Special Offer: Get 50% off your first 2 months when you do one of the following
Personalized offer codes will be given in each session
1605117383-db4a1c4744efb925
Name
Noah Klarmann
Bio
Noah Klarmann received the Master of Science degree in Chemical Engineering at the Technical University of Berlin in 2014. Afterwards, he started as a research assistant at the Chair of Thermodynamics at the Technical University of Munich in 2015. During this time, he developed novel modeling strategies for turbulent combustion in the context of computational fluid dynamics. He received a PhD (summa cum laude) under the supervision of Prof. Dr.-Ing. Thomas Sattelmayer in 2019. In parallel to the doctoral program, Noah Klarmann started to study Computer Science with focus on AI at Technical University of Munich in 2017. In 2019, he joined the Chair of Robotics, Artificial Intelligence and Real-time Systems as a post-doctoral researcher. In his role, he is part of the AI4DI project that focuses on the training of digital twins of industrial robots in virtual environments.

Abstract of Webinar talk: The imminent convergence of artificial intelligence with the industry is one of the most interesting topics of our time. Adopting new methods such as large-scale machine learning has the power to fundamentally change industrial processes and will pave the way for new business models. While the adoption of supervised and unsupervised methods is widespread in many sectors, the paradigm of reinforcement learning is often overlooked. This does not reflect the impressive progress in this area, such as the defeat of world champions in several disciplines without prior domain knowledge (AlphaGo, OpenAI Five, AlphaStar). The talk introduces the exciting field of reinforcement learning as a powerful tool to solve complex sequential decision-making problems in the industrial context. Beyond the discussion of general concepts, two possible applications of reinforcement
learning for industrial applications are presented: (1) Continuous control of industrial manipulators and (2) dynamic production scheduling. Both use cases are examined by employing simplified simulation environments that allow to study basic characteristics.