Recognition of the rebar binding state based on Bag of Features

Abstract

Rebar binding robot is a typical kind of construction robot, replacing manual binding of steel bars in a standard environment. We develop a robot prototype and build an experimental environment to verify the effectiveness of the Bag of Features based rebar binding state recognition for the binding images taken during the operation of the robot. The method successfully classifies two states of bound and unbound of rebar under different lighting conditions, providing guidance for the engineering application. The dataset used in the article has been publicly released.