POSITION-BASED VISUAL SERVOING IN ROBOTIC CAPTURE OF MOVING TARGET ENHANCED BY KALMAN FILTER

The paper develops a Kalman filter (KF)-enhanced position-based visual servo control strategy for autonomous robotic capture of a moving target with an eye-in-hand-configured robotic manipulator.A dual KF scheme is developed to process noisy imaging data and track, intercept, and affect a smooth capture of the target.The first KF concerns the image processing errors and temporary loss of lock of target to enhance tracking robustness while the second KF processes noises resulting from camera's residual vibration due to manipulator's joint flexibility.This framework not only enhances pose and velocity estimation from noisy images, but also provides smooth pose and velocity estimates for the robotic control system, which improves the efficiency of tracking and capturing of a moving target.Furthermore, a composite index of pre-capture measures and a threshold logic function are introduced to activate an automatic grasping of target.Experimental results show that faster and smoother captures of a moving target have been achieved with the proposed control strategy.