A Method of Autonomous Assembly of Large Spacecraft Components Using Robot

When the large components of a spacecraft are installed in a narrow space, there are some problems such as a limited visual field and an invisible target position. The traditional hoisting approach can't accurately adjust the position and attitude of the component, and is prone to cause collision. A path planning method for robot assisted assembly using binocular vision localization is proposed. The binocular vision system is employed to precisely localize the geometric features of installation positions on the spacecraft, based on the 3D model of assembly elements such as the spacecraft, components to be installed and the robot. The relative position and attitude relationship between the robot and the spacecraft can be computed using the measured results, so as to construct the virtual environment of the spacecraft assembly scenes. In the virtual environment, geometric constraints in the assembly process are recognized and a collision detection is implemented using the axis-aligned bounding box (AABB) method. After that, the probabilistic roadmap method is adopted to generate a collision-free path, and then executable assembly sequences for robot are produced through off-line programming. An assembly experiment is carried out for a laser altimeter instrument of a spacecraft. The installation of the large component is completed in a narrow concave cabin without collision. Results show that the proposed method can localize the precise position of thread holes, rapidly generate a collision-free assembly path, and control the robot to safely and efficiently accomplish large component installations in the narrow space of a spacecraft.