Lucas Oliveira Maggi

A strategically minded and exceptionally skilled software engineer with 10+ years of experience in software development. Self-motivated learner and hard-working professional that can work in a team on any position. Passionate for problem-solving and high-performance solutions.

lucas.maggi@usp.br

Currently

Masters Student in Mechanical Engineering (Dynamics and Mechatronics)

University of São Paulo – USP,

São Carlos School of Engineering (EESC),

Brazil.

2022 - Masters Student in Mechanical Engineering (Dynamics and Mechatronics)

Actual status: “Working on something cool with legged and arm robots”.

Education

B.Sc. in Computer Engineering

Universidade Federal de Pernambuco – UFPE,

Centro de Informática (CIn),

Brazil.

2014 - 2021 - B.Sc. in Computer Engineering

Undergraduate final project: “MaggicVision: A Fast and Lighting-invariant Computer Vision System for IEEE Very Small Size Soccer”. Read here

2020 - present

Research Assistant

FedLea project in partnership with HP Inc.

Literature review on Federated Learning, prepared and executed simulations, implemented core and miscellaneous project codes/scripts, configured devops pipeline for deployment, executed continuous code optimizations.

Working with Python for backend, Python and JavaScript for frontend, Docker for containerizing.

2019

Research Assistant (7 months)

+ Internship (5 months)


Robotic deep reinforcement learning

- BOTS project in partnership with HP Inc.

Tackle the reality gap with robotic simulation and implementation in a physical robot.

Configured dockerfiles and scripts to run simulation and on physical robot. Robot modeling, and integration of simulation with a reinforcement learning framework. Implementation of the robot’s drivers for ROS2 based on ROS1 drivers. Implemented vision abstraction for virtual and real environment target detection. Integrated vision, inference network, and control modules.

Open-Source 3D robotic simulator for education – Paper (2019)

Implemented the 3D simulator with three.js library for model loading and rendering. Integrated a C++ interpreter in JavaScript. Created a graphic interface that shows the simulation on the left, and control code development on the right.

Link to Paper Link to Simulator

2015 - 2018

Undergraduate Researcher


NonFlatAR project in partnership with HP Inc.

– Read Curved Visual Marks Patent (2018)

New techniques for non-flat surface tracking in computer vision/augmented reality.

Computer vision project in partnership with Samsung Co. Ltd.

– Patch PlaNet Paper (2018)

Improved photo geolocalization by combining ranking classifier with deep learning patch classification.

Link to Paper

pARagraph 2017

pARagraph text-based tracking technology

Mixed reality TV controlled by body gestures

– Mixed reality TV Paper/Presentation (2017)

Implemented gestures detection, sample blocking-n-block game with hands using fishtank parallax effect.

ARGo - A Simple AR Framework

Voxar Corpora - cases

Voxar Dynamics

Lens Distortion at high framerates on CPU

– Paper/Presentation (2016)

Studies in code optimization and code parallelization in CPU. Used lens distortion effect as a case of study.

2017 - present

Undergraduate Researcher Volunteer

Computer Vision

MaggicSegmentation (undergraduate final project), a fast and lighting-invariant segmentation technique inside MaggicVision (also in undergraduate final project).

Computer Graphics

Graphics – Bezier curves to build attacker path planning, visualization of color segmentation on HSV color space using the MaggicSegmentation.

Parallel Processing

Code optimization and parallel processing optimization in vision pipeline.

Network

Network Communication between vision and control processes.

Mobile Robotics


Implemented autonomous navigation and mapping using TIAGo’s robotic platform – IEEE @Home Competition.