Database-driven model predictive control system for online adaptation of an autonomous excavator to environmental conditions
Tomofumi Okada, Toru Yamamoto, Takayuki Doi, Kazushige Koiwai, Koji Yamashita
Abstract
This paper presents the design of a database-driven model predictive control (DD-MPC) system for the online adaptation of autonomous excavators to environmental conditions. Control systems for autonomous excavators should consider environmental conditions as these affect their performance for a given excavation operation. Moreover, these conditions may change during operation. MPC was performed using an excavator-environment interaction model, which was estimated online using DD-Modeling to represent changes in environmental conditions. The target excavation trajectory was modified by predicting excavation motion using MPC and decision based on the prediction to complete a given excavation operation regardless of the environmental conditions. The proposed system was experimentally verified using a radio-controlled excavator, and it was confirmed that a given operation could be completed by adapting to environmental conditions.