Upfreq Robotics
Describe the robot's behaviour. Agents write the ROS 2 code and prove it in simulation.
- What it is
- AI for robot software
- My role
- Web + AI infrastructure partner
- Status
- Live
- Built with
- ROS 2, NVIDIA Isaac Sim, Python, Claude Code plugins, Digital twins
Overview
Upfreq Studio lets robotics engineers describe behaviour in natural language while AI agents read the robot's codebase, generate ROS 2 software, run tests and verify changes in live simulation. Its testing engineer, ARTE, catches navigation, perception and control failures in CI before code reaches real hardware. Through Vantaverse I built Upfreq's web presence and agent infrastructure, and open-sourced ros2-isaac-agent-suite, a Claude Code plugin for an AI-driven ROS 2 and Isaac Sim test loop.
What it does
- Natural language to ROS 2 code, verified in simulation
- URDF/Xacro import for arms, drones, humanoids and autonomous vehicles
- ros2-isaac-agent-suite: discover a codebase, run seeded Isaac Sim scenarios, trace failures to source
