Automatic Defect Generation Tool in CAD Models

Abstract

Automated inspection and repair of defects on car bodies is a major challenge in the automation of the automotive sector, especially in surface treatment tasks such as sanding or polishing. Since the physical tests required in surface treatment tasks are destructive and costly, the use of virtual environments is proposed to validate new trajectory planning and force control algorithms. This work presents a tool for the semi-autonomous and parameterized generation of defects on CAD models, which will be integrated into an inspection and repair system based on ROS and Gazebo. The system uses the digital twin of a KUKA LBR iiwa robot with integrated force/torque sensors, mounted on a mobile platform, and includes virtual models of tools and consumables. The tool enables the generation of both micro-defects and macro-defects (dings and dents) in various shapes. This facilitates the development and validation of repair algorithms in simulation, accelerating the design of industrial solutions.