Skydio Volume 6 Insurance Appraisal

EBook
Enterprise Drone Inspection:

Picking the Right Platform
Volume 6
High-Precision Insurance Appraisal:
How to get decision-quality data from your drone program
By Guillaume Delepine, Senior Product Marketing Manager
Enterprise
© 2020-1 Drone
Skydio, Inc. Inspection: Picking the Right Platform - Volume 3
All rights reserved. © 2020-1 Skydio, Inc. All rights reserved. 1
AUTHORs
Guillaume Delepine
Senior Product Marketing Manager at Skydio
Guillaume Delepine is a Senior Product Marketing Manager at Skydio, the leading American drone
manufacturer and world leader in autonomous flight. In his tenure at the company, he previously led
Skydio’s efforts to serve first responders as Public Safety lead. Then, as Enterprise Strategy Manager,
Guillaume contributed to the fundraising, planning, and hiring process that led to Skydio’s recent
announcements of a Series D fundraise at a valuation of over $1 billion USD, expanded roadmap, and
growing executive team. Guillaume holds a BA from Princeton University, and took leave of the joint
MPP/MBA program at Harvard University to join Skydio.
Brian Richman
Senior Product Manager at Skydio
Brian Richman is an Enterprise Product Manager at Skydio with almost 10 years of experience in
drones. Brian works on realizing Skydio’s vision for automated drone inspection and mapping by
enabling levels of safety, precision, and data quality beyond what even the best drone pilots in the
world are capable of. Before Skydio, Brian was the product lead for Airware’s Mobile business unit,
launching a pioneering end-to-end commercial drone configuration and flight software. Brian is also
a former R&D team lead for the RQ-23 Tigershark program, with 700+ combined flight hours as a
pilot, payload operator, and mission commander.
Russell Bondi
Image Quality Test Engineer at Skydio
Russell graduated from University California Santa Cruz with a major in Photography. He spent the
majority of his time there working on digital photography from his weekend job at Skydive Hollister.
His love for action sport photography and his understanding of subjective review led him to a career
at GoPro, where he was the lead Image Quality Test Engineer on the Hero 6 and 7. His role was
to uncover all image quality issues and help find the path to a fix. Working closely with ISP Tuning
Engineers, as well as Firmware Engineers, to get products out the door in the best shape possible.
When Russell is not working, he spends the majority of his time outdoors either surfing, mountain
biking, skydiving, or snowboarding.
Enterprise Drone Inspection: Picking the Right Platform - Volume 3 © 2020-1 Skydio, Inc. All rights reserved. 2
Table of Contents
Prologue 3
Data Quality Requirements for Key Use Cases 4
Meeting the Bar for Automated Roof Inspection 6
Skydio Solutions for Inspection 11
Conclusion and Looking Ahead 14
Prologue
The most forward-thinking insurance companies are turning to drones to accelerate and improve their appraisal
processes. With the growing number of enterprise drone options on the market today, operators face an ever-growing
number of choices. Critically, the rise of autonomy offers an entirely new way to compare drones to each other. While the
previous generation of manual drones were all equally limited to 2D waypoint or manual flight, autonomous drones can
perform inspection missions with entirely different flight paths. These closer, more adaptive flight patterns render the old
ways of comparing drones to each other based on rudimentary camera specifications, like the number of megapixels,
outdated.
This paper is the sixth volume of Skydio’s eBook: Enterprise Drone Inspection: Picking the Right Platform, a collection
of white papers that will help enterprise drone operators from various industries make more informed decisions when
choosing their enterprise drones. In this installment, we use 3D model examples and rigorous camera benchmarking to
help roof inspectors understand the tradeoffs of various drone systems, and the impact those tradeoffs will have on the
datasets their uncrewed aerial vehicles (UAVs) can generate.
We hope that this analysis will help your drone program make its decisions based on the quality of the drone’s outputs,
so that you know the end product you will be able to generate as you invest in your next drone system.
Data Quality Requirements for Key Use Cases
Different inspection programs are likely to have different data quality requirements. The true requirements for
a dataset come from the consumer of the data—whoever needs to receive an inspection report and either
certify approval or take corrective action. To provide context as operators assess the level of quality that will
be required to scale their drone inspection programs, we have assembled key requirements from anonymized
sets of customers in the various industries that have adopted drones for inspection.
Skydio’s Solutions Engineering team works with drone programs day in and day out to understand and
document their requirements and make sure our customers end up with the best possible drone operations.
They have been instrumental in helping to document the requirements that are typical for various industries.
While this list is not exhaustive, the industries this eBook will focus most on are:
Common
Typical Use Artifacts to Precision
Industry Asset to Scan
Cases detect Requirement
(GSD)
Construction Primary: Progress Construction Asset damage, Up to 1 cm

Assessments sites, equipment misalignment vs.
inspections tolerance across
Additional: subcontractors
• Inventory
management
(Volumetrics)
• Site Survey
Transportation Primary: Bridge Bridges, Roads, Cracks and Sub-millimeter to

Inspection Railroad Tracks delamination in 5 mm
concrete, bolts,
Additional: nuts, connection
• Inventory points in weight
management bearing areas
(Volumetrics)
• Road grading &
survey
Utilities Primary: Power plants, Rust and loose Up to 0.5 cm

Transmission Transmission Lines, bolts (Primarily photos,
& distribution Oil Pipelines not models)
inspection
Additional:
Long linear
inspection for
wildfire risk
Common
Typical Use Artifacts to Precision
Industry Asset to Scan
Cases detect Requirement
(GSD)
Telecom Primary: Digital Cell Towers Labels on antenna < 2 mm

Twins for Remote arrays, antenna
Management orientation,
structure damage
Insurance Primary: Roof Homes Hail damage < 1 mm

Inspection
Focus of this volume
First Responder Primary: Crime or accident Crumple damage, ~ 1 mm

Collision scene scenes skid marks,
reconstruction weapons, broken
glass
Additional: Crime/
Arson investigation
In this Volume, we will take a deep look at the technologies and outputs available to drone roof inspectors,
using both customer experiences and rigorous camera benchmarking to make sure operators in the industry
have all the data they need to make the right decisions.
Meeting the Bar for Automated Roof Inspection
Residential insurance claims inspectors frequently need to rapidly evaluate damage to rooftops caused by
falling objects or severe weather events. The damage can range from hail damage to full structure cave-
ins. When detecting claims across a nation-wide or international insured base, accuracy is critical. There
is such a high volume of jobs, that systematic error can lead to large sums in over- or under-paid claims.
Furthermore, messing up an inspection on the first pass can require an inspector to commute to the site
again and have to refly the scene — a crippling expense for the inspection program.
As drone inspection and modeling solutions improve, it has become more efficient to perform a drone
inspection that yields a useful 3D model than it previously was to have inspectors carry ladders to scenes
and climb up on them to take a small selection of cell phone pictures. As a result, insurers and their service
providers, such as EagleView, are increasingly training their inspectors to use drones to generate these
models, so they can get better results in less time. For these models, precision is paramount, and even the
most detailed drone inspections are more time-efficient than traditional ground-based techniques (not
to mention, they don’t require the inspector to lean on, and often, walk on the roof they are inspecting for
miniscule damage).
Key requirements for Remote Asset Management
Resolution < 2 mm
Photo Quality Consistent exposure to capture areas backlit by the sun
Photo Location 80%/80% overlap/sidelap. Ideal pattern is consistent orbits at

evenly spaced altitudes around the tower
Relative and Absolute Accuracy Only relative accuracy typically required
Conducting good inspections requires high quality imagery. EagleView is a Skydio customer and partner
using our HouseScan product. They use a default distance offset of 1.25m, which corresponds to about
0.55mm GSD, so that they can detect even the smallest types of damage. Their team is enthusiastic about
the level of quality they will be able to provide from their inspections:
“EagleView is excited to partner with Skydio to deliver breakthrough innovation for residential roof
inspection that puts autonomous drone technology and unparalleled image quality to work for our
customers,” said Chris Jurasek, CEO of EagleView. “EagleView’s images are already 16 times clearer
than those of satellites, and with Skydio’s drones you’ll be able to not only spot a dime on the roof but
see the face of Franklin D. Roosevelt stamped on it. This partnership allows customers to confidently
make decisions while lowering costs.”
We put that statement to the test, and our Image Quality team, led by Russell Bondi, is excited to share the
results below, which can help our readers understand the differences in resolution that come from various
cameras and standoff distances. Note that the two cameras here are both 12MP cameras, but the field of
view makes all the difference in the ground sampling distance.
Skydio 2 - 2m fixed focus Skydio X2 - 2m fixed focus

Camera 12 MP HFOV 81° 12 MP HFOV 45°
1.25 m 0.52 0.25

GS (mm/pix)
0.52 0.25
1.5 m 0.63 0.31

GS (mm/pix)
0.63 0.31
2m 0.84 0.41
GS (mm/pix)
0.84 0.41
A comparison of resolution and photo quality at various standoff distances with two different 12MP cameras that have
fixed focus lengths at 2 meters. Note that these are illustrative comparisons, not representative of Skydio products.
Source: Skydio.
Camera Performance on Text (Eye Chart)
Ground Sampling
Skydio 2 DJI Phantom 4 Pro V2.0 DJI Mavic 2 Pro
Distance (mm/pixel)
0.25 mm 0.59 meters from subject 0.92 meters from subject 1.05 meters from subject
1 mm 2.37 meters from subject 3.67 meters from subject 4.21 meters from subject
This figure compares the camera systems on the Skydio 2, DJI Phantom 4 Pro V2.0, and DJI Mavic 2 Pro drones, at four
levels of ground sampling distance (GSD). The distance from the subject that corresponds to the GSD is unique to each
camera and indicated in each cell. Photos are cropped to the same frame to enable comparisons of which camera
generates fine detail. Note that drones were hand-carried for the study - while Skydio 2’s trustworthy obstacle avoidance
system enables up close flights, capturing imagery at short standoff distances is risky with DJI drones. Source: Skydio
The comparison above reveals several interesting insights that matter to teams looking to choose the right
camera system for their inspection needs. It is important for customers to compare cameras at equivalent
GSDs, because this can be controlled for by flying closer or farther away. So we control for GSD such that
readers can gain insight on how the quality of the camera tuning will impact the photo outputs available to
their teams. Our image quality team notes:
Skydio 2 photo taken at 0.25mm ground sampling distance DJI Phantom 4 Pro V2.0 photo taken at 0.25mm ground
sampling distance
Tone Mapping
Even at the same ground sampling distances, the Skydio photos retain more photo detail in fine
textures to be able to see slight aberrations and the lines between the individual shingles. Comparing
the auto modes in both cameras, both DJI drones overexpose the shingles and lose out on the fine
details of the scene. In fairness, DJI cameras have additional features like touch-to-expose that can
help with these issues. However, for work at scale, having a default setting that adapts to a wide
range of scenes can help inspectors focus on getting the job done instead of tuning their camera
uniquely for every job.
Operators should note that even with a smaller sensor, the Skydio 2 can provide better photos.
Skydio Solutions for Inspection
After reading the previous chapters, it is our hope that you feel more equipped to make a decision regarding
your selection of drones. As you take this knowledge into the field to evaluate the available technologies,
below is a concise summary of the components that make up the Skydio solution for industrial asset
inspection. Your Skydio sales and solutions engineering teams can help you determine which options are
right for your program and make a plan to implement them.
Skydio 2
An entry-level drone to introduce autonomy to
inspection workflows. Features a 12MP camera,
and six 4K color sensors used to support Skydio
Autonomy enabling true 360° obstacle avoidance
in every situation and up to 23 minutes of flight
time. Backpack portable and easy-to-use, Skydio
2 can be provided to any inspector to start taking
advantage of aerial data.
Skydio 2
Skydio X2E
Pairs the breakthrough Skydio Autonomy™ engine
with a ruggedized airframe that features a color
or optional dual color/thermal sensor, long-range
operations, and extended battery life for up to
35 minute flight time. Core autonomy capabilities
include 360° Obstacle Avoidance, Object and Scene
Recognition, and Skydio Visual Navigator.
Skydio X2E
Skydio Enterprise Controller

Ground control for X2E is enabled via the Skydio
Enterprise Controller, which was designed from the
ground up for pilots with demanding operating
requirements. Ground control software is natively
delivered via the Skydio Enterprise App.
Skydio Enterprise Controller
Skydio Autonomy Enterprise Foundation
An add-on software package that augments the
core autonomy engine. It’s designed to assist the
pilot through software capabilities that enhance
flight control in obstacle-dense environments. Key
features of this package include Close Proximity
Obstacle Avoidance, Vertical View, and Visual
Return to Home.
Skydio Autonomy Enterprise Foundation
Skydio 3D Scan™
The first-of-its-kind adaptive scanning software built
on top of Skydio Autonomy. 3D Scan allows the drone
to automate the data capture process needed to
generate 3D models with comprehensive coverage
and ultra-high resolution, so that crews can perform
higher quality scans in less time and with minimal
pilot training.
Skydio 3D Scan
Skydio Cloud
Skydio Cloud serves as a foundation for autonomous connected flight operations as well as API-based
integrations with our partner ecosystem. It includes Skydio Fleet Manager and Skydio Media Sync, which help
distributed fleets manage their operations and data more effectively than they can with spreadsheets and
SD cards. Skydio Streaming and Remote Ops are both coming soon, and allow for real-time monitoring and
execution of drone operations through the cloud.
Skydio Academy
Delivered online, or in-person, in both self-paced and instructor-led configurations, Skydio Academy
provides flexible options for your pilots to achieve the Skydio Professional Operator (SPO) and Skydio Expert
Operator (SEO) certifications that can help a program manager manage and track core competencies
across a distributed pilot fleet.
Conclusion and Looking Ahead
The data product is the most important output of a drone program, and producing a useful dataset can
help a drone program reach scale within a large organization full of consumers of that data. Therefore, we
encourage drone operators to work with their downstream colleagues to make sure that the drones they
purchase can produce sufficient datasets to meet the precision requirements. The tools in this white paper
can help operators look beyond the specifications of the hardware, and instead focus on the outputs the
system can generate. Choosing in this way can greatly improve an organization’s chance of success.
In the next installment of this eBook, we will be taking yet another deep dive using similar analytic techniques,
to help first responders select the right drones for the job. Stay tuned for the next volume!
sales@skydio.com

Skydio Volume 6 Insurance Appraisal

Uploaded by

Copyright:

Available Formats

Skydio Volume 6 Insurance Appraisal

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Skydio Volume 6 Insurance Appraisal

Uploaded by

Copyright:

Available Formats

EBook

Enterprise Drone Inspection:

By Guillaume Delepine, Senior Product Marketing Manager

Data Quality Requirements for Key Use Cases 4

Meeting the Bar for Automated Roof Inspection 6

Skydio Solutions for Inspection 11

Conclusion and Looking Ahead 14

Construction Primary: Progress Construction Asset damage, Up to 1 cm

Transportation Primary: Bridge Bridges, Roads, Cracks and Sub-millimeter to

Utilities Primary: Power plants, Rust and loose Up to 0.5 cm

Telecom Primary: Digital Cell Towers Labels on antenna < 2 mm

Insurance Primary: Roof Homes Hail damage < 1 mm

Focus of this volume

First Responder Primary: Crime or accident Crumple damage, ~ 1 mm

Key requirements for Remote Asset Management

Photo Quality Consistent exposure to capture areas backlit by the sun

Photo Location 80%/80% overlap/sidelap. Ideal pattern is consistent orbits at

Relative and Absolute Accuracy Only relative accuracy typically required

Skydio 2 - 2m fixed focus Skydio X2 - 2m fixed focus

1.25 m 0.52 0.25

1.5 m 0.63 0.31

Skydio Enterprise Controller

Skydio Enterprise Controller

You might also like