Abstract
In these days, motion capture cameras can achieve localization of moving objects with high accuracy. However, it has some problems, for example, it may fail to measurement due to occlusion and the quantization error. In addition, the measurement data with noise or lack of data disturbs the control performance of visual feedback systems. Thus, it is necessary to estimate states suppressing the influence of the noise and occlusion. In this study, we achieve high accuracy of state estimation using Moving Horizon Estimation (MHE) which is able to consider measurement data during the finite-time. We implement MHE with the actual experimental system of 3D static cameras and show the effectiveness for visual feedback control of MHE.
