Design and Development of a Power Transmission Line Inspection Robot

Abstract

Manual crews or helicopters inspecting overhead power transmission lines are dangerous, expensive and inefficient. Existing robotic solutions such as UAVs and walking robots solve parts of the problem but are limited in endurance, payload or obstacle negotiation. This report outlines the design and development of a hybrid Power Transmission Line Inspection Robot (PTLIR) over two semesters of the Nazarbayev University Capstone program. The first semester was devoted to the walking subsystem. Three mechanical iterations were realised, leading to a V-shape chassis with A2212 brushless motors, tested on laboratory conductor mock-ups. The second semester took the platform along two parallel tracks: mechanical refinement, which included the addition of a three point caster wheel support, a servo driven camera gimbal, and JGB37-520 geared DC motors (Iteration 4), and the development of an autonomous vision subsystem. The vision pipeline is a composition of classical computer vision (Gaussian preprocessing, Canny edge detection, Probabilistic Hough Transform using OpenCV) for real-time line tracking and a YOLO11 deep-learning defect detector trained on the InsPLAD-det dataset. A custom To address the detection of small defects under resolution limits, a Manual Slicing inference strategy was developed, which achieved successful detection of defect classes on previously unseen images with a backbone precision of about 0.90. A Flask based MJPEG telemetry server supports remote supervision. Preliminary locomotion was confirmed on a metallic-tube test rig in April 2026. The results demonstrate the feasibility of a safe and inexpensive robotic inspection of high-voltage infrastructure.