Mithril Signs AI Safety Management PoC Contract with Sampyo P&C
Verification of AI Video Analytics Technology in Crane Lifting and High-Altitude Work Sites Field Demonstration of Features Detecting Helmet and Safety Hook Usage as Well as Unauthorized Access to Danger Zones
MITHRIL, an AI deep-tech company providing the industrial safety AI solution 'Guardian-Alpha' and the manufacturing AX solution 'Optima-Alpha', announced that it has signed a PoC (Proof of Concept) contract with Sampyo P&C, a construction company specializing in precast concrete (PC) structures, to build an AI-based real-time safety management system at construction sites.
Sampyo P&C operates PC structure construction sites where heavy loads are lifted using cranes and high-altitude assembly work is performed. The company explained that due to the nature of construction sites where multiple work types are carried out simultaneously, blind spots outside the safety manager's line of sight can occur.
The purpose of this PoC is to apply MITHRIL's AI video analytics technology to the field to verify the feasibility of detecting situations deemed as worker non-compliance with safety rules and notifying managers.
According to MITHRIL, the demonstration system consists of fixed AI cameras attached to cranes and temporary AI cameras that can be moved and installed according to the progress of the construction process. The AI inference will operate on an on-device basis, processing data directly on the camera terminal, allowing site-by-site deployment without constructing a separate central server.
In this PoC, key detection items include:
Workers not wearing safety helmets
Unfastened safety hooks during high-altitude work
Workers approaching hazardous zones beneath lifted loads
The PoC will verify the function of sending notifications—including the detected item, timestamp, and related video clips—to dashboards and mobile devices. The on-site alarm system is planned to be introduced gradually after confirming system stability and field acceptability.
The company explained that it plans to apply a wireless communication method tailored to site conditions and operate event video recordings via local storage. The stored videos will be utilized for site safety management and investigating cause of accidents.
At construction sites, process stages change frequently, and environmental variations such as dust, backlight, and rainfall can affect video analytics performance. MITHRIL stated that during a stabilization period of about one month after installation, it plans to verify detection performance by adjusting models based on field data and reflecting cases of false positives and false negatives.
MITHRIL explained that it plans to use the performance and operational data gathered during the PoC for future system improvements and to evaluate deployment at other sites. Once the demonstration is complete, expanding the application scope to other Sampyo P&C sites will be considered based on the results.
An official from Sampyo P&C stated, "Due to the nature of PC structure construction sites, proactive and systematic safety management is crucial. Through this contract, we will verify the field applicability of AI-based safety management technology and establish an effective, prevention-focused safety management system."
An official from MITHRIL stated, "Safety accidents at construction sites often occur due to momentary carelessness and management blind spots. Through this PoC, we will verify the field suitability of the AI-based real-time safety management system and develop it into a solution that substantially contributes to preventing severe industrial accidents."
By Reporter Doo-yeol Kim
Comments