Gabriel Dorneles
Automation Engineer
NAUTEC Research Group
Brief Bio
Automation Engineer and M.Sc. student in Computer Engineering at FURG, member of the NAUTEC Research Group, researching robotics and intelligent automation. My work focuses on building autonomous robots that perceive, reason, and act in real-world environments, spanning mobile manipulation, computer vision, and language-guided task planning. I develop solutions across R&D projects at iTec/FURG Embrapii and compete internationally with the FBOT Robotics Team in RoboCup@Home.
Education
Interests
- Physical AI
- Mobile Manipulation
- Digital Twins & Simulation
- Domestic Robotics
- Task Planning
Languages
- Portuguese Native
- English Fluent
- Japanese Conversational
Education
Projects
Ongoing
OpenRobiTec/FURG EmbrapiiA mobile manipulation robot prototype guided by language and vision, capable of planning and executing tasks in human-designed environments.
A mobile manipulation robot prototype guided by language and vision, capable of planning and executing tasks in human-designed environments.
- Developing an integrated system combining large language models with visual perception for task planning and execution in unstructured environments.
- Designing the full software stack for autonomous mobile manipulation, from high-level reasoning to low-level motor control.
iCraneiTec/FURG EmbrapiiAn R&D project developing an automated crane to assist operators on Petrobras' Floating Production Storage and Offloading (FPSO) units.
An R&D project developing an automated crane to assist operators on Petrobras' Floating Production Storage and Offloading (FPSO) units.
- Built a Digital Twin of the FPSO in NVIDIA Isaac Sim, integrating contributions from multiple teams into a high-fidelity simulation environment.
- Developed ROS2 packages for real-time 3D scene understanding using RGB-D cameras and LiDARs, enabling collision avoidance between crane payloads and surrounding obstacles.
- Implemented and optimized computer vision algorithms on embedded systems, doubling baseline frame rate performance on NVIDIA Jetson platforms.
FBOT@HomeFBOT Robotics TeamDevelopment of the domestic robots DoRIS and BORIS for the RoboCup@Home competition, aiming to set a benchmark for comparing General Purpose Service Robots in domestic environments.
Development of the domestic robots DoRIS and BORIS for the RoboCup@Home competition, aiming to set a benchmark for comparing General Purpose Service Robots in domestic environments.
- Worked across multiple areas of domestic robotics including ROS2 computer vision, robotic manipulation, 3D scene understanding, human-robot interaction, and general-purpose task planning and execution.
- Contributed to 6 published papers spanning robotic manipulation, LLM-based task planning, and computer vision as part of the FBOT Research division.
- Served as Program Committee Member for the RoboCup Symposium in 2025 and 2026, contributing to peer review of technical submissions.
- Mentored undergraduate students through their first research projects, several of whom have gone on to publish and present their own work.
- 1st place — Latin America Robotics Competition (LARC 2023)
- 1st place — Brazilian Robotics Competition (CBR 2024)
- 7th place — RoboCup Salvador 2025
- 1st place — Brazilian Robotics Competition (CBR 2025)
- A Review on Dataset Collection Strategies for Learning Methods in Robotic Manipulation
- AffordGen: Affordance-Based Dataset Generator for Robot Manipulation Learning
- Digital Environment Description and Reconstruction Using Panoptic Segmentation
- Review of Surface Reconstruction: From Classical Methods to Neural Radiance Fields
- SHARK: Stable hoverboard-driven autonomous robot kit
Past
Robot GardeneriTec/FURG EmbrapiiAn R&D project in partnership with STIHL, focused on developing an autonomous mobile platform to locate and eliminate lawn weeds.
An R&D project in partnership with STIHL, focused on developing an autonomous mobile platform to locate and eliminate lawn weeds.
- Designed the software architecture and custom ROS2 packages for an autonomous outdoor mobile robot integrating navigation, weed detection, targeting, and elimination.
- Developed solutions for identifying invasive plants through 2D image segmentation, transforming positional data into 3D world coordinates to guide a cartesian actuator.
- Led the robotics team with strategic planning, task delegation, and organized robot field tests to validate system performance in real environments.