Overview

In the immediate aftermath of earthquakes, UMC’s top priority is to quickly ensure the safety of equipment and facilities, allowing employees to safely resume their work. Previously, safety inspections were conducted manually, which posed risks to personnel from potential hazards such as collapsing objects, gas leaks, and equipment failure. This approach is not only inefficient but also exposes workers to significant safety risks.

As UMC undergoes digital transformation, our fab operations no longer solely reply on manual work, but are increasingly adopting smarter and automated technologies. Particularly for equipment inspection, recent advances are revolutionizing this critical area of fab management.

In the event of disasters or emergencies, fabs need smarter and safer inspection techniques to ensure normal factory operations. To meet these needs, UMC’s team has developed in-house an Autonomous Mobile inspection Robot (AMR) for smart inspection and risk detection.

Traditional inspections are carried out by experienced personnel who weave their way among equipment, checking the status of each machine through their own auditory and visual observations. However, in a large fab with equipment packed closely together, the noise makes it very difficult for inspectors to precisely detect anomalies and identify potential equipment failure risks on the spot. For example, the sound of leaking gases or liquids is nearly imperceptible, while the sound of anomalous vibrations is easily lost among the ambient noise of operating machines. Similarly, overheating of equipment can only be identified when it reaches a level that is visible to the naked eye. These conditions make early detection difficult. As a result, by the time problems are identified, equipment may have already been damaged, or potential workplace hazards may have occurred.

A more pragmatic concern is that the efficiency of human-based inspection is struggling to keep pace with fab expansion. The current inspection approach requires two inspectors to spend at least 1 hour per day to complete routine inspections, which incurs high labor costs but cannot guarantee timely or comprehensive checks. Therefore, UMC decided to introduce the inspection AMR, which was collaboratively developed by internal departments including fab facility, smart manufacturing, and IT teams.

The inspection AMR is capable of self-navigation and is equipped with powerful sensors, making it the fab’s eyes and ears for monitoring real-time equipment conditions. Three critical technologies have been integrated into the robot:

1. Ultrasonic Acoustic Camera: Catches imperceptible sounds
   This technology detects high-frequency sound waves to pinpoint the faint sounds of gas or liquid leaks or sound wave oscillation due to anomalous mechanic vibrations, even in noisy environments. Problems that could rarely be identified through traditional inspection methods can now be analyzed with AI, which precisely locates potential sources of anomalies and facilitates early intervention by the maintenance team to prevent further equipment damage.
2. Thermal Imaging System (IR): Detects hidden signs of overheating or leaking
   Equipment overheating is usually an indication of malfunctioning internal mechanisms, and liquid leakage is a common hidden hazard in factories. The thermal imaging system can detect liquid residues on the ground that are difficult to identify with the naked eye, discover anomalous oil or chemical leaks in a timely manner, and clearly indicate overheating parts so that engineering personnel can swiftly locate problem areas.
3. Real-time Imaging and AR Technologies: Smart inspection that reduces human interpretation errors
   By employing AI image analysis, the robot can instantly identify workplace safety violations such as personnel not wearing safety helmets and unusual clutter in equipment areas. Moreover, by utilizing augmented reality (AR) technology, the system can indicate the precise location of malfunctioning equipment on the monitoring screen in real time, enabling the swift dispatch of maintenance teams to carry out repairs.

The AMR is more than a robot—it changes how we think about fab inspections. By combining AI analysis with backend robot monitoring functions (iRobot), the AMR will also be key to collecting a fab’s operational data. In addition to recording anomalies, it actively learns from inspection data and continuously optimizes its judgment logic to increase the precision of anomaly detection and enhance predictive maintenance efficiency.

In the next phase, UMC aims to integrate the AMR system into the fab’s monitoring and data collection platforms as well as the facility’s management platform to achieve a more comprehensive digital factory framework.