MAZIN Inc. (Headquarters: 2F Wakasumi Bldg., 3-3-6 Nihonbashi Honcho, Chuo-ku, Tokyo; Representative Director: Takanori Kadoya) will begin offering the AI robot inspection system "TPIM01" for automating the inspection process of machined parts in the manufacturing industry, starting July 2, 2026.
This product autonomously picks products from a mixed-bin state using AI, completing everything from external scratch inspection to sorting without "robot teaching," bringing a new option to sites with high-mix low-volume production.
It can be viewed at "Manufacturing NEXT (Booth No.: E35-30)" during the "38th Manufacturing World Tokyo" held from July 1 (Wed) to 3 (Fri), 2026.
> https://www.manufacturing-world.jp/hub/ja-jp/about/next.html
Background of Development — The Wall of Visual Inspection in High-Mix Low-Volume Production and Market Potential
In recent years, the manufacturing industry has faced a severe labor shortage due to an aging population and the retirement of experienced inspectors, making the automation of inspection processes an urgent task. For many years, visual inspection has relied on human "eyes" and "senses," but risks of overlooking defects due to fatigue and the "personalization" of judgment criteria varying among inspectors have been pointed out as structural issues.
As a solution to these challenges, investment in AI image recognition technology is steadily expanding. According to a survey by Deloitte Tohmatsu Mic Research Institute (2025 edition), the domestic market size for image recognition solutions utilizing deep learning is projected to expand to 44 billion yen in FY2024 (120.5% year-on-year), and is expected to continue growing at an average annual rate of 15.6%, reaching 92 billion yen by FY2029*1.
In particular, the application for defect inspection is growing steadily at 115.5% year-on-year, indicating high expectations for the automation of visual inspection in manufacturing sites.
Indicator
Value
Details
MARKET CAGR
15.6%
Average annual growth rate (projected to expand to 92 billion yen by FY2029)
MARKET SIZE 2026
44 billion yen
Domestic market size for image recognition solutions (120.5% year-on-year)
On the other hand, with the diversification of consumer needs, many factories are shifting to "high-mix low-volume production." However, conventional robot systems and image inspection devices required robot teaching (instruction work) by specialized engineers and retraining of AI using large amounts of image data every time the product type changed. The "shortage of personnel to handle teaching" and the resulting line stoppages (downtime) have been structural barriers preventing automation investment in many factories.
Product Overview — What is the Machined Part Inspection System "TPIM01"?
"TPIM01" is a dedicated system that autonomously performs everything from picking to inspection and sorting of external appearances for machined parts (round metal parts such as shafts, pins, bolts, and rings), which have traditionally relied on manual labor. The preparatory work that was a hurdle for high-mix inspection has been simplified to the extreme.
FEATURE 01: Autonomous Picking Without Teaching
By simply registering a 3D model of the inspection target, the AI automatically determines the product's position and posture, and the gripping point. Stable picking is possible even in a mixed-bin state with metallic luster, which is difficult for image recognition.
FEATURE 02: Overwhelming Labor Saving Completed with Learning from One Good Product
The inspection surfaces and robot inspection trajectories are automatically generated according to the product shape (3D model). No need for large quantities of defective product images or multiple good product images; inspection preparation is completed by learning just "one" good product in the actual image capturing mode.
FEATURE 03: Automatic Recording and Visualization of Inspection Results
Inspection results are automatically saved, and the operating status can be checked in real-time on a monitor. It also supports traceability of inspection history.
The AI robot operates based on the 3D model of the target product and one good product photo. Its major feature is autonomously executing from picking to inspection.
The AI robot operates based on the 3D model of the target product and one good product photo. It autonomously executes from picking to inspection.
Impact of This Product
The introduction of TPIM01 is expected to bring the following changes to manufacturing sites in the inspection of applicable machined parts:
Immediate Adaptation to High-Mix Lines and Simplified Setup Changes By enabling the setup of an automated inspection line with just "preparation of the 3D model" and "shooting of one good product," it reduces the operational burden on sites where setup changes and product additions occur frequently.
Stabilization of Quality Judgment and Elimination of Personalization Suppresses overlooking defects in visual inspection and variations in judgment due to the inspector's physical condition or experience. Since the line can be operated by the site staff without specialized robot engineers, it also helps prevent the personalization of technology.
Liberation from Inspection Tasks and Redeployment of Personnel By automating repetitive visual inspection tasks, it creates opportunities to shift personnel involved in inspection to more value-added work.
[Scope of Application] TPIM01 is a dedicated machine for metal machined parts (with certain guidelines for outer diameter and length) and generally does not cover inspection of features hidden in the depths of sides, such as cross-holes. We recommend confirming applicability to your company's products and lines in advance through the following actual work demo.
Actual Work Demo Reception and Usage Method
To allow you to confirm the performance of the actual machine before implementation, we accept demos using actual workpieces. By sending us the 3D model of your target product in advance, we can confirm applicability, and by bringing your actual workpieces to our research institute (Tsukuba), you can directly observe the entire process from mixed-bin picking to inspection. You can confirm inspection accuracy before implementation, so please feel free to contact us first.
First, please confirm inspection accuracy with an actual work demo.
We can confirm applicability simply by sending us the 3D model of your target product.
At our research institute (Tsukuba), you can see the entire process from mixed-bin picking to inspection on the actual machine.
Application and Inquiries Here
Future Outlook
Through this product, MAZIN Inc. will present a realistic solution to the two major challenges in manufacturing sites: "teaching" and "learning data collection." Furthermore, the technologies of the "mixed-bin picking unit" and "high-mix compatible AI inspection system" used in TPIM01 have potential applications in processes other than machined parts. We will continue to contribute to strengthening the competitiveness and improving the productivity of Japan's manufacturing industry through the latest AI robotics technology.
*At "Manufacturing NEXT," an exhibition within "38th Manufacturing World Tokyo" where this product is being exhibited, you can also see "AI Drawing Unpacking," which ranked first in viewing numbers in the "AI Application" category of the featured product rankings. Please check it out when you visit. (*Reference)
(Photo) Display of featured product rankings at the "Manufacturing NEXT" venue. "AI Drawing Unpacking" ranked first in viewing numbers in the "AI Application" category.
Sources and Notes *1 The domestic market size of AI image recognition (deep learning utilization) solutions (44 billion yen in FY2024 actual, 120.5% year-on-year), average annual growth rate of 15.6%, projected 92 billion yen in FY2029, and 115.5% year-on-year growth for defect inspection applications are based on Deloitte Tohmatsu Mic Research Institute's "Current Status and Outlook of the Image Recognition Solution Market Utilizing AI (Deep Learning) 2025 Edition [8th Edition]."
> Deloitte Tohmatsu Mic Research Institute: Current Status and Outlook of the Image Recognition Solution Market Utilizing AI (Deep Learning) 2025 Edition [8th Edition]
> https://mic-r.co.jp/mr/03540/
Company Profile
Company Name
MAZIN Inc.
Location
2F Wakasumi Bldg., 3-3-6 Nihonbashi Honcho, Chuo-ku, Tokyo
Representative
Takanori Kadoya, Representative Director
Established
June 2018
Capital
430 million yen (including capital reserve)
Business Activities
Development and sales of industrial equipment, provision of human resource services in the manufacturing domain
URL
https://www.mazin.tech
Contact Information for this Matter
Contact Persons: Katayama, Matsumaru
TEL: 03-5981-8078
Email: contact@mazin.tech
Product Details and Application for Actual Work Demo
FACT BOX
- Source: PR TIMES
- Category: 製品リリース