and provides an opportunity to train
and rehearse the inspection (Figure 5).
Recommended Practices for
Robotics-Based Remote Visual
Inspection
Close visual inspection is a top priority
for robotic applications, but there are
discussions about whether robotics-
based remote visual inspection (RVI)
can fully replace close visual inspection
(CVI) performed by a human. Several
RVI limitations have been identified,
including the robot’s distance from
the inspection surface, limited viewing
angles, lack of tactile feedback, absence
of surface preparation or deployment
of inspection aids, and challenges with
artificial lighting. Due to these limita-
tions, it is advised not to claim robotics-
based RVI as a complete replacement for
human CVI. Instead, robotic inspection
should complement conventional CVI
by identifying areas that require further
examination.
Standards such as ASME V Article 9
[6] and BS EN 17637 [7] specify spatial
resolution requirements for CVI and
direct visual inspection (DVI), typi-
cally around 3 line pairs per millimeter
(lp/mm) under optimal viewing con-
ditions, based on human eye acuity.
Although ASME V Article 9 also ref-
erences the visibility of fine lines, this
is not considered a reliable measure
of spatial resolution. To comply with
ASME V Article 9, robotics-based RVI
images should demonstrate a spatial
resolution of approximately 3 lp/mm,
equivalent to that of CVI and DVI.
The “HOIS Guidance on Image
Quality for UAV/UAS–Based External
Remote Visual Inspection in the Oil
&Gas Industry” [5] provides detailed
guidance on maintaining image
quality during uncrewed aerial vehicle
(UAV) inspections within the oil and
gas sector. Its goal is to ensure that
the images obtained are of sufficient
quality for engineering assessments
of component integrity, aiding asset
operators in making critical decisions
about continued operation. While the
HOIS guidance focuses exclusively on
FEATURE
|
ROBOTICVT
New assets
3D CAD data
Asset in 3D Planning &simulation Inspection &reporting
Old assets
3D CAD data
Universal
2D drawing/
sketch available
Existing CAD
3D asset builder
3D dynamic reconstruction
• Feasibility check
• Mission &inspection planning
• Training &rehearsal
Planning
Simulation
3D digital twin
• 3D spatial awareness at any time
• Geotagging of all data
• 100% repeatablility
• No risk for inspectors
• Automated reporting
Mission execution Mission execution
3D digital twin
Figure 5. Integrating planning and simulation, using the 3D virtual representation of the asset and the kinematic representation of the robot with
the camera and the cables.
Thermal impact?
Check with operations Thermal impact?
Check with operations
Figure 4. 3D digital twin: (a) editor to amend and complete findings and recommendations,
possibility to annotate inspection data (b) automatically generated inspection report.
52
M A T E R I A L S E V A L U A T I O N • J U L Y 2 0 2 4

image quality, it does not address safety
and operational aspects of UAV deploy-
ment, which are covered in separate
publications.
The same guidelines for an UAV-
based visual inspection can be applied
to robotic crawlers.
The guidance identifies three priority
applications for UAV and robotics-based
RVI among members of the HOIS orga-
nization [8]: achieving CVI resolution,
assessing coatings to ISO 4628 stan-
dards, and inspecting flare tips/stacks.
While both still images and videos are
considered, the document places more
emphasis on still images, as they are typ-
ically more common in final inspection
reports.
Specific guidance is also provided
regarding spatial resolution require-
ments for each of the priority applica-
tions, along with methods for verifying
that the achieved resolution meets these
standards. Additionally, the importance
of image signal-to-noise ratio (SNR) is
highlighted as a critical quality criterion,
with recommendations for minimum
SNR values and maximum ISO settings
for cameras. Information on these
settings can often be obtained from
resources such as the DxOMark website
or estimated based on the camera’s
sensor element area.
General advice covers various
aspects of UAV and robotics-based RVI,
including considerations for viewing
direction, ambient light levels, and
camera settings. It also addresses file
formats and post-processing software for
both still images and videos.
Overall, the document serves as
a comprehensive guide for ensuring
adequate image quality in UAV-based
RVI within the oil and gas industry. It
offers specific recommendations for key
quality criteria and priority applications
while providing general guidance on
related aspects.
Experimental Validation
To validate the technical capabilities
of both robotics and pole cameras for
confined space inspection, we conducted
an extensive visual examination of a test
vessel. In addition to visual inspection,
we took ultrasonic thickness readings at
designated spots on the hull and con-
ducted 3D surface scans on sections
affected by corrosive pitting. All this data
was geotagged (localized) in a digital twin
optimized for inspection, which was built
from customer drawings (Figure 6).
The test was conducted using an
ultra-mobile robotic platform that allows
it to climb over obstacles [8]. The robot
is equipped with a visual inspection
camera, an ultrasonic probe, and a struc-
tured white light–based surface scanning
system. Utilizing 3DLOC technology, it
can calculate the robot’s pose within the
vessel and geotag the images to the 3D
virtual model (as described previously in
the “Key Technology: Localization and
Data Geotagging” section).
To assess image quality, an USAF
1951 resolution chart was utilized within
the vessel, with measurements taken
from a distance of 1.8 m. Figure 7 depicts
the camera’s capabilities, serving as an
example of the output obtained from the
localization data, images, and other key
notes from the inspection. Typically, a
report of this nature would include:
Ñ a picture captured with the HD camera
Ñ the coordinates of the robot within the
3D model
Ñ the coordinates of the camera hit point
on the surface
Ñ a screenshot of the crawler’s position
and stance at the time the image was
captured
Ñ descriptions and recommendations as
necessary.
Figure 6. Digital twin created from asset drawings: (a) photo of asset (b) digital twin.
Figure 7. USAF test chart at 1.8 m distance and
typical reporting structure.
J U L Y 2 0 2 4 • M A T E R I A L S E V A L U A T I O N 53