실내 디지털 트윈 구축을 위한 로봇 조작 정보 모델 및 시뮬레이션 설계

초록

This study proposes a robotic manipulation information model and simulation design for constructing indoor digital twin environments. The model defines the structure and control flow of key components such as a robot, suction gripper, conveyor system, and 3D vision sensors, ensuring modularity and scalability. To overcome the limitations of CAD-based assets, all objects were converted to the USD format to support extensible attribute representation. The architecture integrates ROS2 messaging with the Action Graph system in NVIDIA Isaac Sim to enable real-time data exchange and control. Trajectory planning is performed using MoveIt 2, with commands transmitted via ROS2 topics and refined through feedback using the Articulation Controller. The proposed design was validated through a simulated postal logistics depalletizing task and provides a reusable foundation for modeling robotic manipulation systems. It supports simulation-based development and evaluation of logistics automation tasks.

키워드