[2F1-GS-10] AI application:

[2F1-GS-10-03] Collaborative Control of Human and Reinforcement Learning in Overhead Crane Operation

〇Ryuji Shiokawa¹, Sachiyo Arai¹ (1. Chiba university)

Keywords:Reinforcement Learning, crane

Heavy machinery such as cranes used in various fields like civil engineering, construction, mining, and agriculture require a substantial amount of time to master their operation, which creates a strong demand for autonomous control. However, the sites where these machines are deployed are often unpaved and cluttered with various parts and debris. Thus, achieving fully autonomous operation would demand robustness to adapt to complex work environments and enhancements on the operational side, such as introducing numerous sensors and extensive environmental preparation.
This research aims to develop a collaborative agent system that can stably guide even inexperienced operators through the control process. Specifically, it focuses on transporting a load with an overhead crane, aiming to accurately and smoothly bring the load to a target location while suppressing any sway that occurs from start to stop. The proposed method achieves collaborative control between machines and humans by combining reinforcement learning–based sway control with human-guided position control.