Path Planning for Construction Machines by Offline Reinforcement Learning

Controlling bulldozers at construction sites requires advanced skills. However, training skilled operators is time consuming and expensive. Additionally, the shortage of young workers is problematic. To address these issues, automation of construction machines using artificial intelligence has been studied, which has yielded positive results. Furthermore, using data collected from construction sites, reducing learning time and improving work accuracy is possible. Therefore, in this paper, we propose a path planning method for heavy equipment using offline reinforcement learning, leveraging existing datasets.