Dynamic OHT routing using distance approximation based on deep neural network

Abstract

Recently, semiconductor fabrication plant has become larger and complex to meet emerging market demand. This trend requires a massive number of transportation moves in a huge rail network. The change of fab environment brought two challenges: dynamic routing and scalable routing. In this thesis, a dynamic OHT routing using distance approximation based on deep neural network is proposed to overcome these challenges. Proposed method consists of local path finding model using shortest path finding approach, and global distance approximation model using deep learning approach. We report our computational results using high fidelity simulation experiments. In experiments, proposed method shows better performance than other existing routing policy and enough scalability .