Design of a Vision-Guided Hierarchical Control System for Robotic Rebar-Tying Tasks

作者：Jinzhu Shen, Hao Duan, Tengfeng Ai, Haili Jiang, Shuai Guo · 年份：2025 · DOI：10.1109/cpsi66656.2025.11343974 · 研究领域：Innovations in Concrete and Construction Materials、BIM and Construction Integration、Modular Robots and Swarm Intelligence

This paper presents a vision-guided hierarchical control system for robotic rebar-tying tasks in construction environments. The system follows a three-layer distributed architecture consisting of an upper-level interface, a C++-based intermediary controller, and a PLC-controlled robotic unit. The intermediary layer coordinates command parsing, visual perception, motion control, and status feedback, using TCP and ADS protocols for decoupled communication. A YOLOv5-based detection module processes RGB-D images to localize rebar intersections, and coordinate transformation maps detections into robot space. Two task scheduling algorithms are implemented: a basic sequential strategy and an optimized parallel-action version where tying gun rotation and descent are executed concurrently. Experiments involving 32-node tying sequences across four stations demonstrate that the optimized algorithm reduces average operation time by over 50% while maintaining stable performance. The results validate the robustness, flexibility, and reusability of the proposed intermediary system, supporting autonomous task execution without code-level changes. This architecture offers a scalable solution for intelligent construction robotics.