Humane Automation: Yuning Wu’s PhD Proposal
PhD candidate in Computational Design Yuning Wu will present her PhD proposal, entitled Towards Humane Automation: An RL-Driven Robotic Framework for Supporting On-Site Construction Workers on December 11, 10:30 AM, at the Mill 19 boardroom. Please join us virtually or in person (details below).
PhD Proposal
Towards Humane Automation: An RL-Driven Robotic Framework for Supporting On-Site Construction Workers
Yuning Wu, PhD Candidate in Computational Design
Date: Dec 11, 2023
Time: 10:30 am – 1:00 pm
Location: Mill 19 boardroom (space limited, please RSVP)
Abstract
How might robots humanely support people in their existing labor-intensive work? Contextualized in a specific construction scenario, this research addresses this question through the development of a robotic framework for adaptively supporting construction workers. Employing a “research by prototyping” approach informed by insights drawn from qualitative studies of an actual construction site, the framework mobilizes state-of-the-art Reinforcement Learning (RL) algorithms and includes five main technical elements, (1) robot hardware prototyping, (2) unstructured site perception and robot state estimation, (3) worker detection and tracking, (4) hierarchical motion planning, and (5) contextualized RL training and deployment. Realized through an actual physical robot prototype, a “work companion rover” for carpentry workers, the framework is examined both qualitatively and quantitatively in lab settings and on an actual construction site. Instead of purporting to replace human workers, this research shows how advanced computational and robotics techniques might be designed to adaptively support them. It advances the state-of-the-art in computational design by innovatively bringing together AI and robotics techniques to actual construction work contexts rather than idealized lab settings. In addition, it also contributes to the fields of robotics and AI by demonstrating a human-centered use of RL for supporting heavy manual labor, and showcasing a technology design process deeply informed by the social and material specificity of construction work.
Thesis Committee
Dr. Daniel Cardoso Llach (Advisor)
Associate Professor, Computational Design Track Chair
School of Architecture
Carnegie Mellon University
Dr. Jean Oh (Co-advisor)
Associate Research Professor
Robotics Institute
Carnegie Mellon University
Dr. Jieliang (Rodger) Luo
Principal AI Research Scientist
Autodesk Research, AI Lab
Category: News