Development of a sensorimotor algorithm able to deal with unforeseen pushes and its implementation based on VHDL

Development of a Sensorimotor Algorithm Able to Deal with Unforeseen Pushes and Its Implementation Based on VHDL is the title of my thesis which concludes my Bachelor Degree in the Escuela Técnica Superior de Ingeniería y Sistemas de Telecomunicación of the Universidad Politécnica de Madrid. It encloses the overall work I did in the Neurorobotics Research Laboratory from the Beuth Hochschule für Technik Berlin during my ERASMUS year in 2015. This thesis is focused on the field of robotics, specifically an electronic circuit called Cognitive Sensorimotor Loop (CSL) and its control algorithm based on VHDL hardware description language. The reason that makes the CSL special resides in its ability to operate a motor both as a sensor and an actuator. This way, it is possible to achieve a balanced position in any of the robot joints (e.g. the robot manages to stand) without needing any conventional sensor. In other words, the back electromotive force (EMF) induced by the motor coils is measured and the control algorithm responds depending on its magnitude. The CSL circuit contains mainly an analog-to-digital converter (ADC) and a driver. The ADC consists on a delta-sigma modulation which generates a series of bits with a certain percentage of 1's and 0's, proportional to the back EMF. The control algorithm, running in a FPGA, processes the bit frame and outputs a signal for the driver. This driver, which has an H bridge topology, gives the motor the ability to rotate in both directions while it's supplied with the power needed. The objective of this thesis is to document the experiments and overall work done on push ignoring contractive sensorimotor algorithms, meaning sensorimotor algorithms that ignore large magnitude forces (compared to gravity) applied in a short time interval on a pendulum system. This main objective is divided in two sub-objectives: (1) developing a system based on parameterized thresholds and (2) developing a system based on a push bypassing filter. System (1) contains a module that outputs a signal which blocks the main Sensorimotor algorithm when a push is detected. This module has several different parameters as inputs e.g. the back EMF increment to consider a force as a push or the time interval between samples. System (2) consists on a low-pass Infinite Impulse Response digital filter. It cuts any frequency considered faster than a certain push oscillation. This filter required an intensive study on how to implement some functions and data types (fixed or floating point data) not supported by standard VHDL packages. Once this was achieved, the next challenge was to simplify the solution as much as possible, without using non-official user made packages. Both systems behaved with a series of interesting advantages and disadvantages for the elaboration of the document. Stability, reaction time, simplicity or computational load are one of the many factors to be studied in the designed systems. RESUMEN. Development of a Sensorimotor Algorithm Able to Deal with Unforeseen Pushes and Its Implementation Based on VHDL es un Proyecto de Fin de Grado (PFG) que concluye mis estudios en la Escuela Técnica Superior de Ingeniería y Sistemas de Telecomunicación de la Universidad Politécnica de Madrid. En él se documenta el trabajo de investigación que realicé en el Neurorobotics Research Laboratory de la Beuth Hochschule für Technik Berlin durante el año 2015 mediante el programa de intercambio ERASMUS. Este PFG se centra en el campo de la robótica y en concreto en un circuito electrónico llamado Cognitive Sensorimotor Loop (CSL) y su algoritmo de control basado en lenguaje de modelado hardware VHDL. La particularidad del CSL reside en que se consigue que un motor haga las veces tanto de sensor como de actuador. De esta manera es posible que las articulaciones de un robot alcancen una posición de equilibrio (p.ej. el robot se coloca erguido) sin la necesidad de sensores en el sentido estricto de la palabra. Es decir, se mide la propia fuerza electromotriz (FEM) inducida sobre el motor y el algoritmo responde de acuerdo a su magnitud. El circuito CSL se compone de un convertidor analógico-digital (ADC) y un driver. El ADC consiste en un modulador sigma-delta, que genera una serie de bits con un porcentaje de 1's y 0's determinado, en proporción a la magnitud de la FEM inducida. El algoritmo de control, que se ejecuta en una FPGA, procesa esta cadena de bits y genera una señal para el driver. El driver, que posee una topología en puente H, provee al motor de la potencia necesaria y le otorga la capacidad de rotar en cualquiera de las dos direcciones. El objetivo de este PFG es documentar los experimentos y en general el trabajo realizado en algoritmos Sensorimotor que puedan ignorar fuerzas de gran magnitud (en comparación con la gravedad) y aplicadas en una corta ventana de tiempo. En otras palabras, ignorar empujones conservando el comportamiento original frente a la gravedad. Para ello se han desarrollado dos sistemas: uno basado en umbrales parametrizados (1) y otro basado en un filtro de corte ajustable (2). El sistema (1) contiene un módulo que, en el caso de detectar un empujón, genera una señal que bloquea el algoritmo Sensorimotor. Este módulo recibe diferentes parámetros como el incremento necesario de la FEM para que se considere un empujón o la ventana de tiempo para que se considere la existencia de un empujón. El sistema (2) consiste en un filtro digital paso-bajo de respuesta infinita que corta cualquier variación que considere un empujón. Para crear este filtro se requirió un estudio sobre como implementar ciertas funciones y tipos de datos (coma fija o flotante) no soportados por las librerías básicas de VHDL. Tras esto, el objetivo fue simplificar al máximo la solución del problema, sin utilizar paquetes de librerías añadidos. En ambos sistemas aparecen una serie de ventajas e inconvenientes de interés para el documento. La estabilidad, el tiempo de reacción, la simplicidad o la carga computacional son algunas de las muchos factores a estudiar en los sistemas diseñados. Para concluir, también han sido documentadas algunas incorporaciones a los sistemas: una interfaz visual en VGA, un módulo que compensa el offset del ADC o la implementación de una batería de faders MIDI entre otras.

Reconfigurable computing for Monte Carlo simulations: results and prospects of the Janus project

We describe Janus, a massively parallel FPGA-based computer optimized for the simulation of spin glasses, theoretical models for the behavior of glassy materials. FPGAs (as compared to GPUs or many-core processors) provide a complementary approach to massively parallel computing. In particular, our model problem is formulated in terms of binary variables, and floating-point operations can be (almost) completely avoided. The FPGA architecture allows us to run many independent threads with almost no latencies in memory access, thus updating up to 1024 spins per cycle. We describe Janus in detail and we summarize the physics results obtained in four years of operation of this machine; we discuss two types of physics applications: long simulations on very large systems (which try to mimic and provide understanding about the experimental non equilibrium dynamics), and low-temperature equilibrium simulations using an artificial parallel tempering dynamics. The time scale of our non-equilibrium simulations spans eleven orders of magnitude (from picoseconds to a tenth of a second). On the other hand, our equilibrium simulations are unprecedented both because of the low temperatures reached and for the large systems that we have brought to equilibrium. A finite-time scaling ansatz emerges from the detailed comparison of the two sets of simulations. Janus has made it possible to perform spin glass simulations that would take several decades on more conventional architectures. The paper ends with an assessment of the potential of possible future versions of the Janus architecture, based on state-of-the-art technology.

Biblioteca Aritmética de operaciones en Tiempo Real para Números en Coma Flotante

Lenguaje de alto nivel utilizado: Java

Boston, Massachusetts Region, Digital Elevation Model with Bathymetry

This raster layer represents surface elevation and bathymetry data for the Boston Region, Massachusetts. It was created by merging portions of MassGIS Digital Elevation Model 1:5,000 (2005) data with NOAA Estuarine Bathymetric Digital Elevation Models (30 m.) (1998). DEM data was derived from the digital terrain models that were produced as part of the MassGIS 1:5,000 Black and White Digital Orthophoto imagery project. Cellsize is 5 meters by 5 meters. Each cell has a floating point value, in meters, which represents its elevation above or below sea level.

Time complexity and convergence analysis of domain theoretic Picard method

We present an implementation of the domain-theoretic Picard method for solving initial value problems (IVPs) introduced by Edalat and Pattinson [1]. Compared to Edalat and Pattinson's implementation, our algorithm uses a more efficient arithmetic based on an arbitrary precision floating-point library. Despite the additional overestimations due to floating-point rounding, we obtain a similar bound on the convergence rate of the produced approximations. Moreover, our convergence analysis is detailed enough to allow a static optimisation in the growth of the precision used in successive Picard iterations. Such optimisation greatly improves the efficiency of the solving process. Although a similar optimisation could be performed dynamically without our analysis, a static one gives us a significant advantage: we are able to predict the time it will take the solver to obtain an approximation of a certain (arbitrarily high) quality.

Approximation Classes for Adaptive Methods

* This work has been supported by the Office of Naval Research Contract Nr. N0014-91-J1343, the Army Research Office Contract Nr. DAAD 19-02-1-0028, the National Science Foundation grants DMS-0221642 and DMS-0200665, the Deutsche Forschungsgemeinschaft grant SFB 401, the IHP Network “Breaking Complexity” funded by the European Commission and the Alexan- der von Humboldt Foundation.

A Mathematical Basis for an Interval Arithmetic Standard

Basic concepts for an interval arithmetic standard are discussed in the paper. Interval arithmetic deals with closed and connected sets of real numbers. Unlike floating-point arithmetic it is free of exceptions. A complete set of formulas to approximate real interval arithmetic on the computer is displayed in section 3 of the paper. The essential comparison relations and lattice operations are discussed in section 6. Evaluation of functions for interval arguments is studied in section 7. The desirability of variable length interval arithmetic is also discussed in the paper. The requirement to adapt the digital computer to the needs of interval arithmetic is as old as interval arithmetic. An obvious, simple possible solution is shown in section 8.

Extension of the C-XSC Library with Scalar Products with Selectable Accuracy

The C++ class library C-XSC for scientific computing has been extended with the possibility to compute scalar products with selectable accuracy in version 2.3.0. In previous versions, scalar products have always been computed exactly with the help of the so-called long accumulator. Additionally, optimized floating point computation of matrix and vector operations using BLAS-routines are added in C-XSC version 2.4.0. In this article the algorithms used and their implementations, as well as some potential pitfalls in the compilation, are described in more detail. Additionally, the theoretical background of the employed DotK algorithm and the necessary modifications of the concrete implementation in C-XSC are briefly explained. Run-time tests and numerical examples are presented as well.

Numerical Issues in Using MATLAB

Михаил Константинов, Весела Пашева, Петко Петков - Разгледани са някои числени проблеми при използването на компютърната система MATLAB в учебната дейност: пресмятане на тригонометрични функции, повдигане на матрица на степен, спектрален анализ на целочислени матрици от нисък ред и пресмятане на корените на алгебрични уравнения. Причините за възникналите числени трудности могат да се обяснят с особеностите на използваната двоичната аритметика с плаваща точка.

Prospective evaluation of three point of care devices for glycemia measurement in a neonatal intensive care unit.

Hypoglycemia, if recurrent, may have severe consequences on cognitive and psychomotor development of neonates. Therefore, screening for hypoglycemia is a daily routine in every facility taking care of newborn infants. Point-of-care-testing (POCT) devices are interesting for neonatal use, as their handling is easy, measurements can be performed at bedside, demanded blood volume is small and results are readily available. However, such whole blood measurements are challenged by a wide variation of hematocrit in neonates and a spectrum of normal glucose concentration at the lower end of the test range. We conducted a prospective trial to check precision and accuracy of the best suitable POCT device for neonatal use from three leading companies in Europe. Of the three devices tested (Precision Xceed, Abbott; Elite XL, Bayer; Aviva Nano, Roche), Aviva Nano exhibited the best precision. None completely fulfilled the ISO-accuracy-criteria 15197: 2003 or 2011. Aviva Nano fulfilled these criteria in 92% of cases while the others were <87%. Precision Xceed reached the 95% limit of the 2003 ISO-criteria for values ≤4.2 mmol/L, but not for the higher range (71%). Although validated for adults, new POCT devices need to be specifically evaluated on newborn infants before adopting their routine use in neonatology.

Effect of the implementation of the Hazard Analysis Critical Control Point (HACCP) prerequisite program in an institutional foodservice unit in Southern Brazil

The aims of this study were to investigate the hygienic practices in the food production of an institutional foodservice unit in Southern Brazil and to evaluate the effect of implementing good food handling practices and standard operational procedures using microbiological hygiene indicators. An initial survey of the general operating conditions classified the unit as regular in terms of compliance with State safety guidelines for food service establishments. An action plan that incorporated the correction of noncompliance issues and the training of food handlers in good food handling practices and standard operational procedures were then implemented. The results of the microbiological analysis of utensils, preparation surfaces, food handlers' hands, water, and ambient air were recorded before and after the implementation of the action plan. The results showed that the implementation of this type of practice leads to the production of safer foods.

Tables of oblique photo scale factors and nadir point distances for 70mm photography. For unit altitude, focal lengths: 1.5, 3, 6, and 12 inches, tilts: 0 ̊to 90.̊

"Scale factor system, oblique photo, developmental. Contract NOas 59-6067-c, Aeronautics specification XPH 118 (modified)."

Tables of oblique photo scale factors and nadir point distances. For unit altitude, focal lengths: 6, 12, 24, 36, and 48 inches, tilts: 3 ̊to 90.̊

"Prepared by Cornell University for the U. S. Navy Bureau of Aeronautics under contract NOas 57-585-c."

Graphs of oblique photo scale factors and nadir point distances. For unit altitude, focal lengths: 6,12,24,36, and 48 inches; tilts: 0⁰ to 90⁰.

"Prepared ... for the U.S. Navy Bureau of Aeronautics, under contract NOas 57-585-c."