Research Seminar on AI: Learning-based Optimization in Manufacturing

 

Dienstag, 13.07.2021, 16.00 Uhr

Vladimir Samsonov will introduce the potentials and current challenges of learning-based optimization methods in manufacturing, as well as talk about adopting reinforcement learning (RL) for two optimization applications on the shop floor and machine levels developed within his work. The part of the talk concentrating on the shop-floor level optimization will look into the use of RL for the job shop scheduling and specific designs of the state- and action-space as well as the reward signal design. The machine-level optimization will cover finding the optimal positioning and orientation of the workpieces for a 5-axis milling process, focusing on the use of unsupervised embeddings of 3D workpiece geometries, the subsequent use of the RL-based optimization and comparison to alternative meta-heuristic solutions.

Since 2017 Vladimir Samsonov is working on the applied research of Artificial Intelligence in manufacturing at the Institute of Information Management (IMA) of the RWTH Aachen University. Before starting his PhD, he studied mechanical engineering at Bauman Moscow State Technical University, Industrial Engineering at Hamburg University of Technology and worked as a manufacturing engineer in the aerospace industry.