Accurate pedestrian indoor navigation by tightly coupling foot-mounted IMU and RFID measurements

We present a new method to accurately locate persons indoors by fusing inertial navigation system (INS) techniques with active RFID technology. A foot-mounted inertial measuring units (IMUs)-based position estimation method, is aided by the received signal strengths (RSSs) obtained from several active RFID tags placed at known locations in a building. In contrast to other authors that integrate IMUs and RSS with a loose Kalman filter (KF)-based coupling (by using the residuals of inertial- and RSS-calculated positions), we present a tight KF-based INS/RFID integration, using the residuals between the INS-predicted reader-to-tag ranges and the ranges derived from a generic RSS path-loss model. Our approach also includes other drift reduction methods such as zero velocity updates (ZUPTs) at foot stance detections, zero angular-rate updates (ZARUs) when the user is motionless, and heading corrections using magnetometers. A complementary extended Kalman filter (EKF), throughout its 15-element error state vector, compensates the position, velocity and attitude errors of the INS solution, as well as IMU biases. This methodology is valid for any kind of motion (forward, lateral or backward walk, at different speeds), and does not require an offline calibration for the user gait. The integrated INS+RFID methodology eliminates the typical drift of IMU-alone solutions (approximately 1% of the total traveled distance), resulting in typical positioning errors along the walking path (no matter its length) of approximately 1.5 m.

Multi-resolution model-based traffic sign detection and tracking

In this paper we propose an innovative approach to tackle the problem of traffic sign detection using a computer vision algorithm and taking into account real-time operation constraints, trying to establish intelligent strategies to simplify as much as possible the algorithm complexity and to speed up the process. Firstly, a set of candidates is generated according to a color segmentation stage, followed by a region analysis strategy, where spatial characteristic of previously detected objects are taken into account. Finally, temporal coherence is introduced by means of a tracking scheme, performed using a Kalman filter for each potential candidate. Taking into consideration time constraints, efficiency is achieved two-fold: on the one side, a multi-resolution strategy is adopted for segmentation, where global operation will be applied only to low-resolution images, increasing the resolution to the maximum only when a potential road sign is being tracked. On the other side, we take advantage of the expected spacing between traffic signs. Namely, the tracking of objects of interest allows to generate inhibition areas, which are those ones where no new traffic signs are expected to appear due to the existence of a TS in the neighborhood. The proposed solution has been tested with real sequences in both urban areas and highways, and proved to achieve higher computational efficiency, especially as a result of the multi-resolution approach.

Traffic sign detection and tracking using robust 3D analysis

In this paper we present an innovative technique to tackle the problem of automatic road sign detection and tracking using an on-board stereo camera. It involves a continuous 3D analysis of the road sign during the whole tracking process. Firstly, a color and appearance based model is applied to generate road sign candidates in both stereo images. A sparse disparity map between the left and right images is then created for each candidate by using contour-based and SURF-based matching in the far and short range, respectively. Once the map has been computed, the correspondences are back-projected to generate a cloud of 3D points, and the best-fit plane is computed through RANSAC, ensuring robustness to outliers. Temporal consistency is enforced by means of a Kalman filter, which exploits the intrinsic smoothness of the 3D camera motion in traffic environments. Additionally, the estimation of the plane allows to correct deformations due to perspective, thus easing further sign classification.

Modelling and Control of Stepper Motors for High Accuracy Positioning Systems Used in Radioactive Environments

Hybrid Stepper Motors are widely used in open-loop position applications. They are the choice of actuation for the collimators in the Large Hadron Collider, the largest particle accelerator at CERN. In this case the positioning requirements and the highly radioactive operating environment are unique. The latter forces both the use of long cables to connect the motors to the drives which act as transmission lines and also prevents the use of standard position sensors. However, reliable and precise operation of the collimators is critical for the machine, requiring the prevention of step loss in the motors and maintenance to be foreseen in case of mechanical degradation. In order to make the above possible, an approach is proposed for the application of an Extended Kalman Filter to a sensorless stepper motor drive, when the motor is separated from its drive by long cables. When the long cables and high frequency pulse width modulated control voltage signals are used together, the electrical signals difer greatly between the motor and drive-side of the cable. Since in the considered case only drive-side data is available, it is therefore necessary to estimate the motor-side signals. Modelling the entire cable and motor system in an Extended Kalman Filter is too computationally intensive for standard embedded real-time platforms. It is, in consequence, proposed to divide the problem into an Extended Kalman Filter, based only on the motor model, and separated motor-side signal estimators, the combination of which is less demanding computationally. The efectiveness of this approach is shown in simulation. Then its validity is experimentally demonstrated via implementation in a DSP based drive. A testbench to test its performance when driving an axis of a Large Hadron Collider collimator is presented along with the results achieved. It is shown that the proposed method is capable of achieving position and load torque estimates which allow step loss to be detected and mechanical degradation to be evaluated without the need for physical sensors. These estimation algorithms often require a precise model of the motor, but the standard electrical model used for hybrid stepper motors is limited when currents, which are high enough to produce saturation of the magnetic circuit, are present. New model extensions are proposed in order to have a more precise model of the motor independently of the current level, whilst maintaining a low computational cost. It is shown that a significant improvement in the model It is achieved with these extensions, and their computational performance is compared to study the cost of model improvement versus computation cost. The applicability of the proposed model extensions is demonstrated via their use in an Extended Kalman Filter running in real-time for closed-loop current control and mechanical state estimation. An additional problem arises from the use of stepper motors. The mechanics of the collimators can wear due to the abrupt motion and torque profiles that are applied by them when used in the standard way, i.e. stepping in open-loop. Closed-loop position control, more specifically Field Oriented Control, would allow smoother profiles, more respectful to the mechanics, to be applied but requires position feedback. As mentioned already, the use of sensors in radioactive environments is very limited for reliability reasons. Sensorless control is a known option but when the speed is very low or zero, as is the case most of the time for the motors used in the LHC collimator, the loss of observability prevents its use. In order to allow the use of position sensors without reducing the long term reliability of the whole system, the possibility to switch from closed to open loop is proposed and validated, allowing the use of closed-loop control when the position sensors function correctly and open-loop when there is a sensor failure. A different approach to deal with the switched drive working with long cables is also presented. Switched mode stepper motor drives tend to have poor performance or even fail completely when the motor is fed through a long cable due to the high oscillations in the drive-side current. The design of a stepper motor output fillter which solves this problem is thus proposed. A two stage filter, one devoted to dealing with the diferential mode and the other with the common mode, is designed and validated experimentally. With this ?lter the drive performance is greatly improved, achieving a positioning repeatability even better than with the drive working without a long cable, the radiated emissions are reduced and the overvoltages at the motor terminals are eliminated.

Modelos para el an��lisis din��mico y la predicci��n a corto plazo de los precios de la electricidad en mercados liberalizados

En esta tesis se va a describir y aplicar de forma novedosa la t��cnica del alisado exponencial multivariante a la predicci��n a corto plazo, a un d��a vista, de los precios horarios de la electricidad, un problema que se est�� estudiando intensivamente en la literatura estad��stica y econ��mica reciente. Se van a demostrar ciertas propiedades interesantes del alisado exponencial multivariante que permiten reducir el n��mero de par��metros para caracterizar la serie temporal y que al mismo tiempo permiten realizar un an��lisis din��mico factorial de la serie de precios horarios de la electricidad. En particular, este proceso multivariante de elevada dimensi��n se estimar�� descomponi��ndolo en un n��mero reducido de procesos univariantes independientes de alisado exponencial caracterizado cada uno por un solo par��metro de suavizado que variar�� entre cero (proceso de ruido blanco) y uno (paseo aleatorio). Para ello, se utilizar�� la formulaci��n en el espacio de los estados para la estimaci��n del modelo, ya que ello permite conectar esa secuencia de modelos univariantes m��s eficientes con el modelo multivariante. De manera novedosa, las relaciones entre los dos modelos se obtienen a partir de un simple tratamiento algebraico sin requerir la aplicaci��n del filtro de Kalman. De este modo, se podr��n analizar y poner al descubierto las razones ��ltimas de la din��mica de precios de la electricidad. Por otra parte, la vertiente pr��ctica de esta metodolog��a se pondr�� de manifiesto con su aplicaci��n pr��ctica a ciertos mercados el��ctricos spot, tales como Omel, Powernext y Nord Pool. En los citados mercados se caracterizar�� la evoluci��n de los precios horarios y se establecer��n sus predicciones compar��ndolas con las de otras t��cnicas de predicci��n. ABSTRACT This thesis describes and applies the multivariate exponential smoothing technique to the day-ahead forecast of the hourly prices of electricity in a whole new way. This problem is being studied intensively in recent statistics and economics literature. It will start by demonstrating some interesting properties of the multivariate exponential smoothing that reduce drastically the number of parameters to characterize the time series and that at the same time allow a dynamic factor analysis of the hourly prices of electricity series. In particular this very complex multivariate process of dimension 24 will be estimated by decomposing a very reduced number of univariate independent of exponentially smoothing processes each characterized by a single smoothing parameter that varies between zero (white noise process) and one (random walk). To this end, the formulation is used in the state space model for the estimation, since this connects the sequence of efficient univariate models to the multivariate model. Through a novel way, relations between the two models are obtained from a simple algebraic treatment without applying the Kalman filter. Thus, we will analyze and expose the ultimate reasons for the dynamics of the electricity price. Moreover, the practical aspect of this methodology will be shown by applying this new technique to certain electricity spot markets such as Omel, Powernext and Nord Pool. In those markets the behavior of prices will be characterized, their predictions will be formulated and the results will be compared with those of other forecasting techniques.

Glucose-Insulin regulator for type 1 diabetes using high order neural networks

In this paper a Glucose-Insulin regulator for Type 1 Diabetes using artificial neural networks (ANN) is proposed. This is done using a discrete recurrent high order neural network in order to identify and control a nonlinear dynamical system which represents the pancreas? beta-cells behavior of a virtual patient. The ANN which reproduces and identifies the dynamical behavior system, is configured as series parallel and trained on line using the extended Kalman filter algorithm to achieve a quickly convergence identification in silico. The control objective is to regulate the glucose-insulin level under different glucose inputs and is based on a nonlinear neural block control law. A safety block is included between the control output signal and the virtual patient with type 1 diabetes mellitus. Simulations include a period of three days. Simulation results are compared during the overnight fasting period in Open-Loop (OL) versus Closed- Loop (CL). Tests in Semi-Closed-Loop (SCL) are made feedforward in order to give information to the control algorithm. We conclude the controller is able to drive the glucose to target in overnight periods and the feedforward is necessary to control the postprandial period.

Validaci��n local de la temperatura superficial del mar del sensor MODIS en aguas someras del Mediterr��neo occidental

La temperatura superficial del mar (SST) estimada a partir de los productos 11 ��m diurnos y nocturnos y 4 ��m nocturnos del sensor MODIS (Moderate Resolution Imaging Spectroradiometer) a bordo de la plataforma Aqua, han sido comparados con datos medidos in situ a tres profundidades diferentes (15, 50 y 100 cm) en una zona costera del Mediterr��neo Occidental. Esta comparaci��n ha permitido analizar la incertidumbre que existe en la estimaci��n de este par��metro en aguas someras y pr��ximas a la costa mediante im��genes de sat��lite de baja resoluci��n espacial. Los resultados obtenidos demuestran que el producto diurno SST_11 ��m, obtiene los estad��sticos RMSE (error cuadr��tico medio) y r2 (coeficiente de correlaci��n de Pearson) m��s ajustados con valores de 1��C y 0,96, respectivamente, para la profundidad 50 cm.

Control of Redundant Joint Structures Using Image Information During the Tracking of Non-Smooth Trajectories

Visual information is increasingly being used in a great number of applications in order to perform the guidance of joint structures. This paper proposes an image-based controller which allows the joint structure guidance when its number of degrees of freedom is greater than the required for the developed task. In this case, the controller solves the redundancy combining two different tasks: the primary task allows the correct guidance using image information, and the secondary task determines the most adequate joint structure posture solving the possible joint redundancy regarding the performed task in the image space. The method proposed to guide the joint structure also employs a smoothing Kalman filter not only to determine the moment when abrupt changes occur in the tracked trajectory, but also to estimate and compensate these changes using the proposed filter. Furthermore, a direct visual control approach is proposed which integrates the visual information provided by this smoothing Kalman filter. This last aspect permits the correct tracking when noisy measurements are obtained. All the contributions are integrated in an application which requires the tracking of the faces of Asperger children.

Estimation of Key Dates and Stages in Rice Crops Using Dual-Polarization SAR Time Series and a Particle Filtering Approach

Information of crop phenology is essential for evaluating crop productivity. In a previous work, we determined phenological stages with remote sensing data using a dynamic system framework and an extended Kalman filter (EKF) approach. In this paper, we demonstrate that the particle filter is a more reliable method to infer any phenological stage compared to the EKF. The improvements achieved with this approach are discussed. In addition, this methodology enables the estimation of key cultivation dates, thus providing a practical product for many applications. The dates of some important stages, as the sowing date and the day when the crop reaches the panicle initiation stage, have been chosen to show the potential of this technique.

Dynamical Approach for Real-Time Monitoring of Agricultural Crops

In this paper, a novel approach for exploiting multitemporal remote sensing data focused on real-time monitoring of agricultural crops is presented. The methodology is defined in a dynamical system context using state-space techniques, which enables the possibility of merging past temporal information with an update for each new acquisition. The dynamic system context allows us to exploit classical tools in this domain to perform the estimation of relevant variables. A general methodology is proposed, and a particular instance is defined in this study based on polarimetric radar data to track the phenological stages of a set of crops. A model generation from empirical data through principal component analysis is presented, and an extended Kalman filter is adapted to perform phenological stage estimation. Results employing quad-pol Radarsat-2 data over three different cereals are analyzed. The potential of this methodology to retrieve vegetation variables in real time is shown.

Piloto autom��tico: Sistema de navega����o aut��noma

Desde o in��cio do crescente interesse na ��rea de rob��tica que a navega����o aut��noma se apresenta como um problema de complexa resolu����o que, por isso, desperta vasto interesse no meio cient��fico. Al��m disso, as capacidades da navega����o aut��noma aliadas �� rob��tica permitem o desenvolvimento de variadas aplica����es. O objectivo da navega����o aut��noma �� conferir, a um dispositivo motor, capacidade de decis��o relativa �� locomo����o. Para o efeito, utilizam-se sensores, como os sensores IMU, o receptor GPS e os encoders, para fornecer os dados essenciais �� navega����o. A dificuldade encontra-se no correcto processamento destes sinais uma vez que s��o suscept��veis a fontes de ru��do. Este trabalho apresenta um sistema de navega����o aut��nomo aplicado ao controlo de um robot. Para tal, desenvolveu-se uma aplica����o que alberga todo o sistema de localiza����o, navega����o e controlo, acrescido de uma interface gr��fica, que permite a visualiza����o em mapa da movimenta����o aut��noma do robot. Recorre-se ao Filtro de Kalman como m��todo probabil��stico de estima����o de posi����o, em que os sinais dos v��rios sensores s��o conjugados e filtrados. Foram realizados v��rios testes de modo a avaliar a capacidade do robot atingir os pontos tra��ados e a sua autonomia no seguimento da traject��ria pretendida.

Validation of the Swiss methane emission inventory by atmospheric observations and inverse modelling

Atmospheric inverse modelling has the potential to provide observation-based estimates of greenhouse gas emissions at the country scale, thereby allowing for an independent validation of national emission inventories. Here, we present a regional-scale inverse modelling study to quantify the emissions of methane (CH���) from Switzerland, making use of the newly established CarboCount-CH measurement network and a high-resolution Lagrangian transport model. In our reference inversion, prior emissions were taken from the "bottom-up" Swiss Greenhouse Gas Inventory (SGHGI) as published by the Swiss Federal Office for the Environment in 2014 for the year 2012. Overall we estimate national CH��� emissions to be 196 �� 18 Gg yr����� for the year 2013 (1�� uncertainty). This result is in close agreement with the recently revised SGHGI estimate of 206 �� 33 Gg yr����� as reported in 2015 for the year 2012. Results from sensitivity inversions using alternative prior emissions, uncertainty covariance settings, large-scale background mole fractions, two different inverse algorithms (Bayesian and extended Kalman filter), and two different transport models confirm the robustness and independent character of our estimate. According to the latest SGHGI estimate the main CH��� source categories in Switzerland are agriculture (78 %), waste handling (15 %) and natural gas distribution and combustion (6 %). The spatial distribution and seasonal variability of our posterior emissions suggest an overestimation of agricultural CH��� emissions by 10 to 20 % in the most recent SGHGI, which is likely due to an overestimation of emissions from manure handling. Urban areas do not appear as emission hotspots in our posterior results, suggesting that leakages from natural gas distribution are only a minor source of CH��� in Switzerland. This is consistent with rather low emissions of 8.4 Gg yr����� reported by the SGHGI but inconsistent with the much higher value of 32 Gg yr����� implied by the EDGARv4.2 inventory for this sector. Increased CH��� emissions (up to 30 % compared to the prior) were deduced for the north-eastern parts of Switzerland. This feature was common to most sensitivity inversions, which is a strong indicator that it is a real feature and not an artefact of the transport model and the inversion system. However, it was not possible to assign an unambiguous source process to the region. The observations of the CarboCount-CH network provided invaluable and independent information for the validation of the national bottom-up inventory. Similar systems need to be sustained to provide independent monitoring of future climate agreements.

Controlled trial of hyperbaric oxygen treatment for alkali corneal burn in the rabbit

Purpose: To evaluate the efficacy of hyperbaric oxygen therapy in the treatment of alkali-induced corneal burns in an animal model. Methods: Twenty-four rabbits were randomized into a control group (n = 12) and hyperbaric oxygen treatment group (n = 12). After induction of anaesthesia, the alkali burn model was established by application of 1 N sodium hydroxide to one eye of each rabbit. The hyperbaric oxygen treatment group was treated each day for 21 days with hyperbaric oxygen at 2.4 Atmospheres Absolute (ATA) for 1 h. The eyes of the animals were examined daily for 2 weeks and then weekly until the end of the trial. The principal endpoint was that of perforation of the cornea at which time the animals were killed with a lethal dose of either intravenous or intraperitoneal barbiturate and the eyes immediately enucleated and fixed in 10% neutral buffered formalin. All animals in which complete healing took placed were also killed, the eyes removed, fixed and examined histologically. Photographs were taken of the rabbit's eyes at weekly intervals and the area of vascularization and epithelial defects in the hyperbaric and control groups were compared. Results: Equal numbers (seven) of the control and hyperbaric oxygen treated groups had perforated corneas and there was no statistical difference in the mean time to perforation (control 30.1 days; treated 30 days). There was also no statistical difference between the two groups with respect to epithelial defect size. Conclusion: Treatment with hyperbaric oxygen for 1 h daily for 21 days had no beneficial effect on alkali-induced corneal burns.

Multisensorial obstacles detection and classification for ADAS and autonomous driving

A reliable perception of the real world is a key-feature for an autonomous vehicle and the Advanced Driver Assistance Systems (ADAS). Obstacles detection (OD) is one of the main components for the correct reconstruction of the dynamic world. Historical approaches based on stereo vision and other 3D perception technologies (e.g. LIDAR) have been adapted to the ADAS first and autonomous ground vehicles, after, providing excellent results. The obstacles detection is a very broad field and this domain counts a lot of works in the last years. In academic research, it has been clearly established the essential role of these systems to realize active safety systems for accident prevention, reflecting also the innovative systems introduced by industry. These systems need to accurately assess situational criticalities and simultaneously assess awareness of these criticalities by the driver; it requires that the obstacles detection algorithms must be reliable and accurate, providing: a real-time output, a stable and robust representation of the environment and an estimation independent from lighting and weather conditions. Initial systems relied on only one exteroceptive sensor (e.g. radar or laser for ACC and camera for LDW) in addition to proprioceptive sensors such as wheel speed and yaw rate sensors. But, current systems, such as ACC operating at the entire speed range or autonomous braking for collision avoidance, require the use of multiple sensors since individually they can not meet these requirements. It has led the community to move towards the use of a combination of them in order to exploit the benefits of each one. Pedestrians and vehicles detection are ones of the major thrusts in situational criticalities assessment, still remaining an active area of research. ADASs are the most prominent use case of pedestrians and vehicles detection. Vehicles should be equipped with sensing capabilities able to detect and act on objects in dangerous situations, where the driver would not be able to avoid a collision. A full ADAS or autonomous vehicle, with regard to pedestrians and vehicles, would not only include detection but also tracking, orientation, intent analysis, and collision prediction. The system detects obstacles using a probabilistic occupancy grid built from a multi-resolution disparity map. Obstacles classification is based on an AdaBoost SoftCascade trained on Aggregate Channel Features. A final stage of tracking and fusion guarantees stability and robustness to the result.

Modelling of non-stationary processes using radial basis function networks

This paper reports preliminary progress on a principled approach to modelling nonstationary phenomena using neural networks. We are concerned with both parameter and model order complexity estimation. The basic methodology assumes a Bayesian foundation. However to allow the construction of pragmatic models, successive approximations have to be made to permit computational tractibility. The lowest order corresponds to the (Extended) Kalman filter approach to parameter estimation which has already been applied to neural networks. We illustrate some of the deficiencies of the existing approaches and discuss our preliminary generalisations, by considering the application to nonstationary time series.