Improved IBVS Algorithm of Robot Arm for Dynamic Tracking in Constrained Environments

摘要

This paper presents an advanced Image-Based Visual Servoing (IBVS) algorithm integrated with Model Predictive Control (MPC) to enhance the performance of robotic arms in dynamic target within complex environments. To address the limitations of conventional IBVS related to field-of-view constraints and occlusion sensitivity, two key contributions are introduced: First, we develop an MPC-based camera pose optimization strategy that extends the motion range of end-effectors while ensuring the stability of visual features, effectively preventing dynamic targets from leaving the field of view. Second, we implement an eye-in-hand vision system combined with MPC-driven online trajectory planning to mitigate hardware limitations and interference from occlusions. Simulation results demonstrate that the proposed algorithm significantly reduces target loss incidents during dynamic tracking compared to traditional IBVS methods. This advancement significantly enhances tracking robustness in constrained environments.