Child programming: an adequate domain specific language for programming specific robots

Dissertação para obtenção do Grau de Mestre em Engenharia Informática

Haptic robot-environment interaction for self-supervised learning in ground mobility

Dissertação para obtenção do Grau de Mestre em Engenharia Eletrotécnica e de Computadores

A volumetric hybrid representation for obstacle detection in all-terrain robots

This dissertation presents an approach aimed at three-dimensional perception’s obstacle detection on all-terrain robots. Given the huge amount of acquired information, the adversities such environments present to an autonomous system and the swiftness, thus required, from each of its navigation decisions, it becomes imperative that the 3-D perceptional system to be able to map obstacles and passageways in the most swift and detailed manner. In this document, a hybrid approach is presented bringing the best of several methods together, combining the lightness of lesser meticulous analyses with the detail brought by more thorough ones. Realizing the former, a terrain’s slope mapping system upon a low resolute volumetric representation of the surrounding occupancy. For the latter’s detailed evaluation, two novel metrics were conceived to discriminate the little depth discrepancies found in between range scanner’s beam distance measurements. The hybrid solution resulting from the conjunction of these two representations provides a reliable answer to traversability mapping and a robust discrimination of penetrable vegetation from that constituting real obstructions. Two distinct robotic platforms offered the possibility to test the hybrid approach on very different applications: a boat, under an European project, the ECHORD Riverwatch, and a terrestrial four-wheeled robot for a national project, the Introsys Robot.

Heterogeneous computing with an algorithmic skeleton framework

The Graphics Processing Unit (GPU) is present in almost every modern day personal computer. Despite its specific purpose design, they have been increasingly used for general computations with very good results. Hence, there is a growing effort from the community to seamlessly integrate this kind of devices in everyday computing. However, to fully exploit the potential of a system comprising GPUs and CPUs, these devices should be presented to the programmer as a single platform. The efficient combination of the power of CPU and GPU devices is highly dependent on each device’s characteristics, resulting in platform specific applications that cannot be ported to different systems. Also, the most efficient work balance among devices is highly dependable on the computations to be performed and respective data sizes. In this work, we propose a solution for heterogeneous environments based on the abstraction level provided by algorithmic skeletons. Our goal is to take full advantage of the power of all CPU and GPU devices present in a system, without the need for different kernel implementations nor explicit work-distribution.To that end, we extended Marrow, an algorithmic skeleton framework for multi-GPUs, to support CPU computations and efficiently balance the work-load between devices. Our approach is based on an offline training execution that identifies the ideal work balance and platform configurations for a given application and input data size. The evaluation of this work shows that the combination of CPU and GPU devices can significantly boost the performance of our benchmarks in the tested environments, when compared to GPU-only executions.

Constraint reasoning with local search for continuous optimization

Optimization is a very important field for getting the best possible value for the optimization function. Continuous optimization is optimization over real intervals. There are many global and local search techniques. Global search techniques try to get the global optima of the optimization problem. However, local search techniques are used more since they try to find a local minimal solution within an area of the search space. In Continuous Constraint Satisfaction Problems (CCSP)s, constraints are viewed as relations between variables, and the computations are supported by interval analysis. The continuous constraint programming framework provides branch-and-prune algorithms for covering sets of solutions for the constraints with sets of interval boxes which are the Cartesian product of intervals. These algorithms begin with an initial crude cover of the feasible space (the Cartesian product of the initial variable domains) which is recursively refined by interleaving pruning and branching steps until a stopping criterion is satisfied. In this work, we try to find a convenient way to use the advantages in CCSP branchand- prune with local search of global optimization applied locally over each pruned branch of the CCSP. We apply local search techniques of continuous optimization over the pruned boxes outputted by the CCSP techniques. We mainly use steepest descent technique with different characteristics such as penalty calculation and step length. We implement two main different local search algorithms. We use “Procure”, which is a constraint reasoning and global optimization framework, to implement our techniques, then we produce and introduce our results over a set of benchmarks.

Wearable monitoring system for assistance dogs

Despite the recent progresses in robotics, autonomous robots still have too many limitations to reliably help people with disabilities. On the other hand, animals, and especially dogs, have already demonstrated great skills in assisting people in many daily situations. However, dogs also have their own set of limitations. For example, they need to rest periodically, to be healthy (physically and psychologically), and it is difficult to control them remotely. This project aims to “augment” the Assistance dog, by developing a system that compensates some of the dog weaknesses through a robotic device mounted on the dog harness. This specific study, involved in the COCHISE project, focuses on the development of a system for the monitoring of dogs activity and physiological parameters.

Towards an algorithmic skeleton framework for programming the Intel R Xeon PhiTM processor

The Intel R Xeon PhiTM is the first processor based on Intel’s MIC (Many Integrated Cores) architecture. It is a co-processor specially tailored for data-parallel computations, whose basic architectural design is similar to the ones of GPUs (Graphics Processing Units), leveraging the use of many integrated low computational cores to perform parallel computations. The main novelty of the MIC architecture, relatively to GPUs, is its compatibility with the Intel x86 architecture. This enables the use of many of the tools commonly available for the parallel programming of x86-based architectures, which may lead to a smaller learning curve. However, programming the Xeon Phi still entails aspects intrinsic to accelerator-based computing, in general, and to the MIC architecture, in particular. In this thesis we advocate the use of algorithmic skeletons for programming the Xeon Phi. Algorithmic skeletons abstract the complexity inherent to parallel programming, hiding details such as resource management, parallel decomposition, inter-execution flow communication, thus removing these concerns from the programmer’s mind. In this context, the goal of the thesis is to lay the foundations for the development of a simple but powerful and efficient skeleton framework for the programming of the Xeon Phi processor. For this purpose we build upon Marrow, an existing framework for the orchestration of OpenCLTM computations in multi-GPU and CPU environments. We extend Marrow to execute both OpenCL and C++ parallel computations on the Xeon Phi. We evaluate the newly developed framework, several well-known benchmarks, like Saxpy and N-Body, will be used to compare, not only its performance to the existing framework when executing on the co-processor, but also to assess the performance on the Xeon Phi versus a multi-GPU environment.

Incremental compilation and deployment for OutSystems Platform

OutSystems Platform is used to develop, deploy, and maintain enterprise web an mobile web applications. Applications are developed through a visual domain specific language, in an integrated development environment, and compiled to a standard stack of web technologies. In the platform’s core, there is a compiler and a deployment service that transform the visual model into a running web application. As applications grow, compilation and deployment times increase as well, impacting the developer’s productivity. In the previous model, a full application was the only compilation and deployment unit. When the developer published an application, even if he only changed a very small aspect of it, the application would be fully compiled and deployed. Our goal is to reduce compilation and deployment times for the most common use case, in which the developer performs small changes to an application before compiling and deploying it. We modified the OutSystems Platform to support a new incremental compilation and deployment model that reuses previous computations as much as possible in order to improve performance. In our approach, the full application is broken down into smaller compilation and deployment units, increasing what can be cached and reused. We also observed that this finer model would benefit from a parallel execution model. Hereby, we created a task driven Scheduler that executes compilation and deployment tasks in parallel. Our benchmarks show a substantial improvement of the compilation and deployment process times for the aforementioned development scenario.

Search-for-yield in Portuguese fixed-income mutual funds and monetary policy

This paper studies the effects of monetary policy on mutual fund risk taking using a sample of Portuguese fixed-income mutual funds in the 2000-2012 period. Firstly I estimate time-varying measures of risk exposure (betas) for the individual funds, for the benchmark portfolio, as well as for a representative equally-weighted portfolio, through 24-month rolling regressions of a two-factor model with two systematic risk factors: interest rate risk (TERM) and default risk (DEF). Next, in the second phase, using the estimated betas, I try to understand what portion of the risk exposure is in excess of the benchmark (active risk) and how it relates to monetary policy proxies (one-month rate, Taylor residual, real rate and first principal component of a cross-section of government yields and rates). Using this methodology, I provide empirical evidence that Portuguese fixed-income mutual funds respond to accommodative monetary policy by significantly increasing exposure, in excess of their benchmarks, to default risk rate and slightly to interest risk rate as well. I also find that the increase in funds’ risk exposure to gain a boost in return (search-for-yield) is more pronounced following the 2007-2009 global financial crisis, indicating that the current historic low interest rates may incentivize excessive risk taking. My results suggest that monetary policy affects the risk appetite of non-bank financial intermediaries.

Performance analysis of an SRI screen based on employment quality in high and low capital industries: International evidence

A Work Project, presented as part of the requirements for the Award of a Masters Degree in Finance from the NOVA – School of Business and Economics and Maastricht University School of Business and Economics

Management consulting labs- Portugal Telecom

The purpose of this work project was to analyze and evaluate the potential impact of a technological innovation in the telecommunications sector, across a wide range of business areas. A cost-benefit and competitive analysis for each pre-selected business area was conducted, as well as national and international benchmarks. As a result of the analysis, a list of prioritized business areas, presenting more immediate opportunities for Portugal Telecom, was created and implications for go-to-market strategies were inferred from the conclusions reached. In addition, a final recommendation that redefined the company’s positioning strategy was made

Growth hacking model for B2B saas startups

The following project introduces a model of Growth Hacking strategies for business-tobusiness Software-as-a-Service startups that was developed in collaboration with and applied to a Portuguese startup called Liquid. The work addresses digital marketing channels such as content marketing, email marketing, social marketing and selling. Further, the company’s product, pricing strategy, partnerships and website communication are examined. Applying best case practices, competitor benchmarks and interview insights from numerous industry influencers and experts, areas for improvement are deduced and procedures for each of those channels recommended.

Avaliação do desempenho de fundos de investimento de obrigações: evidência para o mercado Brasileiro

Dissertação de mestrado em Finanças

Hand gesture recognition system based in computer vision and machine learning

"Lecture notes in computational vision and biomechanics series, ISSN 2212-9391, vol. 19"

Generating timed trajectories foran autonomous robot

Tese de Doutoramento Programa Doutoral em Engenharia Electrónica e Computadores