISSN: 1314-3395 (on-line version)
url: http://www.ijpam.eu
Special Issue
ijpam.eu

AUTOMATIC ROBOTIC SPRAY PAINTING OF LOW VOLUME HIGH VARIANT PARTS

1
G.Vasumathi, 2S.Elakkiya
1
Assistant Professor, Department of Mechatronics Engineering, BIST, BIHER, Bharath University, Chennai-73
2
UG Student, Department of Mechatronics Engineering, BIST, BIHER, Bharath University, Chennai-73
1
vasumathi.mechatronics@bharathuniv.ac.in

Abstract: With the advance in IT technology sensors

and processing power exist to achieve fully automated
Robot programming in industrial tasks. This paper
Reports on the EU-project Flex Paint, which devised a
Methodology to automatically generate robot Programs
for spray painting of unknown parts. The Solution uses
four steps: laser triangulation sensing, geometric feature
detection, tool path planning, and the generation of the
collision-free executable robot program. Demonstrations
at industrial partners show the results.

Keywords: Automated robot programming, feature

detection, path generation, collision avoidance, spray
painting.
1. Introduction
The objective of the European RTD project Flex Paint to
automate robot programming for painting applications of
small lot sizes with a very high number of part variants.
Presently robotic painting is economically feasible for
large lot sizes, since a robot program needs to be
established (using off-line programming and/or manual
teach-in) for each single part variant. The project goal is
to provide economic possibilities for usage of robots in
painting of large part families (e.g., see Figure 1).The
goal is to reduce human programming effort by75 percent
and the human spray [17] painting effort by 90 Percent.
The technical challenge is to develop a method that Can
fulfill the spray painting task similar to a human Painter,
who is able to paint a [16] geometry never seen Before
[1,2,3]. A possible solution is to utilize CAD data to
calculate a paint path and to measure the actual part
Location. However, 3D CAD data is often not fully
available and hence this approach is limited to few
Applications. Within Flex Paint an "inverse approach “Is
developed, which does not require any CAD data [4,5].
The procedure is to automatically obtain robotic paint
Paths from range sensor data and to[18] automatically
Generate a feasible, complete and executable robot
Program [6,7].
2. Related Work
Related work is the automatic generation of a 3D.Paint
path, which has been attempted in the Smart Painter
project. The painting motion was Generated by virtually
folding out the surfaces to be Painted, putting on the
painting motion and folding Back the surfaces and letting
the painting motions Following this folding of
surfaces[7,8] .However, This strategy is only applicable
when 3D models of The objects are available and the
curvature of the Objects is relatively small. The user
decides on the screen where to apply paint Strokes.
[20]The path planning for a robot is then done
automatically [9,10]. First approaches to obtain an
automatic tool path in3D are known from milling turbine
blades by considering planar cross sections or using a
grid cell approach.[19]
3. Flex paint Approach
The Flex Paint approach is based on the observation
That the parts comprise a large number of elementary
Geometries with typical characteristics for an entire
Product family[10,11].

437
 International Journal of Pure and Applied Mathematics
The specification of elementary geometry types is
based on the observed part geometries and on the
constraints of the painting process. The idea is to detect
elementary geometries that can be linked to specific
process model, that is a model for the painting process of
this elementary geometry[12,13]. For example, the
geometry "flat surface" can be painted with a simple
pattern of straight paint strokes. More complex geometric
shapes, such as cavities or ribs, need specific painting
strategies: [21]spraying into the cavity and painting
parallel to the rib orientation, respectively[14].
The elementary geometry types are defined in the
Geometry Library and related to the process
Knowledge, which is specified in the Procedure
3.1 Part Measurement with Laser Triangulation Sensor
When the parts move along a conveyor, a laser range
Sensing device scans the parts and obtains a 3D
Measurement of point data of the object. The scans
Are triggered by the actual part motion of the conveyor
For a part size of one meter[22]a resolution of better than
one millimeter is obtained.
Up to 700 scans per second are taken. This is
Obtained with a calibration
438
Special Issue
Schematic drawing of part measurement Using the sheet
of light system of IVP, Sweden. Often several parts are
mounted on frames or skids .In this case a scan of the
frame/skid is taken and Subtracted from the final
image[15,16]. Figure 5 shows an Example for a part
measured with frame and the part Data automatically
extracted. For every part detected All other processing
steps are performed individually. Finally, if parts are
close to each other, larger robot Painting motions are
obtained by painting several Parts with one stroke[17,18].
Original depth image of part and frame
(top) and the extracted part (bottom).
3.2 Feature Finder
The Feature Finder developed by ACIN has the task
Of detecting the part geometry using the geometric
Definitions of the geometry library. The output is a
Description of the scanned part surface using Elementary
geometries. Feature Finder detects three classes of
features, which are relevant for the painting process: free-
form Surfaces, cavities and rib sections[19,20]. Classical
range Image processing typically starts from
segmentation. Finding features is Then constrained to
defined geometric[15] properties such as planes or
conics. The specific geometry of the three classes of
features is not known; attributes (lower than surrounding
rim, Parallel ribs) define the feature. Hence, a generic
Approach to feature detection is required. This
summarizes the developed image processing
Procedure[21].
A cavity is defined as a region where surface points
 International Journal of Pure and Applied Mathematics
Are locally lower (in the sense of an outward surface
Normal) than a surrounding rim.[35] The challenge is to
Develop a robust procedure to handle noise and
Shadows of the range data.[36] The rim of the cavity is
detected with accuracy in the range of the resolution. By
using interpolation. Points the discovered cavities also
include potential Areas of sensor shadow. This is
important to detect. The full cavity region for the painting
process. The Cavity is then represented as a mesh that
covers the Opening of the cavity[22]. It shows the
misrepresentation of the complete part. The opening
2.3 Paint Planner
In the next process step the painting trajectory for the
Spray gun is calculated for each of the elementary
Geometries. Neighboring paint strokes are fused to
Obtain longer and smoother trajectories.[24] The final
Result is a [11] complete paint path for the spray gun.
The module "Generate Painting Trajectory", shown in
Figure 2, specifies a trajectory of the spray gun, Which
satisfies the desired paint quality [23]. In this Module
only spray gun motions are considered in Relation to
process quality. No restrictions of robots.Are made and
collisions between the spray gun and its surroundings are
not considered.[37] The module uses the Geometry
Library and the Procedure Library in order to plan this
trajectory.
Painting that particular type of geometric primitive.
The painting procedure specifies how to apply spray Gun
motions to the surfaces in order to achieve a Satisfactory
process quality.[14] The Procedure Library is Established
through experimental work[29].
The basic idea is to enable planning of paint strokes
That continues throughout the [30] parts even though
Different geometric primitives must be covered along
Each virtual surface represents only one.Type of
geometric primitive and the same painting. Procedures
can [26] be therefore used for the entire part Surface. In
case different geometric primitives are Present along the
surface, [13]t he system will attempt to Establish
continuous spray gun motions covering Virtual surfaces,
which are in continuation of each Other.[31-34]
3.4 Automatic Generation Of Collision-Free
Robot Programs
Next collision-free robot motion is established based
On the tool trajectory obtained from the Paint Planner.
The actual paint trajectories are Calculated for a given
robot manipulator. The Tool Planner software from
AMROSE Robotics ApS ,Is utilized for generating
collision-free motions[24]. This Tool has been
customized for the Flex Paint project And checks [12]if
any collisions occur between the robot And its
439
Special Issue
surroundings. It uses a simplified solid model(a convex
hull approximation) of the part to include The work piece
into collision avoidance. This Simplified geometry
model, frame and paint paths
4. Experimental Results
The system is already implemented as a prototype and
has been tested in ABB's technical center in Eichen,
Germany (see Figure 4).
The purpose of this experiment was to prove the
basic system concept. It was realized that process quality
has to be optimized by establishing validated painting
procedures for the individual geometric primitives [25].
The painting procedures [38] used was not established by
preceding experiments. Since the surface was scanned
only from one side of the part it was only possible to
perform automatic spray painting of the scanned surface.
5. Conclusions and Outlook
An approach to automatically spray paint families of
Unknown parts has been presented.[39] The approach
uses a sensing cell in front of the painting cell, where the
part geometry is acquired. From the part geometry
process-relevant [10]features are extracted and
Corresponding paint routines are found and grouped To
obtain optimal painting trajectories. Finally a Collision-
free robot path and an executable robot Program are
generated.[40-43]
References
[1] Hertling, P., Hog, L., Larsen, L., Perram, J.W.,
Petersen, H.G.: Task Curve Planning for Painting Robots
- Part I: Process Modeling and Calibration; IEEE
Transactions on Robotics and Automation 12(2), April
1996, 324-330.
 International Journal of Pure and Applied Mathematics
[2] Hoover, A., et.al.: An experimental comparison of
range image segmentation algorithms; IEEE Transactions
on Pattern Analysis and Machine
Intelligence 18(7), 1-17, 1996.
[3] Iqbal, Q., Aggarwal, J.K.: Applying perceptual
grouping to content-based image retrieval: Building
images, IEEE International Conference on Computer
Vision and Pattern Recognition, Fort Collins, Colorado,
Vol. 1, pp. 42--48, 1999.
[4] Kwok, K.S., Louks, C.S., Driessen, B.J.: Rapid 3-D
Digitizing and Tool Path Generation for Complex
Shapes; IEEE International Conference on Robotics and
Automation, 2789-2794, 1998.
[5] Lowe, D.G.: Three-Dimensional Object Recognition
from Single Two-Dimensional Images;Artificial
Intelligence 31(3), 355-395, 1987.
[6]Venkatesan N., Srinivasan V., Fabrication and
mechanical properties of natural composite materials,
International Journal of Applied Engineering Research,
v-9, i-22, pp-7743-7748, 2014.
[7]Mustafa Kamal Basha M., Srinivasan V., Fabrication
of AlSicMmc and analysis of its mechanical properties,
International Journal of Applied Engineering Research,
v-9, i-22, pp-7621-7626, 2014.
[8] Marshall, D.; Lukacs, G.; Martin, R.: Robust
segmentation of primitives from range data in the
presence of geometric degeneracy; IEEE Transactions on
Pattern Analysis and Machine Intelligence 23(3), 304 -
314, 2001.
[9]Gopalakrishnan K., PremJeya Kumar M.,
SundeepAanand J., Udayakumar R., Analysis of static
and dynamic load on hydrostatic bearing with variable
viscosity and pressure, Indian Journal of Science and
Technology, v-6, i-SUPPL.6, pp-4783-4788, 2013.
[10]PremJeya Kumar M., Sandeep Anand J.,
Gopalakrishnan K., Satheesh B., Anbazhagan R.,
Computer modelling of a vehicle system, Indian Journal
of Science and Technology, v-6, i-SUPPL5, pp-4620-
4628, 2013.
[11]Ganeshram V., Achudhan M., Synthesis and
characterization of phenol formaldehyde resin as a binder
used for coated abrasives, Indian Journal of Science and
Technology, v-6, i-SUPPL.6, pp-4814-4823, 2013.
[12]Achudhan M., PremJayakumar M., Mathematical
modeling and control of an electrically-heated catalyst,
International Journal of Applied Engineering Research,
v-9, i-23, pp-23013-, 2014.
[13]Anbazhagan R., Satheesh B., Gopalakrishnan K.,
Mathematical modeling and simulation of modern cars in
the role of stability analysis, Indian Journal of Science
and Technology, v-6, i-SUPPL5, pp-4633-4641, 2013.
[14]PremJeya Kumar M., Gopalakrishnan K., Srinivasan
V., Anbazhagan R., SundeepAanan J., PC modeling and
simulation of car suspension system, Indian Journal of
440
Special Issue
Science and Technology, v-6, i-SUPPL5, pp-4629-4632,
2013.
[15]Jeykar K., Srinivasan V., Performance characteristics
of twin cylinder Di diesel engine operated with three
different non edible vegetable oil blends with diesel,
International Journal of Applied Engineering Research,
v-9, i-22, pp-7601-7607, 2014.
[16]Srinivasan K., Gopikrishnan M., Analysis of a
reduced switch three phase BLDC drive, International
Journal of Applied Engineering Research, v-9, i-22, pp-
6633-6637, 2014.
[17]Selvam M.D., Srinivasan V., Sekar C.B., An attempt
to minimize lubricants in various metal cutting processes,
International Journal of Applied Engineering Research,
v-9, i-22, pp-7688-7692, 2014.
[18]Arun Kumar N., Srinivasan V., Krishna Kumar P.,
Analysing the strength of unidirectional fibre orientations
under transverse static load, International Journal of
Applied Engineering Research, v-9, i-22, pp-7749-7754,
2014.
[19]Srinivasan V., Analysis of static and dynamic load on
hydrostatic bearing with variable viscosity and pressure,
Indian Journal of Science and Technology, v-6,
i-SUPPL.6, pp-4777-4782, 2013.
[20]Praveen R., Achudhan M., Optimization of jute
composite as a noise retardant material, International
Journal of Applied Engineering Research, v-9, i-22, pp-
7627-7632, 2014.
[21]Srinivasan V., Optimizing air traffic conflict and
congestion using genetic algorithm, Middle - East Journal
of Scientific Research, v-20, i-4, pp-456-461, 2014.
[22]Raja Kumar G., Achudhan M., Srinivasa Rao G.,
Studies on corrosion behaviour of borated stainless steel
(304B) welds, International Journal of Applied
Engineering Research, v-9, i-22, pp-7767-7772, 2014.
[23]Ganeshram V., Achudhan M., Design and Mold flow
analysis of piston cooling nozzle in automobiles, Indian
Journal of Science and Technology, v-6, i-SUPPL.6, pp-
4808-4813, 2013.
[24]Udayakumar R., Kaliyamurthie K.P., Khanaa,
Thooyamani K.P., Data mining a boon: Predictive system
for university topper women in academia, World Applied
Sciences Journal, v-29, i-14, pp-86-90, 2014.
[25]Kaliyamurthie K.P., Parameswari D., Udayakumar
R., QOS aware privacy preserving location monitoring in
wireless sensor network, Indian Journal of Science and
Technology, v-6, i-SUPPL5, pp-4648-4652, 2013.
[26] T. Padmapriya and V. Saminadan, “Inter-cell Load
Balancing technique for multi-class traffic in MIMO-
LTE-A Networks”, International Journal of Electrical,
Electronics and Data Communication (IJEEDC), ISSN:
2320- 2084, vol.3, no.8, pp. 22-26, Aug 2015.
441
442