Development and integration of a perceptive robotic grinding system

Abstract

This thesis presents the development and integration of a perceptive robotic system to eliminate the risks of manual grinding for foundry operators by providing a flexible automation solution. At foundries across the world, human operators perform manual grinding to remove excess material and transform castings into their final shape. Extensive research has highlighted the significant health risks associated with manual grinding in foundries, such as excessive exposure to airborne dust, noise-induced hearing loss, and severe musculoskeletal damage. To address these concerns, foundries have begun exploring technologies to aid operators ergonomically or automate the grinding process. However, current automation approaches struggle to adapt to casting variation and a wide product mix without real-time input from a human operator.

To support the grinding of high variety castings in foundry environments, a robotic system that leverages operator markings to determine where to grind was designed and implemented. Inspired by the previous work of Schimpf et al., the Sketch & Grind Robotic System improves grinding capability by integrating industry-leading hardware with perception and control improvements. Hardware and software developments were necessary for each system component to ensure seamless integration and achieve overall system performance. Current industry practices were investigated to establish an initial abrasive grinding tool for the Sketch & Grind Robotic System. An algorithm within the Sketch & Grind process was developed to enable concave boundaries to mark complex quality issues and avoid obstructions. A complementary process, Scan-N-Plan, was implemented on the Sketch & Grind Robotic System hardware, and the Sketch & Grind process was partially integrated to extend functionality for a foundry implementation. Through these developments, system functionality was realized, and the capabilities of Sketch & Grind were extended. This work supported a broader collaboration between research and industry partners to advance perceptive robotic grinding solutions for foundries. It will culminate in an upcoming industry implementation of the system at a steel foundry.