Real-time Filtering Research Articles

Extended Kalman filter steady gain scheduling using k-means clustering

This paper studies the gain scheduling problem for extended Kalman filter (EKF). To save throughput, the steady gain is usually used for Kalman Filter. In the context of EKF, there is no universal steady gain. In this paper, we propose a methodology to schedule the steady gain for EKF. The idea is to offline linearise the nonlinear model at various operating points, and for each of the linearised systems, to compute the steady gain corresponding to conventional Kalman filter by solving the corresponding algebraic Riccati equation. The operating space is then divided into multiple zones, through k-means clustering algorithm, so that within zone, the steady state gains are close to each other. For real time filtering, the centroid of each zone is used as gain for correction step, instead of computing the time-varying gain online, hence saving throughput. We demonstrate the proposed methodology in a two-state nonlinear system.

Adaptive smartphone-based sensor fusion for estimating competitive rowing kinematic metrics.

Competitive rowing highly values boat position and velocity data for real-time feedback during training, racing and post-training analysis. The ubiquity of smartphones with embedded position (GPS) and motion (accelerometer) sensors motivates their possible use in these tasks. In this paper, we investigate the use of two real-time digital filters to achieve highly accurate yet reasonably priced measurements of boat speed and distance traveled. Both filters combine acceleration and location data to estimate boat distance and speed; the first using a complementary frequency response-based filter technique, the second with a Kalman filter formalism that includes adaptive, real-time estimates of effective accelerometer bias. The estimates of distance and speed from both filters were validated and compared with accurate reference data from a differential GPS system with better than 1 cm precision and a 5 Hz update rate, in experiments using two subjects (an experienced club-level rower and an elite rower) in two different boats on a 300 m course. Compared with single channel (smartphone GPS only) measures of distance and speed, the complementary filter improved the accuracy and precision of boat speed, boat distance traveled, and distance per stroke by 44%, 42%, and 73%, respectively, while the Kalman filter improved the accuracy and precision of boat speed, boat distance traveled, and distance per stroke by 48%, 22%, and 82%, respectively. Both filters demonstrate promise as general purpose methods to substantially improve estimates of important rowing performance metrics.

A monitorable and renewable pollution filter based on graphene nanoplatelets

This paper deals with the fabrication, modeling and experimental characterization of a monitorable and renewable graphene-based pollution filter. The main goal is to demonstrate a method to monitor the status of such a filter in real time during its operating phases: pollutant adsorption, saturation, and regeneration. The filter is realized by a disk of pressed graphene nanoplatelets. This is a low-cost type of graphene which has recently drawn great interest due to its potential use in large scale industrial production. Here the nanomaterial is obtained through the exfoliation method assisted by microwave irradiation, by exploiting the thermal expansion of commercial intercalated graphite, according to a low-cost and ecologically friendly procedure. The filter is used here to adsorb acetonitrile, a toxic water-soluble organic compound that is present in some industrial solvents and paints. The monitoring method is based on the interpretation of the time variation of the electrical impedance measured during filter operation. There are two main results of the paper: Firstly, the graphene filter is shown to be effective in adsorbing the above pollutant, with the additional feature of being fully renewable: all the pollutant can be removed from the filter without the need of costly physical or chemical processes. Secondly, monitoring of the time-evolution of the electrical impedance allows efficient detection of the different phases of the filter life cycle: clean, polluted, saturated and regenerated. This feature is of potential interest since it enables the predictive maintenance of such filters.

Fast Ground Filtering of Airborne LiDAR Data Based on Iterative Scan-Line Spline Interpolation

Over the last two decades, a wide range of applications have been developed from Light Detection and Ranging (LiDAR) point clouds. Most LiDAR-derived products require the distinction between ground and non-ground points. Because of this, ground filtering its being one of the most studied topics in the literature and robust methods are nowadays available. However, these methods have been designed to work with offline data and they are generally not well suited for real-time scenarios. Aiming to address this issue, this paper proposes an efficient method for ground filtering of airborne LiDAR data based on scan-line processing. In our proposal, an iterative 1-D spline interpolation is performed in each scan line sequentially. The final spline knots of a scan line are taken into account for the next scan line, so that valuable 2-D information is also considered without compromising computational efficiency. Points are labelled into ground and non-ground by analysing their residuals to the final spline. When tested against synthetic ground truth, the method yields a mean kappa value of 88.59% and a mean total error of 0.50%. Experiments with real data also show satisfactory results under visual inspection. Performance tests on a workstation show that the method can process up to 1 million points per second. The original implementation was ported into a low-cost development board to demonstrate its feasibility to run in embedded systems, where throughput was improved by using programmable logic hardware acceleration. Analysis shows that real-time filtering is possible in a high-end board prototype, as it can process the amount of points per second that current lightweight scanners acquire with low-energy consumption.

A Quantitative Validation of the Control Banding Nanotool.

Eleven years (by publication) years after the development and application of the control banding (CB) Nanotool for the qualitative assessment and control of engineered nanoparticles (ENP), there remains no quantitative gold standard to serve as an alternative to the qualitative assessment. Many CB models have been developed during the years subsequent to the initial development of the CB Nanotool and the literature continues to blossom with comparisons and applications of these various tools; however, these developments have hitherto been made in the absence of validating and verifying their effectiveness using existing, albeit limited, quantitative methods. This paper reviews the existing literature on the CB Nanotool to evaluate its effectiveness in a variety of settings and presents a summary of qualitative and quantitative information from its application in a broad range of ENP handling activities performed in two different research institutions. A total of 28 ENP activities were assessed using the CB Nanotool (Version 2.0). Due to the lack of guidance on a single exposure assessment methodology, a combination of real-time monitoring, filter analysis, and microscopic analysis was used to assess various quantitative metrics, including mass concentration, particle number concentration, and particle speciation. All the results indicated that the control outcomes from the CB Nanotool qualitative assessment were sufficient to prevent workers from being exposed to ENP at levels beyond established exposure limits or background levels. These data represent an independent quantitative validation of CB Nanotool risk level outcomes and give further credence to the use of the CB Nanotool to effectively control worker exposures in the absence of quantitative air monitoring results.

Phenotyping tools for genetic improvement of table grapes in Chile

Although Chile is the world leader for table-grape export by volume, the renewal of table-grape cultivars has just recently taken off. Among these, 'Maylen' ('Iniagrape-One'), a mid-season black cultivar from the Chilean Breeding Program at INIA, stands out for its flavor and excellent postharvest life. In order to accelerate the development of new table-grape cultivars under local conditions, we have developed two software tools to improve the phenotyping of fruit quality characteristics either in the laboratory or in the field. The first of these tools, Vinetracker, is a smartphone application that replaces the use of field blocks: it combines a cloud-located database of the breeding program with the mobile network capacity to allow simultaneous evaluation of vine or fruit phenotypes, resulting in the reduction of cost, time and involuntary errors. It also allows the user to filter and consult data in real time. The second of these tools, Berry Analyzer, is a digital image-analysis tool for fruit phenotyping that aims at fast retrieval of information from berries, such as the polar and equatorial diameters, shape and color, in a small fraction of the time needed in comparison with standard tools or image software that requires human intervention.

Field Sensor Bias Calibration With Angular-Rate Sensors: Theory and Experimental Evaluation With Application to Magnetometer Calibration

Field sensors, such as magnetometers and accelerometers, are widely used sensors for attitude estimation, yet their accuracy is limited by sensor measurement bias. This paper reports a novel methodology for estimating the sensor bias of three-axis field sensors. Our approach employs three-axis angular velocity measurements from an angular-rate gyroscope to estimate the three-axis field sensor measurement bias that, when properly calibrated, can significantly improve attitude estimation. We report three methods implementing this approach based on batch linear least squares, real-time Kalman filter, and real-time adaptive identification. Assuming the field is constant, our methods impose less restrictive conditions for the movements of the instrument required for calibration than previously reported methods, do not require a priori knowledge of the field (e.g., the magnitude of the local magnetic field) or the attitude of the instrument, and also ensure convergence for the estimated parameters. The proposed methods are evaluated and compared with the previously reported methods with numerical simulation and in a comparative laboratory and field experimental evaluation with the sensors onboard an underwater robot vehicle. Finally, as an application example of the magnetometer bias calibration, the proposed methods are used to improve the estimation of the position of an underwater vehicle in Monterey Bay at 2800 in depth.

Real-Time Local Noise Filter in 3-D Visualization of CT Data

Removing noise in computer tomography (CT) data for real-time 3D visualization is vital to improving the quality of the final display. However, the CT noise cannot be removed by straight averaging because the noise has a broadband spatial frequency that is overlapping with the interesting signal frequencies. To improve the display of structures and features contained in the data, we present spatially variant filtering that performs averaging of sub-regions around a central region. We compare our filter with four other similar spatially variant filters regarding entropy and processing time. The results demonstrate significant improvement of the visual quality with processing time still within the millisecond range.

Low-cost didactic platform for real-time adaptive filtering: Application on noise cancellation

Didactic platforms, used in real-time digital signal processing courses, are generally dedicated digital signal processors or field-programmable gate arrays. These devices are expensive and difficult to program, preventing their widespread use in signal processing courses. On the other hand, the new technology of digital signal controllers, microcontrollers with floating point and mathematical operations, can reduce the cost of dedicated platforms for real-time digital signal processors, facilitating the development of digital signal processors projects and educational applications, such as teaching adaptive filters. Here, we present a low-cost didactic platform for developing real-time adaptive filters using the digital signal processors hardware based on the ARM Cortex-M7 processor. We present the theoretical aspects of the least mean squares and normalized least mean squares algorithms and an experimental script to help students learn real-time adaptive filters. We also describe the platform structure and the performance measurement, in terms of mean square error, signal-to-noise ratio, and computational efficiency. Finally, we present a brief discussion on the use of this platform in classes and the improvement in the student engagement and attendance.

Model-based spinal deformation compensation in robot-assisted decompressive laminectomy

Decompressive laminectomy is a common operation for the treatment of lumbar spinal stenosis. A high-speed burr and piezosurgery are often used to remove the lamina to relieve compressed nerves. However, surgeons need to control the cutting margin of the lamina based on experience, and the remaining laminar thickness is difficult to quantify before image measurements. In other surgeries, robot-assisted grinding guided by navigation can make operations accurate, but the deformation of bone tissue caused by instrument interaction cannot be ignored, especially for spinal surgery. Because of the elasticity of the intervertebral disk, the spine deforms under external force, which leads to deviation from the preoperative trajectory for the robot. This paper proposes a model-based compensation method for spinal deformation during decompression operation. A torsion model and bending model are built based on traditional elastic mechanics, and then a coupling model is simplified and modified for improved robot control. The source and influence of residual error caused by the noise is analysed, and data-based estimation of the model parameters is conducted to accommodate different patients. Robot deformation is also considered in the control system, and the model inputs—the feedback force data—are processed using the adaptive Kalman filter in real time. The lumbar data are segmented for finite element analysis, which provides the input and output data with which the parameters of the model are estimated, and the simulation shows that the signal-to-noise ratio of the force sensor has a great effect on system performance.

Design of Video Image Low-Pass Filtering Process Based on FPGA

A real-time video low-pass filter processing system is designed and implemented in this paper through VHDL language programming, applying Altera CycloneIII series FPGA as the main control chip. The changes in display quality of video signals of different bandwidths can be observed on the monitor by changing the bandwidths of the input video signals through low-pass filtering.

An FPGA-Oriented Algorithm for Real-Time Filtering of Poisson Noise in Video Streams, with Application to X-Ray Fluoroscopy

In this paper we propose a new algorithm for real-time filtering of video sequences corrupted by Poisson noise. The algorithm provides effective denoising (in some cases overcoming the filtering performances of state-of-the-art techniques), is ideally suited for hardware implementation, and can be implemented on a small field-programmable gate array using limited hardware resources. The paper describes the proposed algorithm, using X-ray fluoroscopy as a case study. We use IIR filters for time filtering, which largely simplifies hardware cost with respect to previous FIR filter-based implementations. A conditional reset is implemented in the IIR filter, to minimize motion blur, with the help of an adaptive thresholding approach. Spatial filtering performs a conditional mean to further reduce noise and to remove isolated noisy pixels. IIR filter hardware implementation is optimized by using a novel technique, based on Steiglitz–McBride iterative method, to calculate fixed-point filter coefficients with minimal number of nonzero elements. Implementation results using the smallest StratixIV FPGA show that the system uses only, at most, the 22% of the resources of the device, while performing real-time filtering of 1024 × 1024@49fps video stream. For comparison, a previous FIR filter-based implementation, on the same FPGA, in the same conditions and constraints (1024 × 1024@49fps), requires the 80% of the logic resources of the FPGA.

A Hybrid Approach Based on Improved AR Model and MAA for INS/DVL Integrated Navigation Systems

In navigation of autonomous underwater vehicles (AUVs), the estimation of position is an important issue, especially, when the sensors such as gyroscopes contain a lot of noise, and the velocity information of Doppler velocity log (DVL) is affected by the motion attitude of the vehicle. In this paper, based on an improved auto regressive (AR) model, a real-time filter is utilized for gyroscope signal de-noising. Meanwhile, according to the characteristics of the AUV, the influence of the vehicle attitude on the DVL velocity measurement error is analyzed and a motion attitude assist (MAA) method based on error model is introduced for enhancing DVL velocity accuracy. In this paper, using the proposed hybrid approach, an inertial navigation system (INS)/DVL integrated navigation system is designed. The proposed approach is evaluated by simulation and experimental test in different acceleration bound, and the existence of the DVL outage for an AUV. The results indicate that the precisions of the velocity and position are improved effectively, especially in complex motion attitude and long sailing conditions.

改进的MEMS陀螺随机噪声自适应Kalman实时滤波方法

改进的MEMS陀螺随机噪声自适应Kalman实时滤波方法

Research on navigation for underwater unmanned vehicle based on dynamic gray filtering method

According to the problem of large platform noise and external disturbance influence and the data may have outliers for UUV navigation, a dynamic grey filtering method based on fractional order operator is proposed. This method does not require prior knowledge and noise statistical properties of UUV system while establishing a dynamic grey model based on fractional operator according to navigation data. The deviation between the grey prediction value and the next navigation sample value is calculated and the data is fused according to the deviation to realize the real-time filtering. The simulation results show that the method is effective and feasible.

A Study of the Relationship Between the Geographic Locations of the User and Participation in Twitter During Different Types of News Events

Twitter is one of the most active social networks in news sharing. People report local events sometimes faster than news agencies. During major events, such as earthquakes and presidential elections, people share tweets, retweets, images and links related to the event, creating an overwhelming number of posts. The amount of data generated by social media provides a shared resource for the discovery of information for individuals, organizations and governments, but the large stream of tweets makes tracking interesting posts a challenging task. Understanding user behavior during different contexts provides substantial insight to get the most out of the social media. The objective of this work is to investigate the relationship between user participation, news type and geographic locations of Twitter’s users. The results show that financial news tweets have distinct user behavior compared to tweets about political events and tweets about disasters. Tweets about financial events tend to have more original tweets and more links, than tweets about political events, which had larger numbers of retweets. The investigation of the relationship between the style of user’s participation and the type of news provides insight for social media management, specifically, for using information diffusion proactively, real time filtering, and location-aware news recommendation systems. We found that there are relationships between the country of the user, the type of news and user tweeting behavior for news related tweets. News type is strongly correlated to users’ behavior and was found to have a stronger relationship to users’ behavior than geographic location.

Real-Time Filtering with Sparse Variations for Head Motion in Magnetic Resonance Imaging.

Estimating a time-varying signal, such as head motion from magnetic resonance imaging data, becomes particularly challenging in the face of other temporal dynamics such as functional activation. This paper describes a new Kalman filter-like framework that includes a sparse residual term in the measurement model. This additional term allows the extended Kalman filter to generate real-time motion estimates suitable for prospective motion correction when such dynamics occur. An iterative augmented Lagrangian algorithm similar to the alterating direction method of multipliers implements the update step for this Kalman filter. This paper evaluates the accuracy and convergence rate of this iterative method for small and large motion in terms of its sensitivity to parameter selection. The included experiment on a simulated functional magnetic resonance imaging acquisition demonstrates that the resulting method improves the maximum Youden's J index of the time series analysis by 2-3% versus retrospective motion correction, while the sensitivity index increases from 4.3 to 5.4 when combining prospective and retrospective correction.

A real-time filtering method for the drift data of a laser Doppler velocimeter

ABSTRACTTo effectively reduce the random drift of a laser Doppler velocimeter (LDV), a real-time filtering model is presented for filtering the drift data of an LDV, which is a combination of the metabolic grey model (1, 1) and the metabolic time series model AR (2). The basic principle of the metabolic grey-time series model is introduced in detail first. Then, the model is established for the static and dynamic drift data, and a Kalman filter is used to filter the drift data based on the model. The variance analysis method and the Allan variance method are employed to analyse the static drift data. The dynamic drift data are also compared before and after being modelled and filtered. The results demonstrate that the metabolic grey-time series method cannot only effectively reduce the static random drift of an LDV, but can also reduce the dynamic random drift in real time.

Model and Algorithm Improvement on Single Beacon Underwater Tracking

In range-based underwater tracking, the accuracy of the calculated slant range depends largely on the accuracy of the effective sound velocity (ESV). The ESV between a pinger and a receiver certainly is location dependent, and is very difficult to determine accurately. However, a common assumption adopted by many published papers related to single beacon underwater tracking is that the ESV is not only a known quantity, but also location independent. This paper proposes a novel Kalman-based single beacon tracking model that treats the ESV as a state variable. For applying this novel model, a solution method by treating transit time directly as measurement is developed. Traditional Kalman filters for single beacon underwater tracking often update state variables at a uniform rate. However, because the time of arrival measurement arrives at irregular time intervals, implementing a Kalman-based filter for updating state variables should carry out at nonuniform sampling rates. In addition to the realtime filter tracking, this paper also implements the Rauch–Tung–Striebel (RTS) smoother for postprocessing (reconstruction). This paper proposes a hybrid timing algorithm that can achieve both accuracy in filtering and efficiency in smoothing. Through numerical examples of using simulated and field data, both the extended Kalman filter and RTS smoother results show that while implementing the proposed model and algorithm, the tracking accuracy can be significantly improved and the estimated ESV agrees well with its true value.

WAVELET METHODS FOR CARDIO SIGNALS PROCESSING

The article discusses the use of wavelet transforms and algorithms in the digital processing of cardiosignals in a number of medical applications due to their good adaptability to the analysis of non-stationary signals (that is, those whose statistical characteristics change with time). Since an electrocardiogram is a transient signal, wavelet methods can be used to recognize and detect key diagnostic features. In real-time systems for digital signal processing, it is important that mathematical operations are performed quickly, and the time required to execute commands must be known precisely and in advance. For this, both the program and the hardware must be very effective. In digital signal processors, the most important mathematical operation and the core of all digital signal processing algorithms is multiplication, followed by summation. Fast execution of the multiplication operation followed by summation is very important for implementing real-time digital filters, signal processing, matrix multiplication, and graphic image manipulation. Therefore, all this requires the need to improve methods, algorithms and signal processing programs that determine the quality and performance of digital systems. The technique is based on the transformation of the heart rate to simple harmonic oscillations (fast Fourier transform, autoregressive analysis) with different frequencies. In this case, the sequence of heartbeats is converted into a power spectrum of fluctuations of the duration of the RR intervals, which are a sequence of frequencies characterizing HRV. Most often, the area bounded by the spectral power curve corresponding to a certain defined frequency range, that is, the power within a limited frequency range, is estimated.