Eduardo Araújo
MSc in Mechanical Engineering with specialization in Intelligent Control and Automation of Mechanical Systems.
My master’s thesis focused on model predictive control techniques applied to supply chains.
I am passionate about control systems and applying artificial intelligence to real-world problems.
Lisbon, PT • Paris, FR

Control Systems, Multi-Agent Systems, AI

              

Projects

October 2019 – May 2021	Robot Dynamic Scheduling And Allocation To achieve a better resource management as well as improve robot efficiency, a dynamic scheduling and allocation system was developed. Accomplishments: Robots' productivity substantially increased, with almost zero time waste The same number of robots were able to perform more, and more diverse, processes The achieved solution required no financial investment
June 2020	Time tracking application To acquire a better understanding about the team operations, a time tracking application was developed. Accomplishments: The application was developed using windows forms and written in C# It became clear what were the operations team members spent more time in, which improves planning and a more transparent relationship with clients.
January – March 2020	Unit and Integration Testing Framework Since the UiPath products lack in testing tools, a private framework was developed. The solution embodies the Agile Development paradigm and enables Continuous Integration. Accomplishments: Tests become auditable, repeatable and comprehensive The solution may be used in both Test-Driven and Behavioural-Driven Development paradigms The achieved solution may work as a diagnostics tool in any environment, making clear to the developer where to act during bug-fixing Using a unified framework all developers work in a standard way, which decreases allocated time to maintenance activities
May – August 2019	ESoWC 2019 – Data-driven feature selection towards improving forecast-based prediction of wildfire hazard. This project proposes the application of machine learning techniques for the development of prediction models of extreme fire events, based on the knowledge discovery in databases (KDD) framework. This work aims to devise data-driven feature selection approaches adequate for the calibration of the Canadian Forest Fire Weather Index system, to provide a deeper understanding of the datasets open-sourced by ECMWF / Copernicus.     In collaboration with: Maria Sousa.
May 2019	Failure prediction of a water pump system (Phase I). The devised methodologies herein presented focus on data exploration. Two important results are achieved: (1) from a set of 52 sensors, the 8 most relevant measurements are selected; (2) a set of clean, ready to work data sets, which will be used throughout the Phase II are constructed.
April – May 2018	An autonomous mizusumachi system. A multi-agent, autonomous mizusumashi system for an industrial plant comprising three different assembly-lines. In this work, each robot, or agent, must transport materials from downstream processes to upstream ones. Both the demand and availability of resources are stochastic. The proposed project has three major goals: 1) to evaluate the impact of the number of Robots/Agents on the overall system performance; 2) to compare and assess which model of agency (Cooperative or Free-Agent) produces better results; 3) to determine which Ratio of Personalities should one rely on in order to improve the overall system performance.
Oct 2018	MATLAB -- A Multi-Agent Framework.
May 2018	Random wealth game.
Dec 2017	Acquisition and processing of critical psycho-physiological and flight parameters in light-sport aviation This work focuses on developing a predictive system which is able to model the relationship between the pilot’s performance, the aircraft control and the flight safety. The data available to derive the model from is composed of several psycho-physiological and aircraft variables acquired in simulated flight environment – hypobaric chamber. The presented approach makes use of artificial neural networks with sequential feature selection and was implemented resorting to the MATLAB Neural Network Toolbox.
Nov 2017	Regression of a Stock Market Dataset This work presents a fuzzy modelling approach to model stock markets. The dataset used was downloaded from the KEEL Dataset Repository and concerns daily stock prices forten aerospace companies. The task was to approximate the price of the 10th company given the prices of the rest.
Dec 2016	Volume estimation from infrared sensor point-cloud – Computer Vision Estimation of volume of objects using a Microsoft Kinect. RGB component is responsible for localization of the object to be measured, as well as for superficial area determination. The depth camera provides information about the object's height and width. Results have shown the proposed methodology is appropriate for volume estimation, since it presents an error no greater than 1/10 mm regardless of the objects dimensions.     In collaboration with: Maria Sousa, Ricardo Maia.
Nov 2016	Fire Detection based on Thermal Imaging sensor for campsite areas Thermal imaging analysis of FLIR Vue Pro thermal camera. Firstly, color segmentation techniques are used to determine the preponderance of the percentage of red color to throw alerts of possible fire outbreaks. A second phase focus on identifying paths to characterize the campsite, giving information about possible evacuation and/or firefighting paths. The developed algorithm was able to detect both fires and paths with a 92% accuracy.     In collaboration with: Maria Sousa, Ricardo Maia.
March – June 2016	Quad-rotor autonomous in-door navigation This work consisted on turning a commercial quad-rotor into a fully autonomous one, capable of moving to desired in-door locations, i.e. without making use of the GPS system.     In collaboration with: Ricardo Maia.