Manual installation of heavy floor tiles imposes physical demands on construction workers, requiring prolonged bent postures and repetitive lifting of substantial weights. Such high-intensity labor leads to fatigue accumulation, heightened risk of musculoskeletal injuries, and compromised tiling quality. While existing floor tiling robots have demonstrated feasibility for small and lightweight tiles, robots capable of handling large, heavy tiles remain rare.
This research presents the development of a six-degree-of-freedom (6-DOF) Stewart platform-based floor tiling robot designed specifically for automated tiling of heavy floor tiles. The robot comprises a fixed base, a moving platform, and six electrically powered cylinder legs connected via spherical joints, forming a parallel kinematic structure that enhances payload capacity and working flexibility. A vacuum suction system with four overload suction cups provides reliable tile handling for tiles weighing up to 30 kg.
A dual-camera visual recognition system was designed to address the challenge of recognizing large-sized floor tiles. Two cameras mounted on extension rods capture tile corner points from complementary perspectives. A “Reference Line-based Tile Recognition” (RLTR) approach, powered by a pre-trained YOLOv8 model (97% precision, 98% mAP), was introduced to enable accurate tile boundary detection and alignment computation.
On-site experimental validation demonstrated that the robot achieved angle deviations within ±0.001° and distance deviations within 0.5 mm across 30 repeated trials.. Comparative experiments with 10 construction workers, using Borg RPE scales and EMG signals, confirmed significant reductions in physical strain during robotic-assisted tiling compared to manual tiling, with heart rates decreasing from 90–117 bpm to 61–84 bpm.
This research provides a feasible and validated solution for automating the installation of heavy floor tiles, contributing to improved construction safety, reduced manual exertion, and enhanced tiling precision.