다물체 동역학 모델 기반 6 자유도 굴삭기의 작업 경로 계획

Abstract

This study aimed to plan a work path for a 6-DOF excavator and evaluate the applicability of reinforcementlearning. The excavation work caused vibrations in main components of the tilt-rotator, which were experimentallymeasured. Weak components were identified through vibration experimental and signal processing. Next, a multi-bodydynamics model correlated with vibration data was used to generate a stress data set for weak components. After that,various excavator agents were compared in a reinforcement learning environment and the validity of applying the stress data set was verified. Finally, this study analyzed work path planning derived from reinforcement learning and discussed results. Results of this study can be used as basic data for developing operating guides and operating assistance systems to help existing equipment users operate their equipment more safely and efficiently.