Robotic deburring strategy using burr shape recognition

Abstract

Deals with our ongoing work on burr shape recognition based automatic robot deburring strategy with a laser vision sensor. The target workpiece of this project is the complicated burr produced in cast products, which is difficult to recognize due to its shape irregularity. A burr model is established regarding a rigid tool holder to describe several types of burrs including the cast burr, which exist in the typical machining process. An image processing based methodology is applied to restore the original shape of the burr and separate it from its workpiece. Then, the shape parameters of the modeled burr are identified by matching them with the real burr shape. A multiple deburring pass planner utilizing the identified burr is proposed to rapidly remove the burr while satisfying the feasible constraint of grinding mechanics. Experimental results show the validity of our approach.