Image-Based Adaptive Stiffness Visual Servoing Control for Manipulators in Feature Space With Prescribed Performance

With the rapid progression of industry, robotic assembly technologies have garnered significant attention. To enable robots to perform assembly tasks autonomously and intelligently in unstructured environments without relying on precise environmental modeling, many robotic systems incorporate external perception sensors. Among these, the integration of vision and force sensors proves to be both cost-effective and complementary, enhancing system safety and task success by providing real-time sensory feedback. This article proposes a general framework for image-based variable admittance control, which couples force and visual information within the image feature space, thereby avoiding control limitations arising from inconsistent actuation layers. The variable admittance control adjusts stiffness parameters dynamically across different task phases, enhancing flexibility and interaction with the environment. To enhance the transient and steady-state performances of the system, the control design utilizes an asymmetric barrier Lyapunov function with a specified error bound as a time-varying constraint. The asymmetric barrier Lyapunov function enforces a strict predefined range for feature coordinate errors, thereby improving both the transient and steady-state performances of the system. The proposed image-based variable admittance control is experimentally validated. The results show that the proposed framework improves the system’s transient/steady-state response, safety, and environmental interaction.