MUMBAI, India, Sept. 7 -- Intellectual Property India has published a patent application (202641064150 A) filed by Srm Institute Of Science And Technology Tiruchirappalli on May 21, 2026, for Ai-Driven Vision Guided Laser Based Qr Encoding Using Adaptive Clean-Zone Localization For Railway Track Components.
Inventors include Rohan Rose J; Lakshara Anand V V; Kurinji Eswar J A; and Dr. K. B Ajeyprasath.
The application for the patent was published on September 04, 2026, under issue no. 36/2026.
Abstract: Railway track fittings such as fishplates are critical structural components whose condition directly affects operational safety. Traditional inspection techniques for detecting rust and surface degradation rely heavily on manual observation, making them time-consuming, inconsistent, and prone to human error. This work proposes an intelligent, vision-based inspection and targeting system that integrates deep learning–driven semantic segmentation with robotic laser guidance to identify optimal non-rusted regions on railway fishplates. A transformer-based semantic segmentation model is employed to classify rusted and clean metallic surfaces using RGB images acquired from a fixed monocular camera. A custom pixel-level annotated dataset of real-world fishplate images was developed to train and validate the model. Based on segmentation results, the system quantifies rust severity as a percentage of total surface area and automatically identifies the largest contiguous clean-metal region as an optimal targeting location. The selected pixel coordinates are transformed into angular offsets using camera projection geometry, enabling real-time control of a pan–tilt laser turret powered by an ESP32 microcontroller to accurately point toward the selected region.
Disclaimer: Curated by HT Syndication.
