This paper presents the accurate, highly efficient, and learning-free method CFEAR Radarodometry for large-scale radar odometry estimation. By using a filtering technique that keeps the k strongest returns per azimuth and by additionally filtering the radar data in Cartesian space, we are able to compute a sparse set of oriented surface points for efficient and accurate scan matching. Registration is carried out by minimizing a point-to-line metric and robustness to outliers is achieved using a Huber loss. We were able to additionally reduce drift by jointly registering the latest scan to a history of keyframes and found that our odometry method generalizes to different sensor models and datasets without changing a single parameter. We evaluate our method in three widely different environments and demonstrate an improvement over spatially cross-validated state-of-the-art with an overall translation error of 1.76% in a public urban radar odometry benchmark, running at 55Hz merely on a single laptop CPU thread.
This article presents an accurate, highly efficient, and learning-free method for large-scale odometry estimation using spinning radar, empirically found to generalize well across very diverse environments—outdoors, from urban to woodland, and indoors in warehouses and mines—without changing parameters. Our method integrates motion compensation within a sweep with one-to-many scan registration that minimizes distances between nearby oriented surface points and mitigates outliers with a robust loss function. Extending our previous approach conservative filtering for efficient and accurate radar odometry (CFEAR), we present an in-depth investigation on a wider range of datasets, quantifying the importance of filtering, resolution, registration cost and loss functions, keyframe history, and motion compensation. We present a new solving strategy and configuration that overcomes previous issues with sparsity and bias, and improves our state-of-the-art by 38%, thus, surprisingly, outperforming radar simultaneous localization and mapping (SLAM) and approaching lidar SLAM. The most accurate configuration achieves 1.09% error at 5 Hz on the Oxford benchmark, and the fastest achieves 1.79% error at 160 Hz.
New fast sealed lead acid (SLA) battery chargers must be able to charge the fully discharged batteries in a short time. In the same time, the charger must monitor the battery state of health in order to prevent over charge and to extend the battery life time.
In this paper a Fast charger was presented to charge SLA batteries in short time and monitor the battery voltage to prevent over charge. The design was implemented practically. And 150 charger of similar type was produced for commercial use. They are now in service in different Mobile base station sites around Baghdad. It can charge a fully discharged 12V, 4.5Ah battery in less than 5 hours. To supply PLC control system on DC power to about 24 hour of continuous operation during main electricity faults.
During one and half year of continuous operation three faults have been recorded in the 150 chargers. All of the three cases were because of bad components manufacturing.
In the Electronic Warfare receivers, it is important to know the type of modulation of the intercepted Radar signals (MOP modulation on pulse). This information can be very helpful in identifying the type of Radar present and to take the appropriate actions against it. In this paper, a new signal processing method is presented to identify the FM (Frequency Modulation) pattern from the received Radar pulses. The proposed processing method based on Auto Regressive Spectrum Modelling used for digital modulation classification [1]. This model uses the instantaneous frequency and instantaneous bandwidth as obtained from the roots of the autoregressive polynomial. The instantaneous frequency and instantaneous bandwidth together were used to identify the type of modulation in the Radar pulse. Another feature derived from the instantaneous frequency is the frequency rate of change. The frequency rate of change was used to extract the pattern of the frequency change. Results show that this method works properly even for low signal to noise ratios.
This paper investigates the problem of calibrating sensors affected by (i) heteroskedastic measurement noise and (ii) a polynomial bias, describing a systematic distortion of the measured quantity. First, a set of increasingly complex statistical models for the measurement process was proposed. Then, for each model the authors design a Bayesian parameters estimation method handling heteroskedasticity and capable to exploit prior information about the model parameters. The Bayesian problem is solved using MCMC methods and reconstructing the unknown parameters posterior in sampled form. The authors then test the proposed techniques on a practically relevant case study, the calibration of Light Detection and Ranging (Lidar) sensor, and evaluate the different proposed procedures using both artificial and field data.
Accurate and robust mobile robot localization is very important in many robot applications. Monte Carlo localization (MCL) is one of the robust probabilistic solutions to robot localization problems. The sensor model used in MCL directly influence the accuracy and robustness of the pose estimation process. The classical beam models assumes independent noise in each individual measurement beam at the same scan. In practice, the noise in adjacent beams maybe largely correlated. This will result in peaks in the likelihood measurement function. These peaks leads to incorrect particles distribution in the MCL. In this research, an adaptive sub-sampling of the measurements is proposed to reduce the peaks in the likelihood function. The sampling is based on the complete scan analysis. The specified measurement is accepted or not based on the relative distance to other points in the 2D point cloud. The proposed technique has been implemented in ROS and stage simulator. The result shows that selecting suitable value of distance between accepted scans can improve the localization error and reduce the required computations effectively.
We consider the problem of calibrating range measurements of a Light Detection and Ranging (lidar) sensor that is dealing with the sensor nonlinearity and heteroskedastic, range-dependent, measurement error. We solved the calibration problem without using additional hardware, but rather exploiting assumptions on the environment surrounding the sensor during the calibration procedure. More specifically we consider the assumption of calibrating the sensor by placing it in an environment so that its measurements lie in a 2D plane that is parallel to the ground. Then, its measurements come from fixed objects that develop orthogonally w.r.t. the ground, so that they may be considered as fixed points in an inertial reference frame. Moreover, we consider the intuition that moving the distance sensor within this environment implies that its measurements should be such that the relative distances and angles among the fixed points above remain the same. We thus exploit this intuition to cast the sensor calibration problem as making its measurements comply with this assumption that "fixed features shall have fixed relative distances and angles". The resulting calibration procedure does thus not need to use additional (typically expensive) equipment, nor deploy special hardware. As for the proposed estimation strategies, from a mathematical perspective we consider models that lead to analytically solvable equations, so to enable deployment in embedded systems. Besides proposing the estimators we moreover analyze their statistical performance both in simulation and with field tests. We report the dependency of the MSE performance of the calibration procedure as a function of the sensor noise levels, and observe that in field tests the approach can lead to a tenfold improvement in the accuracy of the raw measurements.
We present an improved statistical model of the measurement process of triangulation Light Detection and Rangings (Lidars) that takes into account bias and variance effects coming from two different sources of uncertainty: (i) mechanical imperfections on the geometry and properties of their pinhole lens - CCD camera systems, and (ii) inaccuracies in the measurement of the angular displacement of the sensor due to non ideal measurements from the internal encoder of the sensor. This model extends thus the one presented in [2] by adding this second source of errors. Besides proposing the statistical model, this chapter considers: (i) specialized and dedicated model calibration algorithms that exploit Maximum Likelihood (ML)/Akaike Information Criterion (AIC) concepts and that use training datasets collected in a controlled setup, and (ii) tailored statistical strategies that use the calibration results to statistically process the raw sensor measurements in non controlled but structured environments where there is a high chance for the sensor to be detecting objects with flat surfaces (e.g., walls). These newly proposed algorithms are thus specially designed and optimized for inferring precisely the angular orientation of the Lidar sensor with respect to the detected object, a feature that is beneficial especially for indoor navigation purposes.
This paper describes a new calibration procedure for distance sensors that does not require independent sources of groundtruth information, i.e., that is not based on comparing the measurements from the uncalibrated sensor against measurements from a precise device assumed as the groundtruth. Alternatively, the procedure assumes that the uncalibrated distance sensor moves in space on a straight line in an environment with fixed targets, so that the intrinsic parameters of the statistical model of the sensor readings are calibrated without requiring tests in controlled environments, but rather in environments where the sensor follows linear movement and objects do not move. The proposed calibration procedure exploits an approximated expectation maximization scheme on top of two ingredients: an heteroscedastic statistical model describing the measurement process, and a simplified dynamical model describing the linear sensor movement. The procedure is designed to be capable of not just estimating the parameters of one generic distance sensor, but rather integrating the most common sensors in robotic applications, such as Lidars, odometers, and sonar rangers and learn the intrinsic parameters of all these sensors simultaneously. Tests in a controlled environment led to a reduction of the mean squared error of the measurements returned by a commercial triangulation Lidar by a factor between 3 and 6, comparable to the efficiency of other state-of-the art groundtruth-based calibration procedures. Adding odometric and ultrasonic information further improved the performance index of the overall distance estimation strategy by a factor of up to 1.2. Tests also show high robustness against violating the linear movements assumption.
We propose an expectation maximization (EM) strategy for improving the precision of time of flight (ToF) light detection and ranging (LiDAR) scanners. The novel algorithm statistically accounts not only for the bias induced by temperature changes in the laser diode, but also for the multi-modality of the measurement noises that is induced by mode-hopping effects. Instrumental to the proposed EM algorithm, we also describe a general thermal dynamics model that can be learned either from just input-output data or from a combination of simple temperature experiments and information from the laser’s datasheet. We test the strategy on a SICK LMS 200 device and improve its average absolute error by a factor of three.
We aim at developing statistical tools that improve the accuracy and precision of the measurements returned by triangulation Light Detection and Rangings (Lidars). To this aim we: i) propose and validate a novel model that describes the statistics of the measurements of these Lidars, and that is built starting from mechanical considerations on the geometry and properties of their pinhole lens - CCD camera systems; ii) build, starting from this novel statistical model, a Maximum Likelihood (ML) / Akaike Information Criterion (AIC) -based sensor calibration algorithm that exploits training information collected in a controlled environment; iii) develop ML and Least Squares (LS) strategies that use the calibration results to statistically process the raw sensor measurements in non controlled environments. The overall technique allowed us to obtain empirical improvements of the normalized Mean Squared Error (MSE) from 0.0789 to 0.0046
Electronic warfare (EW) receivers are passive receivers which receive emissions from other platforms, and do certain analysis on these emissions. Some EW receivers receive radar pulses, measure the parameter of each pulse received and group the pulses that belongs to the same emitter together to determine the radar parameters for each emitter. These parameters are then compared with values stored for known radar types, to identify the emitter type. Two parts are focused, emitters deinterleaving and PRF-type identification. The deinterleaving is done through parameters clustering. Two parameters are selected for clustering direction of arrival and radio frequency. A self-organising neural network called Fuzzy ART is proposed for clustering. This algorithm has a very good clustering quality and can run in real-time applications. The PRF-type identification is done through time-of-arrival (TOA) analysis. Three previously presented algorithms are combined in new scheme to do the TOA analysis (or PRF-type identification). These algorithms are difference TOA histogram, TOA folding histogram and sequence search algorithm. The complete proposed system has been tested using three different tests. These tests are simple PRI test, jittered PRI test and staggered PRI test. The proposed system identifies up to 90 simple emitters, 20 jittered emitters and 20 staggered emitters. In all tests, the data were simulated and generated using software.