Publication | Automation in Construction 2021

Robotic assembly of timber joints using reinforcement learning

This paper describes applying Reinforcement Learning to control robot movements in contact-rich and tolerance-prone assembly tasks and presents the first successful demonstration of this approach in the context of architectural construction.

Download publication

Abstract

Robotic assembly of timber joints using reinforcement learning

Aleksandra Anna Apolinarska, Matteo Pacher, Hui Li, Nicholas Cote, Rafael Pastrana, Fabio Gramazio, Matthias Kohler

Automation in Construction 2021

In architectural construction, automated robotic assembly is challenging due to occurring tolerances, small series production and complex contact situations, especially in assembly of elements with form-closure such as timber structures with integral joints. This paper proposes to apply Reinforcement Learning to control robot movements in contact-rich and tolerance-prone assembly tasks and presents the first successful demonstration of this approach in the context of architectural construction. Exemplified by assembly of lap joints for custom timber frames, robot movements are guided by force/torque and pose data to insert a timber element in its mating counterpart(s). Using an adapted Ape-X DDPG algorithm, the control policy is trained entirely in simulation and successfully deployed in reality. The experiments show the policy can also generalize to situations in real world not seen in training, such as tolerances and shape variations. This caters to uncertainties occurring in construction processes and facilitates fabrication of differentiated, customized designs.

Related Resources

Publication

2024

Evaluating Large Language Models for Material Selection

This work evaluates the use of LLMs for predicting materials of…

Publication

2024

Optimized GPU Implementation of Grid Refinement in Lattice Boltzmann Method

Optimized GPU-accelerated algorithm for implementing grid refinement…

Publication

2024

Experiential Views: Towards Human Experience Evaluation of Designed Spaces using Vision-Language Models

Exploratory research on helping designers and architects anticipate…

Publication

2021

UV-Net: Learning from Boundary Representations

We introduce UV-Net, a novel neural network architecture and…

Get in touch

Something pique your interest? Get in touch if you’d like to learn more about Autodesk Research, our projects, people, and potential collaboration opportunities.

Contact us