Abstract: At present, the installation of large spaceborne equipment is usually performed manually. That is because the equipment is expensive and fragile, in addition the assembly process is carried out in a semi-closed space and under non-direct vision condition. Assemble workers operate under human observation and oral instructions. Therefore, the process is usually laborious and inefficient. To overcome this difficulty, an intelligent robotic assembly method based on visual guidance is proposed. First, machine learning and binocular vision techniques are employed to achieve accurate position measurement between assemblies. Then, motion mapping between the vision space and robot space is calculated to transfer the required motion into the robot space by hand-eye calibration. Next, assembly path planning is performed by analyzing the interference condition between two assemblies. Finally, the assembly process is visualized using virtual reality techniques for safety monitoring. The whole assembly process can be completed automatically in less than several minutes. Assembly experiments using satellite and equipment models are carried out, and the assembly error is evaluated to be less than 0.3 mm, confirming the effectiveness of the proposed method.

Key words: robotic assembly, spaceborne equipment, trajectory planning, virtual reality, visual guidance