Jianxun Li

Robust adaptive Kalman filtering for target tracking with unknown observation noise

The Kalman filter (KF) is widely used in the field of target tracking. In practical target tracking systems through, the observation noise is often unknown and characterized by heavier tails named outliers. That will affect the performance of target tracking seriously and even lead to filtering divergence. To overcome this problem, a novel robust Kalman filter (RKF) is proposed based on the maximum a posteriori (MAP) estimation to observation outliers. In addition, the adaptive estimate of observation noise variance R is also given based on the weighted correlation innovation (WCI) sequences of output of a steady state Kalman filter (SSKF). Finally, a robust adaptive Kalman filter (RAKF) algorithm is raised by implementing RKF and adaptive estimate of R simultaneously. The feasibility of the algorithm is demonstrated by an example of target tracking with simulation.

Interactive multiple Kalman filters based 3-D target tracking for two airborne platforms with bearings-only measurements

Target tracking by two airborne platforms with bearings-only measurements has obtained distinct interest recently. It is a nonlinear problem that Kalman filter (KF) can not be applied directly. Using extended Kalman filter (EKF) to solve this problem will get high computational burden involved because the Jacoby and Hessian matrices for the measuring equation should be calculated online. In this paper, the triangular ranging formula is derived for two bearings-only platforms first. Then the location error is proved to be zero mean Gaussian noises. This enables the traditional Kalman filter applicable. To further improve the tracking accuracy, interactive multiple model (IMM) is adopted to estimate the target state from the converted measurements. Simulation for both non-maneuvering and maneuvering target is also included to illustrate the proposed approach has high accuracy with much lower computational burden.

Sparse Wiener Chaos approximations of Zakai equation for nonlinear filtering

Sparse Wiener Chaos approximations of Zakai equation is considered. The objective is to optimize an approach to nonlinear filtering based on the Cameron-Martin version of Wiener Chaos Expansion(WCE). The error of the approximation is obtained. The main feature ofWiener chaos expansion is that it allows one to separate the computations involving the observations from those dealing only with the system parameters and to shift the latter off-line. The sparse truncation can reduce the number of the WCE coefficients dramatically while keeping the same asymptotic convergence rate as the simple truncation.

Path planning for unmanned aerial vehicles based on genetic programming

Path planning system is one of the key component for the unmanned aerial vehicles (UAVs) and mobile robots in modern operational systems used in all sorts of circumstances. Generally, genetic algorithm (GA) plays a big role in dealing with optimization problems. However, compared to GA, genetic programming (GP) displays better modeling and optimizing ability in path planning problem. GP is capable of dealing with UAV and mobile robot path planning problems. GP improves performance by utilizing generalized hierarchical computer programs and optimizing evolutionarily. This paper presents an optimized GP method which applies to path planning problem. Several special designed function and symbol operators are proposed and appended to the binary tree structure, as well as the redesigned decoding system. With the combination of selection and reproduction operation, the optimized GP accomplishes the design of path planning. By using the optimized GP method, experiment results display better fitness paths against GA method.

UAV path planning based on receding horizon control with adaptive strategy

Recently, researchers show great interest in Unmanned Aerial Vehicle(UAV) path planning problem due to development of artificial intelligence algorithms. However, the convergence of these kinds of heuristic methods can not be proved. Therefore, the UAV path planning problem has been modeled as a linear optimal control problem to ensure the convergence. But the computing time will increase exponentially as the scale of the problem enlarge. Hence, the Receding Horizon Control(RHC) is introduced to the problem to guarantee the efficiency. Nevertheless, how to choose a proper receding horizon span to keep balance between efficiency and accuracy becomes a new problem. An adaptive strategy is proposed to choose a proper parameter to meet the real-time requirement and fuel consuming in this paper. The strategy has better performance in the two simulation scenes, which indicates the effectiveness of the algorithm.

Multi-target tracking using the sign of innovations in wireless sensor networks

For wireless sensor networks, target tracking is an important application areas, but the communication consumption and the lower power resource should be taken into account. Currently, the wireless sensor networks for quantitative tracking method are more concentrate on single-target tracking. This paper presents a new method based on quantitative multi-target tracking. First, in order to meet the wireless sensor networks nodes have to save power energy, and reduce communication energy consumption requirements. We introduce a new joint probability data association using the sign of innovations (SOI-JPDA), this algorithm can afford lowest energy consumption, and the performance and complexity are very close to the standard joint probability data association algorithm. Second, we propose a distributed multi-target tracking algorithm based on SOI-JPDA in wireless sensor networks. Simulation results show that this algorithm can successfully track multiple targets with minimal power consumption in wireless sensor networks.

An analysis method of model reduction error based on H ∞ norm

Because of the limitation of linear controllers at the present stage, model reduction of complicated system still plays an important role in engineering application. During the research of electric load simulator with a high-order nonlinear model, no effective method is presented to evaluate the effect of different reduction ways. To solve this problem, an analysis method of model reduction error based on H∞ norm is proposed and applied in this paper. A synthesis error can be obtained by this method including linearization error and order reduction error. Compared with the figures in common use, the quantized result providing the basis for performance evaluation of model reduction is more intuitive and reliable. As the experimental verification, the superiority of Balanced Truncation used in the model reduction of electric load simulator is testified by this proposed method.

Face tracking with an Adaptive Adaboost-based Particle Filter

A novel algorithm, termed a Boosted Adaptive Particle Filter (AAPF), for integrated face detection and face tracking is proposed. The proposed algorithm is based on the synthesis of an adaptive particle filtering algorithm and the AdaBoost face detection algorithm. An Adaptive Particle Filter (AAPF), based on a new sampling technique, is proposed. The APF is shown to yield more accurate estimates of the proposal distribution than the standard Particle Filter thus enabling more accurate tracking in video sequences. In the proposed AAPF algorithm, the AdaBoost algorithm is used to detect faces in input image frames, the APF algorithm incorporate the detection result of AdaBoost algorithm to improve the proposal distribution of the particles. Experimental results show that the proposed AAPF algorithm provides a means for robust face detection and accurate face tracking under various tracking scenarios.

An improved TLD method based on color feature

Visual object tracking is a significant issue in the task of following a target in a stream of images. In this paper, we address this problem by proposing a novel parallel frame based on the original tracking-learning-detection method. The detector and the tracker in our algorithm are working simultaneously and output the candidate regions, then we perform analysis on these regions based on its color feature in HSV color space and the final position of target can be obtained naturally. Besides, this paper introduces BRISK into our tracker to alleviate the instability caused by target rotation and illumination variation. Extensive experimental results on massive benchmark datasets demonstrate that our algorithm has a crucial improvement over the original TLD and other state-of-the-art algorithms.

Improved saliency detection based on manifold ranking algorithm

Saliency detection is a fundamental problem in computational and cognitive sciences. Nowadays, graph-based methods are widely applied to saliency detection including manifold ranking(MR) method, which is shown to be fast and effective. However, because of only using a single feature and imperfect selection strategy for background seeds, MR has a poor performance in some circumstances. In order to complete more challenging saliency detection, an improved method based on manifold ranking algorithm is proposed in this paper. The adopted parallel architecture enables the final detection results generated with the combination of calculations from background seeds and foreground seeds. These seeds are quickly obtained by using priori information and the graph structure is optimized with adding three constraints. Moreover, the graph edge weights are computed by utilizing a adaptive local width parameter and measuring multi-features distance. Three-stage strategy is used to calculate the final saliency map. Experiment results on two large famous datasets demonstrate that the proposed method performs better comparing with MR and other state-of-the-art methods.