Vision-based Casting Positioning System Design

Castings are molded metal objects obtained by various casting methods.They occupy a large proportion of mechanical products and are closely related to daily life.Due to the inherent defects of existing casting methods,grinding of castings is an indispensable procedure in the production process of castings,and most of them are processed manually.How to realize the automatic grinding of castings and raise the productivity of castings is a problem of great urgency for many factories.From the perspective of image processing,we designs a vision-based casting positioning system combined with industrial polishing robots,in order to achieve rapid localization of castings.Based on the positioning information,the path planning of the robot is carried out to realize grinding of the castings automatically.At first,we built a robot vision system.It can convert the relationship between the visual coordinate systems and calibrate cameras to realize acquisition of casting images automatically.Next,traditional image processing algorithms and deep learning based image processing algorithms are used to obtain casting features,which are used to realize the positioning of the casting.Finally,after obtaining the positioning information of the casting,the information is fed back to the robot,and the robot is controlled to move to the designated point to complete the positioning task.Through experiments,our system can quickly and accurately realize the positioning of castings,greatly improve the speed of casting localization,and improve the efficiency of casting grinding.