Automation for Transformable Production

As the fashion industry shifted toward on-demand orders, Transformable Production emerged as the manufacturing model for personalised clothing: high model variety, small batches, and a factory floor that must reconfigure constantly. This project treated the floor as a heterogeneous multi-robot system — static robots, mobile transporters, and human workers — and asked how tasks should be scheduled and allocated across it. The work ran from scheduling heuristics through deep reinforcement learning to multi-agent pickup and delivery, and it is where the coordination questions in the current program were first posed concretely.