robo-spect robotic system control for news · robo-spect robotic system at the research tunnel ......
TRANSCRIPT
4rth ROBO-SPECT Newsletter, November 2015
ROBO-SPECT
ROBO-SPECT
NEWS
In This Issue
Editorial…………………… 1
The First Evaluation of the
ROBO-SPECT Robotic
System in the Research
Tunnel Infrastructure of VSH
in Switzerland..………….... 2
The ROBO-SPECT Work-
shop in London on Novem-
ber 5, 2015………..… 3
News and Events………... 6
Consortium……………….. 8 ROBO-SPECT
Editorial Welcome to this ROBO-SPECT's fourth newsletter.
ROBO-SPECT is a European 7th Framework project funded under the ICT
programme on Robotics Use Cases (contract No. 611145), implemented by
10 partners from 6 European countries.
The main objective of the project consortium is to provide an automated,
speedy and reliable tunnel inspection and assessment solution that can
combine in one pass both inspection and detailed structural assessment
and that only minimally interfere with tunnel traffic. The proposed robotic
system has been evaluated at the research infrastructure of VSH in
Switzerland, and will be further evaluated at London Underground and at
the tunnels of Egnatia Motorway in Greece.
The project officially launched its activities in October 2013. During the first
year of the project the partners identified an extended group of end-users,
derived user requirements and specifications and based on these designed
the first prototypes of all components of the proposed robotic system.
Inside this issue you will find the results of the first evaluation of the
ROBO-SPECT robotic system at the research tunnel infrastructure of VSH
in Switzerland, as well as information on the ROBO-SPET workshop on
November 5, 2015. Additionally, in this issue you will find information on the
consortium's participation in conferences and other events related to the
project's research field.
Angelos Amditis
Project Coordinator
PROJECT FACTS
Duration:
1 October 2013 –
30 September 2016
Total Cost:
4.592.196,00€
EU contribution:
3.306.599,00€
www.robo-spect.eu
ROBOtic System
with Intelligent Vision and
Control for
Tunnel Structural INSPECTion
and Evaluation
Issue 4 November 2015
1st ROBO-SPECT Newsletter, May 2014 Page 2 4rth ROBO-SPECT Newsletter, November 2015
The ROBO-SPECT project is combining recent advances in robotic systems and sensors, computer vision, ultrasonic sensing, as well as laser scanning in an integrated manner to achieve structural assessment of reinforced concrete linings in transportation tunnels in a semi-autonomous way.
The robotic platform of the system is the main system vehicle incorporating all hardware and software systems. It is responsible for the autonomous navigation and positioning of the robotic system as well as crane and robotic arm movements.
The computer vision system is responsible for the crack and other defects detection on the surface of the tunnels, calculation of the crack coordinates as well as the creation of 3D model of the area of interest. A 3D laser scanner is also used to create a 3D representation of the tunnel slice of interest to detect any deformations of the tunnel perimeter.
An ultrasonic sensing device is used to provide very accurate measurements of the identified cracks with a sub-millimeter precision.
A structural assessment software (not evaluated in this lab test) compiles all sensor results and performs the overall tunnel structural assessment and also incorporates the tunnel operator interface.
First Evaluation of the ROBO-SPECT Robotic System at the Research Tunnel Infrastructure of VSH in Switzerland
Robotic Platform
The first tests of the integrated ROBO-SPECT system
were executed on the week 20-24 July 2015 at the Re-
search Infrastructure Tunnels of VSH (consortium part-
ner). VSH is fully owned by the Amberg Group – lead-
ing producers of mobile tunnel inspection equipment.
During the last 40 years, VSH has developed – on a
private base – an underground research infrastructure
of tunnels totaling about 5.1 km in length. These tun-
nels, that have a height of 5.5 meters, include sections
with cast in place concrete linings some of which are
damaged.
ROBO-SPECT started evaluating and benchmarking
their robotic system at the above road,
cast-in-place concrete, damaged and undamaged sec-
tions of the VSH tunnels. For this, version1 of the inte-
grated system was transported to VSH tunnels
(Switzerland).
All system components were tested in a semi-
integrated and individual scenarios into actual parts of
the VSH tunnel network and also over damaged testing
concrete slabs that were created by the VSH team. The
evaluation and benchmarking included the performance
of the navigation and positioning system, the vision sys-
tem, the ability to control the arm, stability, the sensing
system, the speed as a function of damage and the
structural assessment models. First results of each
submodule testing can be found below:
4rth ROBO-SPECT Newsletter, November 2015
ROBO-SPECT
Robotic Platform Navigation and
Positioning
At first, all the safety lasers were
placed and reprogrammed under
operating circumstances, in order
to allow the collision avoidance
tool for the movement of the
platform. The axis PTZ frontal
camera was also reconfigured in
order to offer real time images
through the WiFi network, which
were used for remotely controlling
the platform and supervising the
operation.
Tests were performed with the
movement and positioning of the
crane, revealing technical issues to
be revised. The general
performance of the movement was
also tested and led to some in-situ
adjustments to the low level
controllers. Communications and
interfaces for the mission controller
within the robotic system were also
revised.
Robotic Arm Testing
The tests regarding the robotic arm integrated with the complete system
included:
Communication between robotic arm and other subsystems: The
communication between the arm software and the ICCS vision system
and ROB mobile vehicle was achieved using an internal network and
the YARP libraries. The first tests on communications consist in input
and output client/service messages through TCP/IP ports. This
mechanism will be used to transmit the information during the
autonomous inspection.
Hardware testing: The robotic arm complete system fits well on the
crane platform. The movements of the arm were tested in a
tele-operated way and the ultrasonic sensors were placed successfully
in contact with the wall (see “Ultrasonic System” below). The range laser
attached to the robotic arm operates correctly and the DC power supply
installed on the crane platform provides the necessary power.
Software testing: The trajectory generator system developed for the arm
was tested on a tunnel-lining sample provided by VSH. The system was
able to scan the surface of the wall and position the robotic tip at a
constant distance to the wall with normal orientation.
Robotic Arm System during VSH tests. Note the scanned surface and the
robot state in the monitor.
1st ROBO-SPECT Newsletter, May 2014 Page 4 4rth ROBO-SPECT Newsletter, November 2015
Crack Detection Computer Vision System The robotic system crack detection system was
successfully tested by ICCS at both the actual VSH
tunnels (pre casted concrete sections) and also the
testing slabs constructed by VSH team.
The system was tested over visible cracks and was
able to perform crack detections even in areas with
multiple cracks on the surface.
Different moving and lighting conditions were
successfully tested giving significant feedback to the
future versions of the computer vision algorithms.
Tunnel Defects Detection System The Tunnel Defect Detection System was tested
mainly on two locations of the VSH tunnel one at the
main VSH workshop and another at a location inside
the VSH tunnel (with pre-casted concrete blocks).
The experiments were conducted under varying
luminance (manually controlled from a light source
installed on the robot platform) and intensity values.
Images captured inside the VSH tunnels were using
the cameras from the robotic platform. The defect
detection model, which was previously trained with a
database consisting of images captured from an auto-
focus camera and diffused background lighting, it was
now tested on images captured by a camera with
manual focus and different background lighting.
Despite the different conditions, the pre-trained model
was tested successfully and gave a high enough
accuracy. Furthermore, a retraining of the model with
original images captured from the same camera
settings and lighting conditions as in the actual tunnel
scenario can be used to further enhance its
robustness.
One crack detection results
Spalling detection results
4rth ROBO-SPECT Newsletter, November 2015
ROBO-SPECT
Laser Scanning System
The 3D laser scanner of the system was also tested by ICCS at many positions inside the VSH testing tunnels
network. The laser scanner was triggered by the robotic platform and so the messaging system was also tested.
The 3D laser scanner successfully captured slices of the positions of interest inside the tunnel that were locally
stored for further processing.
Ultra Sonic System
On-field testing of the ultrasonic sensor system for crack characterization was performed in VSH facility on cracks
artificially produced on a two-meters high sample of tunnel lining with length of several meters, suitable for
emulating real working conditions for the ultrasonic sensor system. In these tests, the robot arm was brought close
to the lining and brought in contact with it using the four-point supports of the ultrasonic sensor module. The
procedure was completed successfully and vibrations during arm movement seemed not to interfere with the preci-
sion of the placement and the effectiveness of the crack depth measurements.
Laser Scanner in road tunnel and VSH tunnel (slices)
Testing of the ultrasonic sensor system on tunnel lining sample in VSH facility.
1st ROBO-SPECT Newsletter, May 2014 Page 6 4rth ROBO-SPECT Newsletter, November 2015
ROBO-SPECT Workshop on 'Ageing Tunnels: Inspection and Assessment' on November 5, 2015, in London The aim of this workshop was to:
share information on the current needs and issues in tunnel
structural inspection from several end users and technology
providers
present the first ROBO-SPECT results
discuss some of the challenges the tunnel structural
inspection is facing and receive feedback on the
ROBO-SPECT research.
Fourteen representatives of end users attended the meeting.
They represented the following organizations:
London Underground,
Crossrail,
CH2M,
SITAF,
Attico Metro,
Egnatia Odos,
VSH,
Aegean Motorways
In the first part of the workshop the partners presented their
results in the ROBO-SPECT project.
Following this, the partners learned about current needs and
issues in tunnel inspection from end users and received feedback
about the ROBO-SPECT research.
Regarding needs for a robotic system for tunnel inspection the
end users mentioned:
a. In case of metros: To be able to get the robot to location it
should be possible to assemble it at the level of the train rails.
Moreover, inspection should finish in no more than 3 hours so
that trains can start running again.
b. In a future version: It would be desirable to have a com-
puter vision system that is operational while moving. Further-
more, to add to the robot spring loaded hammers.
IROS 2015 in Hamburg from Sept 28 to Oct. 2
ROBO-SPECT attended the IROS 2015
conference and supported the ROBO-
SPECT dissemination activities through a
booth at the conference exhibition. The
conference was attended by more than
2000 attendees from various fields of
expertise such as robotics, computer
vision, sensing systems, navigation and
other proving very relevant to the ROBO-
SPECT activities. ROBO-SPECT was
disseminated through many discussions
with attendees and booth followers
through leaflets, posters and newsletters.
ROBO-SPECT evaluation Video
A video has been produced
on the ROBO-SPECT
evaluation at the VSH tunnel
research infrastructure and
can be seen at youtube.
1st ROBO-SPECT Newsletter, May 2014 Page 7 4rth ROBO-SPECT Newsletter, November 2015
News and Events
ROBO-SPECT Presentations/Publications During the Last Six Months
Makantasis, K., Protopapadakis, E., Doulamis, A.D., Doulamis, N.D., Loupos, C., "DEEP CONVOLUTIONAL NEURAL NETWORKS FOR EFFICIENT VISION BASED TUNNEL INSPECTION," IEEE International Confer-ence on Intellignence, Communication, and Processing, 3-5 September, Cluj-Napoca, Romania (2015)
Eftychios Protopapadakis and Nikolaos Doulamis - IMAGE BASED APPROACHES FOR TUNNELS' DEFECTS RECOGNITION VIA ROBOTIC INSPECTORS," International Symposium on Visual Computing, Las Vegas, USA, December 2015
K. Loupos, A. Amditis, C. Stentoumis, P. Chrobocinski, J. Victores, A. Roncaglia, S.Camarinopoulos, N. Ko-modakis, R. Lopez - ROBOTIC SYSTEM WITH INTELLIGENT VISION FOR TUNNEL STRUCTURAL AS-SESSMENT - SYSTEM ARCHITECTURE - THE ROBO-SPECT-EC PROJECT – Third Conference on Smart Monitoring, Assessment and Rehabilitation of Civil Structures, SMAR 2015, 7-9 September 2015, Antalya, Tur-key, ISBN: 978-3-905594-65-2
K. Makantasis, E. Protopapadakis, N. Doulamis, A. Doulamis, K. Loupos - DEEP CONVOLUTIONAL NEURAL NETWORKS FOR EFFICIENT VISION BASED TUNNEL INSPECTION, 11th International Conference on Intel-ligent Computer Communication and Processing, IEEE ICCP 2015, 2-5 September 2015, Cluj-Napoca
C. Stentoumis, A. Amditis, G. Karras - CENSUS-BASED COST ON GRADIENTS FOR MATCHING UNDER ILLUMINATION DIFFERENCES, International Conference on 3D Vision (3DV 2015), Lion, October 2015
C Stentoumis, E. Karkalou, G. Karras - A review of penalty functions in semi-global matching Evaluation under radiometric differences - 11th International Conference on Intelligent Computer Communication and Pro-cessing, IEEE ICCP 2015, 2-5 September 2015, Cluj-Napoca
K. Makantasis, A. Doulamis, K. Loupos - VARIATIONAL INFERENCE FOR BACKGROUND SUBTRACTION IN INFRARED IMAGERY, 11th International Symposium on Visual Computing (ISVC 2015), 14-16 December 2015, Las Vegas, Nevada, USA
Loupos, C. et al. 'Robotic System with Intelligent Vision for Tunnel Assessment - System Architecture - The ROBO-SPECT EC Project,' Third Conference on Smart Monitoring, Assessment and Rehabilitation of Struc-tures, 7-9 Sept. 2015, Andalusia, Spain.
N. Doulamis, A, Doulamis, K. Makantasis, K. Karantzalos, K. Loupos. 'Thermal Imaging for Active Evacuation Routes' 8th ACOM International Conference on Pervasive Technologies Related to Assistive Environments, 1-3 July 2015, Corfu, Greece
K. Loupos et al. 'Robotic System with Intelligent Vision for Tunnel SHM - ROBO-SPECT EC Project,' Structural Health Monitoring of Intelligent Infrastructure (SHMII), July 1-3, 2015, Torino, Italy.
R. Montero, J.G. Victores, S. Martinez, A. Jardon, C. Balaguer 'Past, Present and Future of Robotic Tunnel In-spection' Automation in Construction, vol.59, 2015, pp.99-112
S. Zagoryuko and N. Komodakis, 'Learning to Compare Image Patches via Convolutional Neural Networks,' CVPR 2015, June 8-10, 2015, Boston, USA
W. Zou and N. Komodakis, 'HARF: Hierarchy-Associated Rich Features for Salient Object Detection,' ICCV 2015, Dec. 13-16, Santiago, Chile
1st ROBO-SPECT Newsletter, May 2014 Page 8 4rth ROBO-SPECT Newsletter, November 2015
Contact Us Project Coordinator:
Angelos Amditis ([email protected])
Dissemination Manager:
Stephanos Camarinopoulos ([email protected])
Visit us on the web at www.robo-spect.eu
Join us on robo_spect Twittter group
Join us on ROBO-SPECT LinkedIN group
Consortium
End users that participate as partners, associate partners, subcontractors or members of an
extended users' group that has been formed are: EUROTUNNEL, London Underground,
RATP (Parisian metro), Attico Metro (Athinean metro), Harcrow Group Ltd., Egnatia Metorway
S.A., Aegean Motorway S.A., SITAF, and CETU (in the French Public Roads Administration).
END USERS IN THE GROUP
This project has received funding
from the European Union’s
Seventh Framework Programme
under grant agreement no
[611145]
Institute of Communications and Computer Systems (ICCS)
AIRBUS DS
University Carlos III, Madrid Dpto. De Ingeniera de Sistemas y Automatica Robotics Laboratory(UC3M)
VSH Hagerbach Test Gallery Ltd. (VSH)
Egnatia Motorway S.A. (EOAE)
Institute of Microelectronics and Microsystems, National Research Council of Italy (CNR)
RISA Sicherheitsanalysen GmbH (RISA)
Techniche e Consulenze Nell’ Ingegneria Civile S.p.A - Consulting Engineers S.p.A (TECNIC)
D. Bairaktaris & Associates Structural Design Office Ltd. (DBA)
Ecole des Ponts ParisTech (ENPC)
ROBOTNIK (ROB)