During this internship, I migrated many projects, from AWD autonomous rovers to orbital servicing robot arms assembly. It was an opportunity for me to master Isaac Sim.

NVIDIA is putting a lot of effort into developing the future of robotics, starting with IsaacSim, a robotic simulator that embeds all NVIDIA technologies (RTX, DLSS, etc.), offers smooth ROS2 integration, and provides an API to easily generate synthetic data to train policies and DNNs. That makes Isaac Sim a great versatile tool for all roboticists.

During these 6 months, I learned Isaac Sim and also got in touch with NVIDIA developer teams through video calls, participation in the official Discord, and by supporting and contributing to open-source projects such as OmniLRS (a lunar simulator in Isaac Sim).

I also created documentation and internal training for the team to provide everything they needed to start using Isaac Sim themselves.

I think this internship was a great opportunity because I discovered many different robotics projects and developed very strong skills in Isaac Sim, which I personally believe will be the future of robotics.

If you want to see my report and my presentation slides or just ask question, send me a mail at contact@rodriguez-esteban.com

Rodriguez Esteban Robotic – Embedded Systems – Software Engineer
	E: contact@rodriguez-esteban.com

What is Tiqit ?

Tiqit is a solution providing a secure way to authenticate physical products by linking them to a digital twin through the blockchain. We used the blockchain solely as a decentralized and secure database, not as a marketing tool (which, in my opinion, is not an effective approach).

The tech stack

I was the technical leader of the product and built the entire ecosystem on my own:

• A dashboard and professional web application
• An end-user and professional mobile application (iOS and Android)
• A backend for the product
• A blockchain backend for the product

As for the frameworks and languages used, all apps were developed with Flutter, the backend was built with Firebase, and the blockchain was implemented using Solidity.

A double action

But it was more than just an authentication app, it also provided an experience for the end user. As a product seller, you could create your own product page using our page builder, offering digital content and a unique experience when the user brought their phone close to the chip embedded in the product.

The end of the story

Today, we have permanently paused the project, as everyone has different goals and objectives. We also struggled to identify our true target audience and couldn’t position ourselves effectively.

But it was a wonderful experience that gave me valuable soft skills and sharpened my programming skills.

Today, the app is still available and maintained, as we still have some clients using it, but no new features will be added in the future.

Access to tiqit.fr website

Rodriguez Esteban Robotic – Embedded Systems – Software Engineer
	E: contact@rodriguez-esteban.com

Context

I built this tool for the University of Canterbury in New Zealand, specifically for the Forestry Department, to study seed behavior—especially for Pinus radiata, which is an invasive species there.

The goal was also to document and open-source the project to provide an easy-to-build, low-cost hardware solution that any researcher at a university can adapt for their own purposes.

Main Features

• Raspberry Pi-based system with a main unit and worker units connected via a crossover bridge
• Image processing pipelines using two cameras with stereo vision
• Velocity computation algorithms that can be customized and improved
• CLI tools for administration, testing, and calibration
• Modular algorithm architecture making it easy to plug in and test new implementations

Seed Tracker is intended as a platform to:

• experiment with computer vision pipelines,
• analyze seed behavior during free fall,
• and contribute to open-source innovation in agriculture.

Other medias

Source

Source code (only the backend without documentation)

Report (french)

Rodriguez Esteban Robotic – Embedded Systems – Software Engineer
	E: contact@rodriguez-esteban.com

Persistence of Vision 2024

Embedded system project during my robotics specialization at Enseirb Matmeca. The goal of the project was to build a holographic fan on an ATmega328P using pure C. It involved a comprehensive understanding of real-time clocks and concurrency.

Here you can mine but as it is turning very fast it was complicated to get the full clock but you can see better exemple on the project main page bellow :)

Project main page

Source code

Robotic State-of-Art 2024

During our specialization semester, we had to carry out a project for a company. The project was split into two parts: the first part was to research the state of the art of the project and present existing solutions on the market. This study was condensed into a report.

The project involved aiming with an arm embedded on a mobile base, specifically, a Leo Rover. We were responsible for the integration and simulation parts. In this report, we explored possible solutions that could be used in this project to point at a target (in the air) with the arm.

Source

Report (french)

AI hornet detection 2024

It was a small school project/hackathon over four days where we had to develop a complete solution based on AI, image processing, or both. We were only given video samples from a real system—a laser mounted on a beehive with a camera. We trained our AI and also proposed some alternative solutions.

Source

Source code (french)

Report (french)

Graph algorithm 2023

This short project aimed to develop well-known navigation algorithms like Dijkstra on famous problems such as the Traveling Salesman Problem. Our goal was to implement these algorithms and test their performance on concrete problems like drone navigation.

Source code and report (french)

Operations research 2023

Another project involved implementing the SIMPLEX algorithm on Lot-Sizing problems. The project had a theoretical part, which consisted of writing the MILP model, followed by implementing it in Python and then comparing the performance of different models.

Source code and report (french)

Rodriguez Esteban Robotic – Embedded Systems – Software Engineer
	E: contact@rodriguez-esteban.com

Across the time – May 2022

Game made by Gaetan Poncelet and Esteban Rodriguez. Photography across the time was our inspiration and we're offering you a journey into the history of Gothenburg.

A little game made for a jam with very restrictive rule, it was kind a challenge to me. I was with just one friend for the music. I got to program and make graphism on my own. Very proud of the atmosphere we built even if the game is very... artistic.

Links

Itch.io

Source code

Therapy – April 2020

Welcome to our game called “Therapy“ ! We are a group of high schoolers in senior year trying to have fun developing a game all together in spite of the quarantine. This is our first time developing a game and participing to this kind of event and we are really proud of the outcome of it !

This was my first game jam, which I attended during COVID with my high school friends who didn't know how to program, so I was the only one responsible for the programming.

Links

Ludum Dare page

Rodriguez Esteban Robotic – Embedded Systems – Software Engineer
	E: contact@rodriguez-esteban.com

1. 🧠 Map-Based Texture Decoding Uses a Map Texture system where uniquely colored pixels define regions for mapping—ensuring accurate reuse and easy manipulation of sprite data.

2. 🧩 Layer/Skin Texture System Supports Layer Textures that define how mapped regions should render, enabling retexturing, color shifts, and dynamic skinning.

3. 🧱 Texture-to-Map Abstraction Separates sprite visual data (textures) from structural layout (mapping), perfect for character customization, equipment overlays, and skin systems.

4. 🎛️ Multiple Mapping Workflows Offers a variety of options to map your textures:

   • Shader-based (fastest, GPU-driven, low overhead)
   • Sprite Component (UI-based, supports multi-layering)
   • Animator Override Component (frame-based UV mapping)
   • Compute Sprite Component (GPU-powered for large textures)
   • Engine Class API (manual scripting control with fine-grained access)

5. 🎮 Runtime Modifiability Both shader and component properties can be updated in real time—allowing for in-game evolution of sprites, dynamic skin swaps, or procedural generation.

6. 🧪 Built-in Map Generator Easily generate Map Textures from existing sprites via a dedicated UI window, defining clean region data with unique colors.

7. 🔁 Multi-Layer Support With components like PAUMSprite, stack and blend multiple rendering layers for complex visual effects and animations.

Technical Stack:

• Unity C# Editor Scripting
• ScriptableObjects & Mesh API
• Custom Inspector UI
• GLSL Shaders for texture analysis and feature detection
• GPU computation using ComputeShader and GPU-based pixel parsing
• Unity Asset Database & Serialization

Impact: Now available on the Unity Asset Store and used by indie developers, Pixel Art UV Mapper enhances both performance and workflow simplicity. It’s particularly well-suited for procedurally generated content, and resource-constrained platforms.

Rodriguez Esteban Robotic – Embedded Systems – Software Engineer
	E: contact@rodriguez-esteban.com

Sprite Engine is a little script interface for Unity which allows you to animate basic sprite easier and faster than with the unity animator. It's very powerful for UI basic sprite animation. You can also create event and customize each sprite duration. The package also provides a class to handle basic animation by scripting if you don't want to use the component.

A lot of features are available :

- Set custom time ration for each frame

- Custom events can be set at any frame, you can subscribe to them easily

- Start and End event

Links

Asset Store

Rodriguez Esteban Robotic – Embedded Systems – Software Engineer
	E: contact@rodriguez-esteban.com