The manual assembly of objects characterized by intricate geometry and materials, such as concentrator photo-voltaics solar panel units, often results in suboptimal accuracy, efficiency, and throughput. This paper responds to a genuine industrial imperative by proposing the development of a robotic system for the automated assembly of precision industrial components. To realize this goal, a dual robotic system is employed as the manipulation subsystem, complemented by the camera serving as computer vision (CV) subsystem. To address the crucial need for precise object 3D localization during assembly, an enhanced pose estimation algorithm grounded in convolutional neural networks (CNN) is proposed. By incorporating shallow information into the feature fusion network, precise object pose estimation is guaranteed. In instances of inevitable occlusion, a process planning scheme is founded on cooperative manipulation, leveraging the unique characteristics of different robots for pose refinement and flexible operation. Taking the solar-cell of the concentrator as an example object, customized datasets are established and experiments are conducted within a predefined workspace. The calibration accuracy, offline and online performance of the proposed pose estimation algorithm and practical robotic gripping capability are evaluated. Results underscore the effectiveness of the developed robotic system and its potential for industrial precision assembly.