Using Drones to Uncover the Risks of Falling from Roofs!

Roof construction accidents at cattle sheds, factories, warehouses, and other
sites resulted in 125 fatalities over the three-year period from 2020 to 2022.
These fatal accidents are particularly prevalent in small-scale, extremely shortterm projects that last just one to two days.

 The Korea Occupational Safety and Health Agency (KOSHA), a subsidiary of
the Ministry of Employment and Labor, launched a pilot program from August
to November 2023 to establish a roof construction accident prevention system
utilizing digital twin* technology, with a specific focus on preventing roof
collapses. This system enables the assessment of roof material, aging, and other
risk factors through drone photography. As the project advances, specialized
accident prevention measures for high-risk roof sites are expected to become
feasible.
* Digital twin: a virtual representation of real-world facilities or locations, implemented as
a digital space for various applications

KOSHA has used its pilot program and developed digital maps for cattle sheds
with limited access. Collaborating with Hapcheon-gun in Gyeongsangnam-do,
405 cattle sheds were selected for digital twin representations, and drone aerial
photography (with a resolution of 5 cm/pixel) was conducted following community
outreach.

 To assess the risk level of photographed cattle shed roofs, the agency
collaborated with experts to derive risk assessment criteria. These criteria
incorporate factors such as material, aging, damage, and skylights/solar panels/
openings; and include area and quantity information. The finalized criteria were
applied to the 405 cattle sheds, categorizing them into five risk levels (safe,
caution, warning, severe, high risk).

 KOSHA has developed digital maps illustrating the locations of cattle sheds and
their associated risk levels. These initiatives include offering direct technical
support to high-risk cattle sheds, disseminating one-point sheet (OPS: a type
of safety checklist used for conducting risk assessments) and other safety
and health materials, collaborating on inspections and campaigns with local
governments (Hapcheon-gun), and executing various activities to prevent
industrial accidents.

In the upcoming year, KOSHA will expand digital mapping to encompass not
only cattle sheds but also factories and industrial complexes prone to roof
construction risks. Plans are also in the works for developing an AI-powered
automatic recognition system capable of identifying roof material types and risk
factors. To this end, satellite and aerial photography data will be collected and
associated dataset* will be established. Moreover, various pilot measures will be
undertaken such as sensor installation (including displacement/access detection
sensors, with future sensor types to be determined) to detect fall risks among
roof construction workers.
* Data including objects, scope, risk information, etc., for AI learning

 The ultimate goal is to apply the developed system to industrial accident
prevention projects in densely populated areas with roof construction risks.
Rapid establishment of a roof risk database in vast areas where direct access are
challenging will enable efficient industrial accident prevention activities such as
fall risk zone management and close work progress monitoring.

 KOSHA President Ahn Jong-ju expressed optimism, stating: “We anticipate an
efficient reduction in fatal accidents by establishing industrial accident prevention
models using digital twin and other smart technologies. We will continue to
explore various smart safety and health systems to take the lead in mitigating
serious accidents.”