Balakumar Balasingam

Efficient distributed resampling for particle filters

In particle filtering, resampling is the only step that cannot be fully parallelized. Recently, we have proposed algorithms for distributed resampling implemented on architectures with concurrent processing elements (PEs). The objective of distributed resampling is to reduce the communication among the PEs while not compromising the performance of the particle filter. An additional objective for implementation is to reduce the communication among the PEs. In this paper, we report an improved version of the distributed resampling algorithm that optimally selects the particles for communication between the PEs of the distributed scheme. Computer simulations are provided that demonstrate the improved performance of the proposed algorithm.

Mathematical modeling and parameter estimation of blood pressure oscillometric waveform

In this paper, a mathematical model for the blood pressure oscillometric waveform (OMW) is developed and a statespace approach using the extended Kalman filter (EKF) is proposed to adaptively estimate and track parameters of clinical interest. The OMW model is driven by a previously proposed pressure-lumen area model of the artery under the deflating cuff. The arterial lumen area is a function of vessel properties, the cuff pressure, and the arterial pressure. Over the deflation period, the arterial pressure causes lumen area oscillations while the deflating cuff pressure adds a slow-varying component to these oscillations. In the previous literature, it has been demonstrated that the oscillometric pulses are proportional to the arterial area oscillations. In this paper, the OMW is modeled as the difference between the whole lumen area model and the slow-varying component of the lumen area caused by the deflating cuff pressure. The OMW model is then represented in the statespace and the extended Kalman filter (EKF) is incorporated to estimate and track the time-varying model parameters during the cuff deflation period. The parameter tracking performance of the EKF is evaluated on a simulated noisy OMW.

Battery charging optimization for OCV-resistance equivalent circuit model

In this paper, we present a closed-form solution to the problem of optimally charging a Li-ion battery. The objective function is considered as a combination of two cost functions: time-to-charge (TTC) and energy losses (EL). For the case where cost function is a combination of TTC and EL, the optimal charging strategy is a Constant Current-Constant Voltage (CC-CV) policy with the value of the current in the CC stage being a function of the ratio of weighting on TTC and EL and of the resistance of the battery. The case where the cost function is a weighted sum of TTC, EL and a temperature rise index (TRI) is also considered and an analytical solution for the problem is derived. This analytical solution can be approximated by a CC-CV with the value of current in the CC stage being a function of ratio of weighting on TTC and EL, resistance of the battery and the effective thermal resistance. The effects of weights in the objective function on the optimal charging profile is discussed and the behavior of different kinds of commercial batteries are analyzed.

Robust battery fuel gauge algorithm development, part 0: Normalized OCV modeling approach

The open circuit voltage (OCV) characterization of Li-ion batteries as it applies to battery fuel gauging (BFG) in portable applications is considered in this paper. Accurate knowledge of the nonlinear relationship between the OCV and the state of charge (SOC) is required for adaptive SOC tracking during battery usage. BFG in portable applications requires this OCV-SOC characterization to be defined with a minimum number of parameters. With the help of OCV characterization data collected from 34 battery cells each at 16 different temperatures ranging from -25°C to 50°C, we present a novel normalized OCV modeling approach that dramatically reduces the number of OCV-SOC parameters and as a result simplifies and generalizes the BFG across temperatures and aging.

Robust battery fuel gauge algorithm development, part 3: State of charge tracking

In this paper, we present a novel SOC tracking algorithm for Li-ion batteries. The proposed approach employs a voltage drop model that avoid the need for modeling the hysteresis effect in the battery. Our proposed model results in a novel reduced order (single state) filtering for SOC tracking where no additional variables need to be tracked regardless of the level of complexity of the battery equivalent model. We identify the presence of correlated noise that has been so far ignored in the literature and use this for improved SOC tracking. The proposed approach performs within 1% or better SOC tracking accuracy based on both simulated as well as HIL evaluations.

Multiple target tracking from images using the maximum likelihood HPMHT

In this paper, we address the problem of passive tracking of multiple targets with the help of images obtained from passive infrared (IR) platforms. Conventional approaches to this problem, which involve thresholding, measurement detection, data association and filtering, encounter problems due to target energy being spread across multiple cells of the IR imagery. A histogram based probabilistic multi-hypothesis tracking (H-PMHT) approach provides an automatic means of modeling targets that are spread in multiple cells in the imaging sensor(s) by relaxing the need for hard decisions on measurement detection and data association. Further, we generalize the conventional HPMHT by adding an extra layer of EM iteration that yields the maximum likelihood (ML) estimate of the number of targets. With the help of simulated focal plane array (FPA) images, we demonstrate the applicability of MLHPMHT for enumerating and tracking multiple targets.

Robust battery fuel gauge algorithm development, part 1: Online parameter estimation

In this paper, we present a novel voltage drop model for battery SOC tracking and develop a robust, realtime approach for model parameter estimation. The proposed model avoids the need to model hysteresis voltage that hard to model and estimate in practical applications. Another advantage of the proposed voltage drop model is that the parameters of the model is estimated linearly, regardless of the model complexity, i.e., number of RC elements considered in the model. We identify the presence of correlated noise that has been so far ignored in the literature and use it to enhance the accuracy of model identification. The proposed parameter approach enables a robust initialization/re-initialization strategy for continuous operation of the battery fuel gauge (BFG). The performance of the online parameter estimation scheme was evaluated through objective measures.

Robust battery fuel gauge algorithm development, part 2: Online battery-capacity estimation

In this paper we present an approach for robust, real time capacity estimation in Li-ion batteries. The proposed capacity estimation scheme has the following novel features: it employes total least squares (TLS) estimation in order to account for uncertainties in both model and the observations in capacity estimation. The TLS method can adaptively track changes in battery capacity. We propose a second approach to estimate battery capacity by exploiting rest states in the battery. This approach is devised to minimize the effect of hysteresis in capacity estimation. Finally, we propose a novel approach for optimally fusing capacity estimates obtained through different methods. We demonstrate the performance of the algorithm through objective experiments.

On the use of hidden Markov models for gaze pattern modeling

Some of the conventional metrics derived from gaze patterns (on computer screens) to study visual attention, engagement and fatigue are saccade counts, nearest neighbor index (NNI) and duration of dwells/fixations. Each of these metrics has drawbacks in modeling the behavior of gaze patterns; one such drawback comes from the fact that some portions on the screen are not as important as some other portions on the screen. This is addressed by computing the eye gaze metrics corresponding to important areas of interest (AOI) on the screen. There are some challenges in developing accurate AOI based metrics: firstly, the definition of AOI is always fuzzy; secondly, it is possible that the AOI may change adaptively over time. Hence, there is a need to introduce eye-gaze metrics that are aware of the AOI in the field of view; at the same time, the new metrics should be able to automatically select the AOI based on the nature of the gazes. In this paper, we propose a novel way of computing NNI based on continuous hidden Markov models (HMM) that model the gazes as 2D Gaussian observations (x-y coordinates of the gaze) with the mean at the center of the AOI and covariance that is related to the concentration of gazes. The proposed modeling allows us to accurately compute the NNI metric in the presence of multiple, undefined AOI on the screen in the presence of intermittent casual gazing that is modeled as random gazes on the screen.

Cognitive context detection using pupillary measurements

In this paper, we demonstrate the use of pupillary measurements as indices of cognitive workload. We analyze the pupillary data of twenty individuals engaged in a simulated Unmanned Aerial System (UAS) operation in order to understand and characterize the behavior of pupil dilation under varying task load (i.e., workload) levels. We present three metrics that can be employed as real-time indices of cognitive workload. In addition, we develop a predictive system utilizing the pupillary metrics to demonstrate cognitive context detection within simulated supervisory control of UAS. Further, we use pupillary data collected concurrently from the left and right eye and present comparative results of the use of separate vs. combined pupillary data for detecting cognitive context.