标题可翻译为一个火辣的   滤波无用论 ？

贝叶斯大名鼎鼎主宰整个估计理论领域，滤波可以说是自卡尔曼以来控制界的宠爱，有人敢说他们无用？

Since the groundbreaking work of the Kalmanfilter in the 1960s, considerable effort has been devoted to various discretetime filters for dynamic state estimation, especially including dozens ofdifferent types of suboptimal implementations of the Bayes filters. This has been accompanied by the rapid development of simulation/approximation theories and technologies.The essence of filters is to make the best use of the observation information intime sequence based on the hidden Markov model, which however is advisable onlyunder the premise that the modeling errors are relatively small and that the approximation used is not too much. While admitting the success of filters in manycases, this study investigates the failure cases when they are in fact ineffective for state estimation. Several classic models have shown that the straightforward observation-only (O₂) inference that does not need modeling thestate dynamics can perform better (in terms of both accuracy and computing speed) for estimation than filters in certain cases. Special attention has beenpaid to quantitatively analyze when and why a filter will not outperform the O₂ inference from the information fusion perspective.

Thanks tothe rapid development of advanced sensors, the O₂ inference is notonly engineering friendly and computationally fast but can also be veryaccurate and reliable by fusing the information received from multiple sensors. The statistical attributesof the multi-sensor O₂ inference are analyzed and demonstrated through simulations.  In the situation with limited sensors, the O₂ approachcan work jointly with existing clutter filtering and data association algorithms for multi-target tracking in clutter environments. Given an adequate number of sensors, the O₂ approach can employ the multi-sensor datafusion to deal with clutter and can handle the very general multi-targettracking scenario with no background information.

=============================

For the detail of this work, please go to  http://arxiv.org/abs/1408.4636