Shared Control for a Robotic Wheelchair with Two Arms
Control a dual-arm robotic wheelchair using a standard joystick. Demonstrates intuitive pick-and-place tasks while optimizing ergonomics and task efficiency.
Teleoperating Spot with Arm using Hand Tracking
Demo of a Boston Dynamics Spot with a Kinova Gen3 arm and Inspire-Robots hand, controlled entirely through a Leap Motion sensor. The interface maps both human hands to the robot, enabling smooth demonstrations of complex tasks with 25 DoF.
First VS Third-Person Perspective for Imitation Learning
Compare first- and third-person teleoperation interfaces for training a robot on bimanual tasks (shoelace insertion).
2024
3D Eye-Gaze Tracking for Assessing Trust in spatial HRI
Showcases a 3D eye-gaze tracking system to analyze human trust in robots. Demonstrates key metrics linked to trust during interactive inspection tasks.
Augmented-Reality Multi-User Interface for spatial HRI
Demonstrates an AR interface for multi-user human-robot interaction. Highlights trust calibration across different interaction scenarios with a quadruped robot.
2023
Florespiña – Tribute to Ana Kiro
I took part in the AI Song Contest 2023 as a member of the Spanish team PAMP. This music and multimedia project celebrates Galician singer Ana Kiro, highlighting her feminism and resilience. The song was the public favourite and won the Audience Trophy. The video portrays her career metaphorically through the thorny flower (florespiña) and combines traditional music with AI-driven production, including artificial voice models and generative visuals.
2022
Privacy-preserving learning for Smart Wheelchair Mobility
Shows a robotic wheelchair learning from distributed demonstrations while protecting user data. Highlights autonomous navigation and shared control in real environments.
Explaining my PhD thesis in 3 Minutes – 3MT – USC (Galician)
This short outreach talk (in Galician) was my winning presentation at the University of Santiago de Compostela’s Three Minute Thesis (3TM) competition.
2021
AI Agent for Smartphone Racing Game (Traffic Rider)
A casual project made just for fun. I trained an AI agent to play the mobile racing game Traffic Rider by imitating real players. The agent learns directly from vision-only inputs (game frames), achieving high speeds and high scores while avoiding crashes.
