RomaLab Robotics

RomaLab Robotics is a collaborative robotics research and development group focused on Physical AI, Robot Learning, and intelligent robotic systems.

Our goal is to develop robotic systems that can perceive, reason, learn, and safely interact with the physical world.

Research Areas

• Vision-Language-Action Models (VLA)
• Robot Learning
• Imitation Learning
• Physical AI
• Vision-Language Models (VLM)
• Large Language Models (LLM) for Robotics
• Robot Safety
• Control Barrier Functions (CBF)
• End-to-End Robotic Manipulation

Current Focus

We are currently exploring learning-based robotic manipulation systems, including:

• Vision-Language-Action policy learning
• Imitation learning for manipulation
• Safe execution of learned robot policies
• Multimodal robot perception
• Sim-to-Real and Real-to-Sim workflows
• Industrial robotic manipulation

Models

Trained and fine-tuned robot policies will be published here.

Datasets

Robot demonstration datasets used for imitation learning and VLA training will be published here.

Projects

Our projects include robotic manipulation tasks such as:

• Pick-and-Place
• Safe Human-Robot Interaction
• Long-Horizon Robotic Tasks

Models, datasets, benchmarks, and experimental results will be continuously updated.