In-field Calibration Research Articles

Unveiling the potential of a novel portable air quality platform for assessment of fine and coarse particulate matter: in-field testing, calibration, and machine learning insights.

Although low-cost air quality sensors facilitate the implementation of denser air quality monitoring networks, enabling a more realistic assessment of individual exposure to airborne pollutants, their sensitivity to multifaceted field conditions is often overlooked in laboratory testing. This gap was addressed by introducing an in-field calibration and validation of three PAQMON 1.0 mobile sensing low-cost platforms developed at the Mining and Metallurgy Institute in Bor, Republic of Serbia. A configuration tailored for monitoring PM2.5 and PM10 mass concentrations along with meteorological parameters was employed for outdoor measurement campaigns in Bor, spanning heating (HS) and non-heating (NHS) seasons. A statistically significant positive linear correlation between raw PM2.5 and PM10 measurements during both campaigns (R > 0.90, p ≤ 0.001) was observed. Measurements obtained from the uncalibrated NOVA SDS011 sensors integrated into the PAQMON 1.0 platforms exhibited a substantial and statistically significant correlation with the GRIMM EDM180 monitor (R > 0.60, p ≤ 0.001). The calibration models based on linear and Random Forest (RF) regression were compared. RF models provided more accurate descriptions of air quality, with average adjR2 values for air quality variables in the range of 0.70 to 0.80 and average NRMSE values between 0.35 and 0.77. RF-calibrated PAQMON 1.0 platforms displayed divergent levels of accuracy across different pollutant concentration ranges, achieving a data quality objective of 50% during both measurement campaigns. For PM2.5, uncertainty ( ) was below 50% for concentrations between 9.06 and 34.99μg/m3 in HS and 5.75 and 17.58μg/m3 in NHS, while for PM10, it stayed below 50% from 19.11 to 51.13μg/m3 in HS and 11.72 to 38.86μg/m3 in NHS.

Measurement System for the Calibration of Accelerometer Arrays

This paper addresses accelerometer array calibration, focusing on determining the errors between multiple sensors. Micro-electromechanical system (MEMS) based triaxial accelerometers, key components of Inertial Measurement Units (IMUs), are used in localization, robotics, and navigation systems. The requirements of these applications necessitate low-cost sensors, which makes MEMS IMUs a reasonable choice. However, these low-cost IMUs are significantly affected by systematic (i.e., bias, misalignment, scale-factor) and random errors. Achieving reliable sensor output depends on the precision of the executed calibration method. While traditional laboratory-based sensor calibration using specialized equipment (i.e., three-axis turntable) is accurate, it is time-consuming and costly. In contrast, in-field calibration techniques, which can be performed using a mechatronic actuator or a robotic arm, have gained popularity. These techniques involve comparing sensor measurements to established reference values. The MEMS sensors are increasingly being used in multi-sensor applications, which demands not only individual sensor error calibration but also important to determine the axis misalignment between the used sensors. During calibration process, various optimization algorithms (e.g., GA, PSO) can also be used to find the error parameters. The proposed measurement system allows for individual calibration of misalignment, bias, and scale factor of the sensor array, and eliminates between-sensor misalignment errors.

In-field calibration of triaxial accelerometer based on PE-ANGO

Abstract This paper presents a new method for in-field calibration of accelerometers to address the problems of low efficiency and high cost associated with traditional calibration methods. A nonlinear mathematical model of the accelerometer is established, and the cost function is analysed and deduced the cost function. Then, an adaptive Northern Goshawk Optimisation (NGO) algorithm based on prior knowledge enhancement is introduced. A method of collecting multi-position data with a hand-held accelerometer is introduced, and the proposed algorithm is used to in-field calibrate nine parameters of the accelerometer’s nonlinear error model. In addition, simulation is used to compare the results of calibrating the accelerometer using the proposed algorithm and the original algorithm, demonstrating the superiority of the proposed algorithm. Finally, experimental results confirm that the proposed method can rapidly calibrate accelerometer error parameters without relying on complex equipment and with greater accuracy than traditional methods.

Improved modeling and fast in-field calibration of optical flow sensor for unmanned aerial vehicle position estimation

To better reveal the principle of off-the-shelf OFS that used for UAV position estimation, two measurement models are derived, namely the continuous-time and discrete-time models. Then the intrinsic equivalence of these two models is also proven. Based on these two models, a calibration method for OFS is introduced, which only makes use of in-situ rotation to determine the OFS’s parameters including its focal length and installation angles with respect to the UAV. Furthermore, an OFS-based UAV position estimation algorithm is presented, which stems from the proposed discrete-time model. The proposed OFS models, calibration method, and position estimation algorithm form a full solution for UAV displacement measurement via OFS, and its feasibility and effectiveness is verified through flight tests on a quadcopter UAV in outdoor environment.

Integrating Remote Sensing and Weather Variables for Mango Yield Prediction Using a Machine Learning Approach

Accurate pre-harvest yield forecasting of mango is essential to the industry as it supports better decision making around harvesting logistics and forward selling, thus optimizing productivity and reducing food waste. Current methods for yield forecasting such as manually counting 2–3% of the orchard can be accurate but are very time inefficient and labour intensive. More recent evaluations of technological solutions such as remote (satellite) and proximal (on ground) sensing have provided very encouraging results, but they still require infield in-season sampling for calibration, the technology comes at a significant cost, and commercial availability is limited, especially for vehicle-mounted sensors. This study presents the first evaluation of a ”time series”—based remote sensing method for yield forecasting of mango, a method that does not require infield fruit counts and utilizes freely available satellite imagery. Historic yield data from 2015 to 2022 were sourced from 51 individual orchard blocks from two farms (AH and MK) in the Northern Territory of Australia. Time series measures of the canopy reflectance properties of the blocks were obtained from Landsat 7 and 8 satellite data for the 2015–2022 growing seasons. From the imagery, the following vegetation indices (VIs) were derived: EVI, GNDVI, NDVI, and LSWI, whilst corresponding weather variables (rainfall (Prec), temperature (Tmin/Tmax), evapotranspiration (ETo), solar radiation (Rad), and vapor pressure deficit (vpd)) were also sourced from SILO data. To determine the relationships among weather and remotely sensed measures of canopy throughout the growing season and the yield achieved (at the block level and the farm level), six machine learning (ML) algorithms, namely random forest (RF), support vector regression (SVR), eXtreme gradient boosting (XGBOOST), RIDGE, LASSO and partial least square regression (PLSR), were trialed. The EVI/GNDVI and Prec/Tmin were found to be the best RS and weather predictors, respectively. The block-level combined RS/weather-based RF model for 2021 produced the best result (MAE = 2.9 t/ha), marginally better than the RS only RF model (MAE = 3.4 t/ha). The farm-level model error (FLEM) was generally lower than the block-level model error, for both the combined RS/weather-based RF model (farm = 3.7%, block (NMAE) = 33.6% for 2021) and the RS-based model (farm = 4.3%, block = 38.4% for 2021). Further testing of the RS/weather-based RF models over six additional orchards (other than AH and MK) produced errors ranging between 24% and 39% from 2016 to 2020. Although accuracies of prediction did vary at both the block level and the farm level, this preliminary study demonstrates the potential of a ”time series” RS method for predicting mango yields. The benefits to the mango industry are that it utilizes freely available imagery, requires no infield calibration, and provides predictions several months before the commercial harvest. Therefore, this outcome not only presents a more adoptable option for the industry, but also better supports automation and scalability in terms of block-, farm-, regional, and national level forecasting.

A Variational Bayesian Blind Calibration Approach for Air Quality Sensor Deployments

Air pollution has become a global threat to urban environments and public health. Low-cost air quality sensor systems have been deployed to support fine-grained monitoring, and in-field calibration methods are necessary to assure the accuracy of sensor observation. Existing methods use in-field reference to train some specific calibration models. Nevertheless, collecting sufficient reference data after deployment is challenging. First, the synchronized data pair of reference observation is hard to obtain. Thus, many blind calibration approaches rely on some alternative reference like a historical reference before sensor observation. However, since the relationship model between alternative reference and sensor observation becomes more complicated, the amount of collected data is still insufficient to learn the complex model. To address the above challenge, we propose a Variational Bayesian Blind Calibration Algorithm. Our method introduces a supplement set that only involves historical reference and synchronized reference measurement without sensor observation to alleviate the few-reference pressure. Large amounts of the introduced supplement sets have been collected in different cities by accurate stations, which we adopt to conduct prediction tasks and build a Bayesian model to learn how to combine the formulated prediction task and target calibration task in a theoretically better way. Furthermore, we design a variational Bayesian framework to make good use of the easily obtained supplement set to train a better prediction model for improving calibration performance and relieving the pressure of collecting costly reference measurements after deployment. Evaluations of real-world and synthetic datasets show that the proposed approach has a better performance than previous baselines.

Validation of In-field Calibration for Low-Cost Sensors Measuring Ambient Particulate Matter in Kolkata, India

Validation of In-field Calibration for Low-Cost Sensors Measuring Ambient Particulate Matter in Kolkata, India

Clustering-Based Segmented Regression for Particulate Matter Sensor Calibration

Nowadays, sensor-based air pollution sensing systems are widely deployed for fine-grained pollution monitoring. In-field calibration plays an important role in maintaining sensory data quality. Determining the model structure is challenging using existing methods of variable global fitting models for in-field calibration. This is because the mechanism of interference factors is complex and there is often insufficient prior knowledge on a specific sensor type. Although Artificial-Neuron-Net-based (ANN-based) methods ignore the complex conditions above, they also have problems regarding generalization, interpretability, and calculation cost. In this paper, we propose a clustering-based segmented regression method for particulate matter (PM) sensor in-field calibration. Interference from relative humidity and temperature are taken into consideration in the particulate matter concentration calibration model. Samples for modeling are divided into clusters and each cluster has an individual multiple linear regression equation. The final calibrated result of one sample is calculated from the regression model of the cluster the sample belongs to. The proposed method is evaluated under in-field deployment and performs better than a global multiple regression method both on PM2.5 and PM10 pollutants with, respectively, at least 16% and 9% improvement ratio on RMSE error. In addition, the proposed method is insensitive to reduction of training data and increase in cluster number. Moreover, it may bear lighter calculation cost, less overfitting problems and better interpretability. It can improve the efficiency and performance of post-deployment sensor calibration.

Installation error calibration for MEMS angular measurement system in hypersonic wind tunnel

The accuracy of model attitude measurement has an important impact on wind tunnel test results. Microelectromechanical System Inertial Measurement Unit (MEMS IMU) provides a feasible way to measure model attitudes with high accuracy. However, the installation error between MEMS IMU coordinate system and the body coordinate system of test models can make the accuracy of the model attitude measurement decrease. In wind tunnel tests, the installation error depends on the relationship between the IMU and the model mechanism before tests. Therefore, in-field calibration in wind tunnel tests is necessary to reduce installation errors. To improve attitude measurement accuracy, the least squares quaternion calibration method based on MEMS IMU and six-position calibration procedure are proposed. High-precision three-axis turntable tests are performed. The pitch accuracy after calibration is higher than that before calibration in the angle of attack sweeping tests. The Root-Mean-Square Errors (RMSE) in the roll and yaw are within 0.01°, which are smaller than those before calibration. In the roll sweeping tests, RMSE of three attitude angles decrease significantly. In hypersonic wind tunnel tests, the pitch errors before and after calibration are within 0.05° and 0.02° in the angle of attack sweeping tests without wind. In five angle of attack sweeping tests with wind, the deviation between the mean of the pitch and the pitch after the elastic angle correction is within 0.03° and the standard deviation of five tests is within 0.01°. The proposed method is confirmed to enhance the accuracy of attitude measurement effectively, which is convenient for engineering applications.

MAIC: Metalearning-Based Adaptive In-Field Calibration for IoT Air Quality Monitoring System

Air pollution has become a global threat to human health. Fine-grained air quality monitoring has attracted much attention in recent years. Low-cost calibrated sensors make it possible for the large-scale deployment of IoT air quality monitoring systems. In practice, the calibration performance degrades after deployment due to the dynamic and diversity of system conditions. However, it is infeasible to collect sufficient in-field reference data to train calibration models for these new conditions. To address the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">multicondition</i> and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">few-data</i> challenge, we proposed metalearning-based adaptive in-field calibration (MAIC), a metalearning-based adaptive in-field calibration algorithm. Specifically, MAIC adopts metalearning to learn how to adapt to new conditions quickly. To effectively leverage historical data, we first develop task generation strategies for sensor calibration under this scheme. Then, task-oriented optimization is introduced to train a model with superior adaptability in the offline training phase. Furthermore, an adaptation method is presented to learn the task-specific data distribution without forgetting the metaknowledge, enabling continual learning to utilize the temporal dependencies between multiple conditions. Our evaluations on synthetic and real-world data sets show that MAIC has high robustness and adaptability under multiple complicated conditions. Our proposed method outperforms the state-of-the-art calibration algorithms by 4.23%–29.46% in the real-world deployment data set, but with fewer requirements for the available in-field reference data.

Correlation-based transceiver in-phase/ quadrature skew in-field calibration in dual-polarization coherent optical transmission system.

In this paper, we propose a novel transceiver in-phase/quadrature (I/Q) skew in-field calibration scheme with correlation-based method for the dual-polarization coherent optical system. Simultaneous dual-polarization calibration of transceiver I/Q skews after fiber transmission is experimentally performed. Rx/Tx correlation-based skew estimations (CBSEs) are proposed to accurately estimate the transceiver I/Q skews with dual-polarization OFDM signal. By simulation, the robustness of the Rx/Tx CBSEs is investigated against various transceiver I/Q imbalances and channel impairments including carrier frequency offset (CFO), phase noise (PN), and chromatic dispersion (CD). The simultaneous measurement of large transceiver skews is studied within a range of ±128 ps. The bit error rate (BER) improvement brought by the CBSEs is studied in 80 km single-mode fiber (SMF) transmissions under various Rx/Tx skews. In the experiments, the Rx/Tx skew is measured in the range of 1 to 128 ps w/ and w/o the presence of 5 ps Tx/Rx skew. Simultaneous dual-polarization measurements are performed with the X/Y polarization Tx/Rx skews set to 2.5 ps, 5 ps, 7.5 ps and 10 ps, respectively. The measurement errors are within ±0.2 ps. The 80 km SMF dual-polarization transmission after in-field calibration for inter-data center interconnection (inter-DCI) is implemented, with a data rate of 400 Gb/s for both 16QAM and 32QAM modulation formats.

Cognitive Raman Amplifier Control Using an Evolutionary Optimization Strategy

In this work, a cognitive Raman amplifier controller using an evolutionary optimization strategy for both in-field device calibration and optimal pump power configuration design is presented. The conceived methodology allows to jointly characterize the physical layer properties of the optical fiber span, such as the fiber attenuation profile, the entities of the lumped losses due to mechanical connectors or splices, the polarization coefficients between each Raman pump pair and the Raman efficiency curve. The developed framework expects to work using optical channel monitors (OCMs) as telemetry device and it has been experimentally validated, obtaining outstanding results in terms of precision on a C-band setup.

Sub-arcsecond imaging with the International LOFAR Telescope

The International LOFAR Telescope is an interferometer with stations spread across Europe. With baselines of up to ~2000 km, LOFAR has the unique capability of achieving sub-arcsecond resolution at frequencies below 200 MHz. However, it is technically and logistically challenging to process LOFAR data at this resolution. To date only a handful of publications have exploited this capability. Here we present a calibration strategy that builds on previous high-resolution work with LOFAR. It is implemented in a pipeline using mostly dedicated LOFAR software tools and the same processing framework as the LOFAR Two-metre Sky Survey (LoTSS). We give an overview of the calibration strategy and discuss the special challenges inherent to enacting high-resolution imaging with LOFAR, and describe the pipeline, which is publicly available, in detail. We demonstrate the calibration strategy by using the pipeline on P205+55, a typical LoTSS pointing with an 8 h observation and 13 international stations. We perform in-field delay calibration, solution referencing to other calibrators in the field, self-calibration of these calibrators, and imaging of example directions of interest in the field. We find that for this specific field and these ionospheric conditions, dispersive delay solutions can be transferred between calibrators up to ~1.5° away, while phase solution transferral works well over ~1°. We also demonstrate a check of the astrometry and flux density scale with the in-field delay calibrator source. Imaging in 17 directions, we find the restoring beam is typically ~0.3′′ ×0.2′′ although this varies slightly over the entire 5 deg2field of view. We find we can achieve ~80–300 μJy bm−1image rms noise, which is dependent on the distance from the phase centre; typical values are ~90 μJy bm−1for the 8 h observation with 48 MHz of bandwidth. Seventy percent of processed sources are detected, and from this we estimate that we should be able to image roughly 900 sources per LoTSS pointing. This equates to ~ 3 million sources in the northern sky, which LoTSS will entirely cover in the next several years. Future optimisation of the calibration strategy for efficient post-processing of LoTSS at high resolution makes this estimate a lower limit.

Statistical Analysis of In-Field Magnetometer Calibration for Two Representative Methods

In this paper, we propose a procedure for the investigation of the feasibility and effectiveness of recently developed in-field magnetometer calibration methods. Extensive numerical analyses under different noise levels are conducted to identify the proper measurement conditions of each method. The variance analysis is also presented to study the observability of model parameters based on the information matrix under different inclination angles. In accordance with the analysis, several important conclusions are drawn including the condition to ensure the validity of the parameter estimation and the approach to improve the performance of the calibration results. Numerical simulation and experimental validation support our analytical findings. Accordingly, this study provides an in-depth exploration into the limitations of typical in-field magnetometer calibration methods.

Accurate and Computational-Efficient Analytical Solutions for the Extended Two-Step Magnetometer Calibration

One of the biggest challenges involving low-cost inertial and magnetic sensors, especially the latter, is the inherent distortion/corruption of their measurements by systematic errors. Such errors can be so compromising that using the corrupted measurements for navigation purposes becomes an impossible task. In addition to sensor fusion, which aims at improving performance, calibration techniques can precisely estimate these errors, which, then, can be compensated for. This paper revisits the problem of in-field magnetometer calibration, focusing on the traditional and well-established ellipsoid fitting-based Extended Two-Step (ETS) method. Despite being largely employed, especially for initializing more sophisticated and iterative optimization-based calibration techniques, ETS implementation is not straightforward, as the mapping between its estimated intermediate parameters and the sensor error parameters of interest (biases, scale factors, and misalignments) has not been clearly provided in the literature yet. In this paper, hence, we carefully derive the latter, aiming at facilitating ETS numerical implementation. Withal, and also figuring as main contribution of this paper, we provide analytical closed-form solutions for ETS, which are especially suitable for low-cost, real-time embedded applications. In order to validate their accuracy, insensitivity to local-minima convergence issues, and computational efficiency, we present simulations - including a Monte Carlo analysis - and hardware implementation, alongside a comprehensive comparison with state-of-the-art magnetometer calibration techniques.

Local Reference-Free In-Field Calibration of Low-Cost Air Pollution Monitoring Sensors

Local Reference-Free In-Field Calibration of Low-Cost Air Pollution Monitoring Sensors

Domain Adaptation-Based Deep Calibration of Low-Cost PM₂.₅ Sensors

Air pollution is a severe problem growing over time. A dense air-quality monitoring network is needed to update the people regarding the air pollution status in cities. A low-cost sensor device (LCSD) based dense air-quality monitoring network is more viable than continuous ambient air quality monitoring stations (CAAQMS). An in-field calibration approach is needed to improve agreements of the LCSDs to CAAQMS. The present work aims to propose a calibration method for PM <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2.5</sub> using domain adaptation technique to reduce the collocation duration of LCSDs and CAAQMS. A novel calibration approach is proposed in this work for the measured PM <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2.5</sub> levels of LCSDs. The dataset used for the experimentation consists of PM <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2.5</sub> values and other parameters (PM <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">10</sub> , temperature, and humidity) at hourly duration over a period of three months data. We propose new features, by combining PM <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2.5</sub> , PM <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">10</sub> , temperature, and humidity, that significantly improved the performance of calibration. Further, the calibration model is adapted to the target location for a new LCSD with a collocation time of two days. The proposed model shows high correlation coefficient values (R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> ) and significantly low mean absolute percentage error (MAPE) than that of other baseline models. Thus, the proposed model helps in reducing the collocation time while maintaining high calibration performance.

Advanced NO&amp;lt;sub&amp;gt;2&amp;lt;/sub&amp;gt; retrieval technique for the Brewer spectrophotometer applied to the 20-year record in Rome, Italy

Abstract. A re-evaluated data set of nitrogen dioxide (NO2) column densities over Rome for the years 1996 to 2017 is here presented. This long-term record is obtained from ground-based direct sun measurements with a MkIV Brewer spectrophotometer (serial number #067) and further reprocessed using a novel algorithm. Compared to the original Brewer algorithm, the new method includes updated NO2 absorption cross sections and Rayleigh scattering coefficients, and it accounts for additional atmospheric compounds and instrumental artefacts, such as the spectral transmittance of the filters, the alignment of the wavelength scale, and internal temperature. Moreover, long-term changes in the Brewer radiometric sensitivity are tracked using statistical methods for in-field calibration. The resulting series presents only a few (about 30) periods with missing data longer than 1 week and features NO2 retrievals for more than 6100 d, covering nearly 80 % of the considered 20-year period. The high quality of the data is demonstrated by two independent comparisons. In the first intensive campaign, Brewer #067 is compared against another Brewer (#066), recently calibrated at the Izaña Atmospheric Observatory through the Langley method and there compared to reference instrumentation from the Network for the Detection of Atmospheric Composition Change (NDACC). Data from this campaign show a highly significant Pearson's correlation coefficient of 0.90 between the two series of slant column densities (SCDs), slope 0.98 and offset 0.05 DU (Dobson units; 1.3×1015 molec.cm-2). The average bias between the vertical column densities is 0.03 DU (8.1×1014 molec.cm-2), well within the combined uncertainty of both instruments. Brewer #067 is also independently compared with new-generation instrumentation, a co-located Pandora spectrometer (#117), over a 1-year-long period (2016–2017) at Sapienza University of Rome, showing linear correlation indices above 0.96 between slant column densities, slope of 0.97, and offset of 0.02 DU (5.4×1014 molec.cm-2). The average bias between vertical column densities is negligible (−0.002 DU or -5.4×1013 molec.cm-2). This, incidentally, represents the first intercomparison of NO2 retrievals between a MkIV Brewer and a Pandora instrument. Owing to its accuracy and length, the Brewer data set collected in Rome can be useful for satellite calibration/validation exercises, comparison with photochemical models, and better aerosol optical depth estimates (NO2 optical depth climatology). In addition, it can be employed to identify long-term trends in NO2 column densities in a metropolitan environment, over two decades witnessing important changes in environmental policies, emission loads and composition, and the effect of a worldwide economic recession, to offer just a few examples. The method can be replicated on the more than 80 MkIV spectrophotometers operating worldwide in the frame of the international Brewer network. The NO2 data set described in this paper can be freely accessed at https://doi.org/10.5281/zenodo.4715219 (Diémoz and Siani, 2021).

Missing Data Imputation on IoT Sensor Networks: Implications for on-Site Sensor Calibration

IoT sensors are becoming increasingly important supplement to traditional monitoring systems, particularly for in-situ based monitoring. Data collected using IoT sensors are often plagued with missing values occurring as a result of sensor faults, network failures, drifts and other operational issues. Missing data can have substantial impact on in-field sensor calibration methods. The goal of this research is to achieve effective calibration of sensors in the context of such missing data. To this end, two objectives are presented in this paper. 1) Identify and examine effective imputation strategy for missing data in IoT sensors. 2) Determine sensor calibration performance using calibration techniques on data set with imputed values. Specifically, this paper examines the performance of Variational Autoencoder (VAE), Neural Network with Random Weights (NNRW), Multiple Imputation by Chain Equations (MICE), Random Forest-based Imputation (missForest) and K-Nearest Neighbour (KNN) for imputation of missing values on IoT sensors. Furthermore, the performance of sensor calibration via different supervised algorithms trained on the imputed dataset were evaluated. The analysis showed VAE technique to outperform the other methods in imputing the missing values at different proportions of missingness on two real-world datasets. Experimental results also showed improved calibration performance with imputed dataset.

Automatic Targetless Extrinsic Calibration of a 3D Lidar and Camera by Maximizing Mutual Information

This paper reports on a mutual information (MI) based algorithm for automatic extrinsic calibration of a 3D laser scanner and optical camera system. By using MI as the registration criterion, our method is able to work in situ without the need for any specific calibration targets, which makes it practical for in-field calibration. The calibration parameters are estimated by maximizing the mutual information obtained between the sensor-measured surface intensities. We calculate the Cramer-Rao-Lower-Bound (CRLB) and show that the sample variance of the estimated parameters empirically approaches the CRLB for a sufficient number of views. Furthermore, we compare the calibration results to independent ground-truth and observe that the mean error also empirically approaches to zero as the number of views are increased. This indicates that the proposed algorithm, in the limiting case, calculates a minimum variance unbiased (MVUB) estimate of the calibration parameters. Experimental results are presented for data collected by a vehicle mounted with a 3D laser scanner and an omnidirectional camera system.