Lars Quaedvlieg

  Faculty of Science and Engineering | Bachelor Data Science and Artificial Intelligence

"Multi-Agent Reinforcement Learning With Graph Neural Networks For Online Multi-Hoist Scheduling"

Lars' elevator pitch
“The multi-hoist scheduling arises when multiple hoists are used to lift loads. The goal is to determine when to use each hoist so that the loads are lifted in the most efficient manner. Unfortunately, it is extremely computationally expensive to compute efficient ways to solve this problem. In this thesis, I apply reinforcement learning to the problem, which is a field in artificial intelligence that allows machines to learn through experience. I propose an approach that computes an efficient solution to the problem quickly, by making hoists communicate with each other and learn with the experiences they observe.”