Digital workflow for printability checking and prefabrication in robotic construction 3D printing based on Artificial Intelligence planning
Résumé
This paper introduces an innovative digital workflow for printability and prefabrication checking in robotic construction 3D printing based on Artificial Intelligence (AI) planning techniques. The proposed method
generates operational sequences for a robotic manipulator tasked with concrete 3D printing by utilizing Planning Domain Definition Language (PDDL) 2.1-based AI planners. The planned sequences consider diverse
geometry specifications, aligning with printability checks and prefabrication requisites. The sequences can then be executed in a robotic simulation environment, allowing operators to assess the 3D printing process in every aspect before real-world deployment, and potentially avoiding costly mistakes and the waste of resources. Compared to existing methods, the use of AI planning makes our approach more modular and easy to adapt across a variety of robotic platforms, diverse designs, materials, and printing conditions; and it is released as open-source software to the benefit of the community. The proposed method was tested by processing objects with multiple geometries and a varying number of printing layers, and by employing two different PDDL planners. Results prove the viability of the approach but also hint at possible scalability limitations for complex geometries that involve a large number of edges to be printed.