Investigation of Multi-Stage Visual Servoing in the context of autonomous assembly

This paper introduces a novel strategy, Multi -Stage Based Visual Servoing (MSBVS), to tackle the significant challenge of planning complex trajectories in robotic systems without hand-eye calibration, particularly in some assembly tasks. The MSBVS strategy decomposes a complete visual servoing process into three distinct phases, each characterized by unique functionalities. By assigning specific image features and control objectives to each phase, the strategy indirectly facilitates the planning and tracking complex trajectories. Furthermore, a novel image feature, the vanishing angle, is introduced, enhancing the precision in representing an object's spatial position. Subsequently, a robotic vision system is established, and real-time simulations and experiments are performed to verify the efficacy of the MSBVS strategy. The results show that MSBVS efficiently plans and executes complex trajectories in robotic assembly tasks.