A New Prediction Algorithm for Embedded Real-Time Applications

Luis GRACIA  Carlos PEREZ-VIDAL  

Publication
IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences   Vol.E93-A   No.1   pp.272-280
Publication Date: 2010/01/01
Online ISSN: 1745-1337
DOI: 10.1587/transfun.E93.A.272
Print ISSN: 0916-8508
Type of Manuscript: PAPER
Category: Systems and Control
Keyword: 
filtering,  prediction,  parallel computting,  visual servoing,  

Full Text: PDF>>
Buy this Article




Summary: 
In this research a new prediction algorithm based on a Fuzzy Mix of Filters (FMF) is developed. The use of a fuzzy mix is a good solution because it makes intuitive the difficult design task of combining several types of filters, so that the outputs of the filters that work closer to their optimal behavior have higher influence in the predicted values. Therefore the FMF adapts, according to the motion of the tracked object or target, the filter weights to reduce the estimation error. The paper develops the theory about the FMF and uses it for applications with hard real-time requirements. The improvement of the proposed FMF is shown in simulation and an implementation on a parallel processor (FPGA) is presented. As a practical application of the FMF, experimental results are provided for a visual servoing task.