A Floating-point Extended Kalman Filter Implementation for Autonomous Mobile Robots

Localization and Mapping are two of the most important capabilities for autonomous mobile robots and have been receiving considerable attention from the scientific computing community over the last 10 years. One of the most efficient methods to address these problems is based on the use of the Extended Kalman Filter (EKF). The EKF simultaneously estimates a model of the environment (map) and the position of the robot based on odometric and exteroceptive sensor information. As this algorithm demands a considerable amount of computation, it is usually executed on high end PCs coupled to the robot. In this work we present an FPGA-based architecture for the EKF algorithm that is capable of processing two-dimensional maps containing up to 1.8 k features at real time (14 Hz), a three-fold improvement over a Pentium M 1.6 GHz, and a 13-fold improvement over an ARM920T 200 MHz. The proposed architecture also consumes only 1.3% of the Pentium and 12.3% of the ARM energy per feature.

Pairs trading : aplicação no mercado acionário brasileiro

Esta dissertação estuda a aplicação da estratégia Pairs Trading no mercado acionário brasileiro. Envolve basicamente a identificação de pares de ações que tenham movimentos de preço semelhantes e posteriormente a operação do diferencial entre seus preços. É possível observar no mercado a existência de um valor de equilíbrio de longo prazo para alguns pares de ações e quando o diferencial divergir de certa quantidade pré-definida opera-se o diferencial no intuito de que o mesmo retorne ao seu valor de equilíbrio de longo prazo, ou seja, espera-se que ocorra uma reversão à média do diferencial. A metodologia para a identificação desses pares de ações que descrevem movimentos semelhantes vem do conceito de cointegração. Essa metodologia é aplicada sobre as ações do índice Bovespa de 04-Jan-1993 a 30-Jun-2005. Inicialmente é feita uma pré-seleção dos pares de ações via correlação, ou medida de distância. Para a seleção final é feito o teste de cointegração, onde é utilizado o método de Dickey-Fuller aumentado (Augmented Dickey-Fuller test – ADF) para verificar a existência de raiz unitária da série de resíduo da combinação linear do logaritmo dos preços. Após a seleção, os pares são simulados historicamente (backtesting) para se analisar a performance dos retornos dos pares de ações, incluindo também os custos operacionais.

Pairs trading: uma aplicação ao mercado acionário brasileiro

Neste trabalho, verificamos viabilidade de aplicação da estratégia de pairs trading no mercado acionário brasileiro. Diferentemente de outros estudos do mesmo tema, construímos ativos sintéticos a partir de uma combinação linear de preços de ações. Conforme Burgeois e Minko (2005), utilizamos a metodologia de Johansen para a formação dos pares a serem testados. Após a identificação de pares cointegrados, para assegurar a estacionaridade do ativo sintético contruído a partir da relação linear de preços das ações, utilizamos os testes DF-GLS e KPSS e filtramos àqueles que apresentavam raiz unitária em sua série de tempo. A seguir, simulamos a estratégia (backtesting) com os pares selecionados e para encontrar os melhores parâmetros, testamos diferentes períodos de formação dos pares, de operação e de parâmetros de entrada, saída e stop-loss. A fim de realizarmos os testes de forma mais realista possível, incluímos os custos de corretagem, de emolumentos e de aluguel, além de adicionar um lag de um dia para a realização das operações

Estratégia de cointegração dinâmica empírica para arbitragem estatística e trading

Este trabalho primeiramente explora fundamentos teóricos básicos para análise e implementação de algoritmos para a modelagem de séries temporais. A finalidade principal da modelagem de séries temporais será a predição para utilizá-la na arbitragem estatística. As séries utilizadas são retiradas de uma base de histórico do mercado de ações brasileiro. Estratégias de arbitragem estatística, mais especificamente pairs trading, utilizam a característica de reversão à média dos modelos para explorar um lucro potencial quando o módulo do spread está estatisticamente muito afastado de sua média. Além disso, os modelos dinâmicos deste trabalho apresentam parâmetros variantes no tempo que aumentam a sua flexibilidade e adaptabilidade em mudanças estruturais do processo. Os pares do algoritmo de pairs trading são escolhidos selecionando ativos de mesma empresa ou índices e ETFs (Exchange Trade Funds). A validação da escolha dos pares é feita utilizando testes de cointegração. As simulações demonstram os resultados dos testes de cointegração, a variação no tempo dos parâmetros do modelo e o resultado de um portfólio fictício.

Localization of a mobile robot based on odometry and natural landmarks using extended kalman filter

SANTANA, André M.; SOUZA, Anderson A. S.; BRITTO, Ricardo S.; ALSINA, Pablo J.; MEDEIROS, Adelardo A. D. Localization of a mobile robot based on odometry and natural landmarks using extended Kalman Filter. In: INTERNATIONAL CONFERENCE ON INFORMATICS IN CONTROL, AUTOMATION AND ROBOTICS, 5., 2008, Funchal, Portugal. Proceedings... Funchal, Portugal: ICINCO, 2008.

A Discussion Related to Orbit Determination Using Nonlinear Sigma Point Kalman Filter

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Comparative analysis of synchronization algorithms based on PLL, RDFT and Kalman Filter

This paper discusses the main characteristics and presents a comparative analysis of three synchronization algorithms based respectively, on a Phase-Locked Loop, a Kalman Filter and a Discrete Fourier Transform. It will be described the single and three-phase models of the first two methods and the single-phase model of the third one. Details on how to modify the filtering properties or dynamic response of each algorithm will be discussed in terms of their design parameters. In order to compare the different algorithms, these parameters will be set for maximum filter capability. Then, the dynamic response, during input amplitude and frequency deviations will be observed, as well as during the initialization procedure. So, advantages and disadvantages of all considered algorithms will be discussed. ©2007 IEEE.

Trade-growth relationship in Cuba: estimation using the Kalman filter

Includes bibliography

Kalman Filter approaches on crane swing

The purpose of this study is to explore a Kalman Filter approach to estimating swing of crane-suspended loads. Measuring real-time swing is needed to implement swing damping control strategies where crane joints are used to remove energy from a swinging load. The typical solution to measuring swing uses an inertial sensor attached to the hook block. Measured hook block twist is used to resolve the other two sensed body rates into tangential and radial swing. Uncertainty in the twist measurement leads to inaccurate tangential and radial swing calculations and ineffective swing damping. A typical mitigation approach is to bandpass the inertial sensor readings to remove low frequency drift and high frequency noise. The center frequency of the bandpass filter is usually designed to track the load length and the pass band width set to trade off performance with damping loop gain. The Kalman Filter approach developed here allows all swing motions (radial, tangential and twist) to be measured without the use of a bandpass filter. This provides an alternate solution for swing damping control implementation. After developing a Kalman Filter solution for a two-dimensional swing scenario, the three-dimensional system is considered where simplifying assumptions, suggested by the two-dimensional study, are exploited. One of the interesting aspects of the three-dimensional study is the hook block twist model. Unlike the mass-independence of a pendulum's natural frequency, the twist natural frequency depends both on the pendulum length and the load’s mass distribution. The linear Kalman Filter is applied to experimental data demonstrating the ability to extract the individual swing components for complex motions. It should be noted that the three-dimensional simplifying assumptions preclude the ability to measure two "secondary" hook block rotations. The ability to segregate these motions from the primary swing degrees of freedom was illustrated in the two-dimensional study and could be included into the three-dimensional solution if they were found to be important for a particular application.

An Ensemble Kalman Filter multi-tracer assimilation: Determining uncertain ocean model parameters for improved climate-carbon cycle projections

An Ensemble Kalman Filter is applied to assimilate observed tracer fields in various combinations in the Bern3D ocean model. Each tracer combination yields a set of optimal transport parameter values that are used in projections with prescribed CO2 stabilization pathways. The assimilation of temperature and salinity fields yields a too vigorous ventilation of the thermocline and the deep ocean, whereas the inclusion of CFC-11 and radiocarbon improves the representation of physical and biogeochemical tracers and of ventilation time scales. Projected peak uptake rates and cumulative uptake of CO2 by the ocean are around 20% lower for the parameters determined with CFC-11 and radiocarbon as additional target compared to those with salinity and temperature only. Higher surface temperature changes are simulated in the Greenland–Norwegian–Iceland Sea and in the Southern Ocean when CFC-11 is included in the Ensemble Kalman model tuning. These findings highlights the importance of ocean transport calibration for the design of near-term and long-term CO2 emission mitigation strategies and for climate projections.

Methodology for adapting the parameters of a fuzzy system using the extended Kalman filter

When we try to analyze and to control a system whose model was obtained only based on input/output data, accuracy is essential in the model. On the other hand, to make the procedure practical, the modeling stage must be computationally efficient. In this regard, this paper presents the application of extended Kalman filter for the parametric adaptation of a fuzzy model

Application of the Extended Kalman filter to fuzzy modeling: Algorithms and practical implementation

Modeling phase is fundamental both in the analysis process of a dynamic system and the design of a control system. If this phase is in-line is even more critical and the only information of the system comes from input/output data. Some adaptation algorithms for fuzzy system based on extended Kalman filter are presented in this paper, which allows obtaining accurate models without renounce the computational efficiency that characterizes the Kalman filter, and allows its implementation in-line with the process

Improved Animal Tracking Algorithm using Distributed Kalman Filter-based Algortihms

Animal tracking has been addressed by different initiatives over the last two decades. Most of them rely on satellite connectivity on every single node and lack of energy-saving strategies. This paper presents several new contributions on the tracking of dynamic heterogeneous asynchronous networks (primary nodes with GPS and secondary nodes with a kinetic generator) motivated by the animal tracking paradigm with random transmissions. A simple approach based on connectivity and coverage intersection is compared with more sophisticated algorithms based on ad-hoc implementations of distributed Kalman-based filters that integrate measurement information using Consensus principles in order to provide enhanced accuracy. Several simulations varying the coverage range, the random behavior of the kinetic generator (modeled as a Poisson Process) and the periodic activation of GPS are included. In addition, this study is enhanced with HW developments and implementations on commercial off-the-shelf equipment which show the feasibility for performing these proposals on real hardware.