GenerativeMPC
VLM-RAG-guided whole-body MPC with virtual impedance for bimanual mobile manipulation.
MPCVLM-RAGImpedance
Welcome
Marcelino Fernando
Robotics software engineer working on Physical AI — bringing robot foundation models and vision-language-action systems from simulation to safe, contact-rich deployment on bimanual and humanoid robots.
My work centers on vision-language-action systems, humanoid and bimanual manipulation, and the compliant control that lets them act safely on real hardware. I build the learning, perception, and control stacks that turn language- and vision-derived intent into safe, contact-rich physical interaction.
News
Jun 2026
MilestoneCompleted my Master's degree with honours at Skoltech.
Jun 2026
PublicationGenerativeMPC accepted at IEEE SMC 2026.
Jun 2026
PublicationHapticVLA accepted at IEEE/RSJ IROS 2026.
Jan 2026
PublicationSafeHumanoid published at ACM/IEEE HRI 2026.
Oct 2025
AwardBest Industrial Immersion Project Award at Skoltech.
Oct 2025
ExperienceCompleted research internship at MTS AI on real-time Cartesian MPC for Unitree H1.
Featured Projects
HapticVLA
Tactile-aware vision-language-action learning for contact-rich manipulation, no tactile sensing at inference.
VLAIsaac SimTactile
SafeHumanoid
VLM-RAG-driven impedance control for humanoid robots, ensuring safe physical interaction.
HumanoidImpedanceSafety
Get in touch
Open to collaborations and roles in robotics control, perception, and Physical AI. The fastest way to reach me is email.