LOAS, a startup innovating industrial sites with acoustic AI that detects abnormal noise in just 0.8 seconds

Broken machines make noise. But in noisy industrial environments, detecting subtle signs of abnormalities is never easy. LOAS, an AI acoustic solution startup, removes surrounding noise through proprietary 3D beamforming technology and detects defects with 99% accuracy. This story introduces LOAS’s unique technology, capable of visualizing problems in just 0.8 seconds to simultaneously improve both safety and quality. It also follows Jaehyun Lee, who participated in the second batch of D.CAMP Batch, as he demonstrates the market viability of the technology and expands toward the global stage.

Q1. Please introduce LOAS, a company that analyzes industrial sounds using artificial intelligence.

LOAS is a deep-tech company that analyzes acoustic data generated across various industrial environments using artificial intelligence to diagnose abnormal signs and anomalies.

Its flagship product, SMART, is deployed in production-line quality inspection processes to detect abnormal sounds and identify defective products based on acoustic analysis. In particular, the system delivers diagnostic results within approximately 0.8 seconds, enabling rapid feedback to users.

Another key solution, ARQOS, combines LOAS’s existing AI diagnostic technology with Physical AI technologies such as drones and robots to diagnose abnormalities in industrial facilities including factories and power plants. Operating remotely 24 hours a day, 7 days a week, ARQOS can patrol entire industrial sites and inspect hazardous areas that are difficult for humans to access directly.

Q2. What inspired you to start the business?

Even before LOAS was founded, people in manufacturing environments were already using sound to diagnose abnormalities. However, the process heavily relied on the subjective senses and experience of skilled workers, which often resulted in inconsistencies in inspection outcomes. In many cases, final inspections were still performed by manually placing one’s ear close to the product to check for abnormal sounds.

We became convinced that if these sounds could be precisely converted into data, it would be possible not only to maintain more consistent product quality but also to prevent accidents across industrial environments. In particular, we believed that if abnormal sounds could be classified and identified among the wide variety of noises generated within three-dimensional spaces, much more precise diagnostics would become possible.

“Proprietary 3D beamforming technology precisely detects abnormal sounds.”

Q3. Industrial environments are extremely noisy. How are you able to accurately isolate only abnormal sounds?

We utilize “AI Square,” an AI acoustic detection engine developed with LOAS’s proprietary technology. AI Square performs two primary functions. First, it effectively filters out various types of environmental noise. Then, based on the refined acoustic data, it diagnoses abnormal sounds generated by specific products.

At the core of this process is our proprietary “3D beamforming”-based spatial acoustic separation technology. The system divides space into approximately 4,000 acoustic matrices and visualizes them, then rapidly scans the environment in intervals of 0.1 seconds (100 milliseconds). During this process, AI algorithms analyze sound movement and reflection patterns to accurately identify which part of the product is generating the abnormal sound, while simultaneously visualizing the location in real time.

Q4. To be applied in industrial environments, it seems that everything from diagnosis to visualization needs to take place almost in real time.

That’s correct. To address this, we advanced the SMART system so it can identify the location of abnormal sounds within approximately 0.8 seconds. However, achieving this level of performance was extremely challenging. When acoustic data is visualized in the form of imaging data, the amount of information that must be processed in real time is roughly 16 times greater than that of conventional vision inspection systems.

Nevertheless, we could not allow the product production Takt Time to slow down, which meant additional technological innovation was essential. To solve this challenge, we independently developed a proprietary data compression format that significantly reduced processing time. As a result, we were able to accurately identify the location of abnormal sounds within 0.8 seconds.

Once the computation process is completed, the acoustic data is provided to users in the form of visual information. Including the visualization stage, users can intuitively identify the exact location of the issue within approximately 1.2 seconds.

Q5. You also introduced ARQOS earlier. Could you explain in more detail how it diagnoses abnormalities in industrial facilities?

We deploy our acoustic detection technology on robots and drones to inspect various areas within industrial facilities. Many industrial environments — such as high-voltage substations, thermal power plants, and steel manufacturing plants — are too dangerous for humans to access directly. In these environments, robots and drones are deployed in place of people to detect subtle sound changes generated by facilities and machinery in real time.

Through this approach, the system can identify abnormal signals such as tiny gas leak sounds or friction noises from components before accidents occur. Ultimately, this enables safer and more precise unmanned monitoring even in industrial sites where continuous human operation is difficult.

Q6. Your accuracy must be extremely high if the system can isolate abnormal sounds across various industrial environments.

Our technology can detect only the actual abnormal sounds with over 99% accuracy, even in environments with extremely high levels of surrounding noise. For example, even from a distance of approximately 25 meters, we can precisely identify the source location of a sound within an error range of less than 5 millimeters.

We have also conducted experiments in extremely noisy environments, such as the middle of Gangnam-daero in Seoul, where the system successfully detected abnormalities as subtle as a single strand of hair lodged inside an air purifier. Furthermore, we have secured technology capable of tracking and detecting small drones in real time from distances of up to 350 meters.

“Achieved a 100% Paid Conversion Rate for PoC Projects with Over 99% Accuracy”

Q7. What have been the most meaningful achievements over the past year?

First, the SMART acoustic inspection system has been deployed across 15 domestic and overseas production lines of LG Electronics, as well as at two sites affiliated with and partnered with Hyundai Motor Company, generating cumulative revenue of approximately KRW 4 billion.

In addition, through repeat orders from existing clients and expansion into six overseas sites, we secured approximately KRW 500 million in additional revenue. In particular, we consider our expansion beyond the previous LG Electronics-centered pipeline into the automotive sector — including Hyundai Motor Company, Hyundai Transys, Hyundai Mseat, and Daewon Precision Industry — to be one of our most meaningful achievements.

We launched ARQOS in 2025, and after PoC projects, the solution immediately transitioned into paid contracts. We believe this was possible because LOAS products were able to quantitatively demonstrate reductions in operational costs.

In addition, among companies possessing related technologies, LOAS became the only company to establish international R&D collaboration with Fraunhofer IDMT in Germany. Through this partnership, our acoustic AI engine was recognized as meeting global-level technological standards and capabilities.

Q8. As LOAS continued growing both technologically and commercially, you participated in the second batch of the D.CAMP Batch program. Which part of the program was the most helpful?

The most valuable support was definitely the mentoring program. Our mentor, Hwang Hee-cheol, who had experience in the same sector, provided clear solutions to the challenges LOAS was facing, which was extremely helpful.

More specifically, we received guidance on how to scale SMART and ARQOS not as standalone products, but as expandable service models. We gained important insights into transitioning from a PoC-centered approach toward a structure capable of generating recurring revenue and scalable expansion.

The PR support also played a major role in producing tangible results. Until then, we had primarily focused our messaging on technology and field performance. However, through D.CAMP’s support, we were able to reposition LOAS not simply as a technology-focused company, but as a “solution company validated in real industrial environments.”

As a result, our press releases evolved beyond simple exposure to potential customers and began generating meaningful business opportunities, including new meetings, PoC discussions, and partnership consultations.

Q9. If you had to describe the D.CAMP Batch program in one word, what would it be?

I would describe it as “belief.”

The reason is that, as someone running a company myself, there are moments when I lose confidence in what I’m doing. I often wonder, “Did I make the right decision?” or “Am I heading in the right direction?” During those moments, the program helped me stay grounded, continue moving forward without wavering, and ultimately build confidence and belief in our direction.

Q10. Now that you have completed the D.CAMP Batch program and achieved growth in various areas, what is your vision moving forward?

Our goal over the next year is to fully enter the global market and generate meaningful business results. Based on the performance already validated across domestic industrial sites, we are pursuing PoC projects primarily in the Middle East and Asia, where demand for plant manufacturing is especially high.

Through these efforts, we aim to secure overseas customer references and expand our business onto the global stage. Ultimately, we want LOAS’s AI acoustic engine to become a core system responsible for ensuring safety across industrial environments worldwide.

Source: D.CAMP > News > Story

View Original Content>>https://dcamp.kr/news/story/7321

─

LOAS Inc.

(주)로아스

Contact

Tel. 02-6486-6411

Email. info@loas.ai