Visual Motion Tracking with Full Adaptive Extended Kalman Filter: An Experimental Study
| Authors: | Lippiello Vincenzo, Università degli Studi di Napoli FEDERICO II, Italy
Villani Luigi, Università degli Studi di Napoli FEDERICO II, Italy
Siciliano Bruno, Università degli Studi di Napoli FEDERICO II, Italy |
|---|
| Topic: | 4.3 Robotics |
|---|
| Session: | Autonomous Robots and Systems |
|---|
| Keywords: | Pose Estimation, Vision, Motion Tracking, Visual Servoing, Adaptive Extended Kalman Filter |
|---|
Abstract
An algorithm for real-time estimation of the pose of a movingobject of known geometry is considered. The algorithm is based ona discrete-time Extended Kalman Filter which computes the objectpose on the basis of visual measurements of the object features.The robustness of the algorithm with respect to measurement noiseand modelling errors is improved by considering a full adaptiveversion of the Extended Kalman Filter. A complete experimentalstudy is presented to test the performance and feasibility of theapproach.
