Density-based spatial clustering and discriminative modeling for automatic recognition and localization of cast-in hoist rings

Abstract Manual hooking the lifting gear of a crane onto the cast-in hoist rings (CHRs) of precast concrete components (PCCs) is the first step when loading PCCs for hoisting and transportation. However, the hooking process is dangerous and inefficient due to the complicated site environment and the falling risk workers may face when they need to work at height. This paper presents a method for automating the first step of the process: recognition and localization of CHRs of PCCs. We collect point clouds by placing laser radars on the surface of PCCs, and then the method follows three steps to implement the automation: denoising laser point clouds with a set of filtering rules, extracting rough recognition areas of CHRs using a spatial clustering algorithm, and determining the location of CHRs through pattern recognition. The experimental results indicate that the method achieves desired performance in terms of 0.864 precision and 0.822 recall.