Marcelino Fernando/About
CV Email

About

Biography

I am a robotics software engineer working on Physical AI — bringing robot foundation models into the physical world. My focus is the gap between what vision-language-action (VLA) models can reason about and what robots can actually do, safely and reliably, in contact-rich tasks on bimanual and humanoid platforms.

I connect high-level reasoning to low-level compliant control, turning language- and vision-derived intent into safe physical interaction with people and objects. This covers the full sim-to-real stack: synthetic-data and digital-twin pipelines, tactile-aware VLA learning, and foundation-model-guided whole-body control. Recent work includes HapticVLA, contact-rich manipulation without tactile sensors at inference; GenerativeMPC, VLM-RAG-guided whole-body MPC for bimanual mobile manipulation; and SafeHumanoid, retrieval-augmented impedance control for humanoids.

Looking ahead, I am drawn to world models as the next layer — moving from hand-built digital twins toward learned models of physics that can both generate training data at scale and give policies the foresight that reactive VLAs lack.

Research Interests

Bimanual robotic manipulation Whole-body model predictive control Vision-Language-Action models Reinforcement learning World models Sim-to-real transfer Humanoid robotics Contact-rich manipulation Impedance control

Experience

  • Sep 2024 — Present

    Graduate Researcher

    Intelligent Space Robotics Laboratory (ISR Lab) · Moscow

    Whole-body MPC, vision-language-action models, and impedance control for bimanual and humanoid manipulation, bridging semantic reasoning and compliant physical interaction.

  • Sep 2025 — Mar 2026 · Moscow

    Engineer Intern

    Sber Robotics Center — R&D Department

    Vision-Language-Action foundation models: large-scale manipulation data engineering, teleoperation-based demonstration collection, and policy evaluation for bimanual and humanoid robotic systems.

  • Jun 2025 — Aug 2025

    Research Intern

    MTS AI Research Division · Moscow

    Developed real-time Cartesian MPC for the Unitree H1 humanoid for bimanual manipulation, achieving control latency under 10 ms with hardware-in-the-loop testing and sim-to-real calibration. Recognized with the Best Industrial Immersion Project award.

Awards & Honours

  • Jun 2026Certificate of Achievement for Academic Excellence — Skoltech.
  • Feb 2026Academic Excellence Scholarship — Skoltech Scholarship Council (also Apr 2025 – Jan 2026).
  • Feb 2026Scholarship for Top Paper Development — Skoltech.
  • Oct 2025Best Industrial Immersion Project 2025 — Skoltech Industrial Studies Department.
  • Mar 2023Best Conference Paper Award — 29th International Scientific Conference.
  • Jul 2022Diploma of Merit — Embassy of Angola, for Outstanding Academic Performance.

Personal Interests

Outside robotics, I enjoy coffee, reading, photography, sports, and attending robotics or technology events.

Coffee
Reading
Photography
Sports
Robotics Events