Abstract
This article introduces a velocity observer-based adaptive visual servoing control scheme for closed-architecture robotic systems. A novel dual-observer framework, consisting of an image-space observer and a joint-space sliding observer, is developed to avoid the usage of both image-space velocity and joint velocity. Leveraging this framework, an adaptive visual servoing loop control solution is designed, which dynamically generates velocities estimations-based joint position or velocity commands for the torque control loop. Rigorous Lyapunov-based stability analysis demonstrates the asymptotic convergence of both image-space tracking error and observation error, without necessitating modifications to the inner torque controller. This feature renders the proposed method particularly suitable for closed-architecture robotic systems, where the torque controller is unmodifiable due to safety/commercial strategy, and restricted to accepting only joint position or velocity commands. Experimental validation on a robotic manipulator demonstrates the superior performance of the proposed approach.
Funding Information
  • Central Guidance on Local Science and Technology Development Fund of Hebei Province (246Z1812G)
  • National Natural Science Foundation of China (62203380)
  • S&T Program of Hebei (2023HBQZYCSB007)
  • Natural Science Foundation of Hebei Province (F2023203079)
  • National Natural Science Foundation of China (52205060)