Robot Autonomy for Scrap Cutting in Metal Recycling

Akl, James

Etd

Robot Autonomy for Scrap Cutting in Metal Recycling

Public Deposited

This dissertation develops an automation framework to address the challenges, problems, and opportunities of the metal recycling industry. This is achieved by integrating a variety of components and functionalities into a diverse cognitive architecture. The aim is to endow robotic systems with task-specific autonomy against four main problems found in metal scrap cutting and recycling. These are: (1) Cutting path generation, using viewpoint planning and active perception; (2) Autonomous oxy-fuel cutting, using visual feedback for conditioning and control; (3) Cutting task validation, using learning-based inference via neural network models; and (4) Safe structural disassembly, using sequential decision planning. In this dissertation, we formalize and discuss the design and evaluation of each of these functionalities. Additionally, we analyze the broader impacts of this research from a socio-technical, economic, and ethical perspective. Ultimately, while this proposed framework is tailored towards metal recycling, many of the components’ underlying techniques may be applicable or transferable to tasks of similar nature. In effect, we demonstrate the leverage and flexibility of diverse component-based architectures for augmenting an agent’s capability, intelligence, and autonomy towards a particular goal.

Creator