Excavators, as common construction machinery, involve certain risks in operation and driving. Operators work for long periods in outdoor environments like open-pit mines, facing significant physical demands. Builder[X] is dedicated to developing a remote control platform for excavators to reduce safety risks for operators and enhance working conditions.
However, there is still room for improvement in remote-controlled excavators. Firstly, due to camera angle limitations, operators cannot fully observe their surroundings on the monitor, making it difficult to accurately perceive slopes and inclines in 3D space, thus requiring careful attention to avoid collisions with surrounding objects. Secondly, for tasks involving high repetition such as digging, rotating, and loading, automation technologies can improve efficiency and reduce labor burdens.
To enhance the safety and user experience of remote-controlled excavators, acquiring a digital twin model of the excavator is essential. These models need to accurately reflect the excavator's geometric shape, connections, colors, and materials, and should be compatible with various robot simulation software for kinematic and dynamic simulations. However, obtaining corresponding digital twin models for excavators of different brands and sizes from various customers poses certain difficulties.
To address this challenge, we have developed an efficient and precise toolchain for building digital twin models of excavators. Initially, the 3DScanner App on a mobile phone is used to scan and capture the external shape and color information of the excavator model. Subsequently, open-source software such as MeshLab and CloudCompare are utilized to process point cloud models and measure key dimensions. The measured dimension parameters are then imported into a parameterized 3D model in SolidWorks for reconstruction. Finally, the model is exported in URDF format for simulation in robot simulation software like Gazebo and V-rep. The entire process from scanning the excavator to obtaining the URDF robot description file takes less than two hours, with modeling errors in critical components not exceeding five millimeters, and basic collision avoidance functionalities included.
This project aims to develop an accurate and efficient toolchain for excavator digital twins to meet the needs of Builder[x]. The toolchain is straightforward, highly efficient, has a low entry barrier, and requires low-cost and readily available equipment and software. It exhibits high adaptability to excavators of different sizes. Apart from adjustments needed in the scanning step to accommodate actual excavators, this toolchain can directly be applied to kinematic and dynamic simulations and visualizations of commercial excavators. Moreover, this toolchain can be expanded in the future for fine modeling of other construction machinery.
