Maeda Corporation Develops Green Cut AI Judgment System

Maeda Corporation (Headquarters: Chiyoda-ku, Tokyo; President: Soji Maeda; hereinafter "the Company") has developed a "Green Cut AI Judgment System" and specialized "shooting robots" to standardize and elevate the quality of "green cut" work in dam construction. Green cutting involves removing fragile surface thin films (laitance) after concrete placement to prepare for subsequent layers. This work previously relied heavily on the experience of technical staff, but the new system enables objective judgment using AI and robots.

This system allows AI to substitute for criteria that previously relied on the intuition of skilled technicians. By integrating with the shooting robot, high-level quality control becomes achievable regardless of the technician's experience.

Green cutting is a critical process for dam strength, directly affecting the integration and durability of concrete joints. Traditionally, quality judgment was done solely through visual inspection, which made it difficult for less experienced technicians to perform accurately. Given that dam projects are long-term, personnel often change, leaving judgment standards vulnerable to individual subjectivity. These challenges drove the development of an objective, consistent system that can be used by less experienced staff.

The shooting robot was created to address inconsistencies in height and angle during manual shooting. It consists of an aluminum frame, GNSS, web camera, light-shielding sheet, communication equipment, and a control PC. Since construction areas in dams are typically wet, the light-shielding sheet suppresses solar reflections from water surfaces and robot shadows.

Shooting modes include 'Auto Shooting Mode' and 'Remote Shooting Mode.' In Auto Shooting Mode, users simply send pre-set routes from a tablet to the control computer and press start.

The system analyzes photos taken by the robot using AI, automatically judges the green cut quality, and outputs the results. To handle photos taken in wet areas where judgment might be difficult, the system applies weighting to water puddle areas and considers surrounding results for final judgment.

The developed robot and system confirm:

- Establishment of shooting methods under stable conditions
- Realization of surface quality management through AI
- Objective visualization of green cut work quality

Following the previously developed 'Automatic Green Cut Machine' (released Jan 19, 2026), the Company continues to aim for system development that fully automates the workflow from execution to quality judgment in the green cut process.