| Title |
An AMR-Based Text-Tracking Lighting System Using Pixel-Deviation-Based Tilt Calibration |
| Authors |
Min-Seo Kim ; An-Seop Choi |
| DOI |
https://doi.org/10.5207/JIEIE.2026.40.5.346 |
| Keywords |
Autonomous mobile robot; Dynamic lighting control; PaddleOCR; DMX512; Text-tracking lighting |
| Abstract |
This study implements a text-tracking lighting system that recognizes the position of text using PaddleOCR, a lightweight deep-learning OCR model, and converts it into control values for a DMX512 luminaire. The system calculates the pixel deviation between the center of the camera image and the centroid of the detected text bounding box, and maps this deviation to the tilt angle of the luminaire. The relationship between pixel deviation and tilt value was first derived through direct measurement in a fixed configuration with a camera-to-text distance of 1,200mm. This distance was determined by considering that text falls outside the camera’s field of view below 500mm, while text recognition becomes unreliable beyond 1,200mm. The system was subsequently mounted on an autonomous mobile robot, in which the robot detects text through rotational scanning and aligns both the horizontal direction and the target distance based on proportional control, while the tilt of the luminaire governs the vertical direction, thereby extending the system into a two-axis coordinated control scheme. As a result, it was confirmed that the luminaire illuminates the detected text. This study removes the positional constraints of fixed luminaires and demonstrates the feasibility of a lighting infrastructure that responds actively to the position of a target object. |