From Sim to Real: Train, Deploy, Iterate
Our open-source Sim-to-Real pipeline simplifies reinforcement learning deployment, allowing developers to move from simulation to real-world robots with minimal setup and tuning.
Reinforcement Learning
Wheeled-Legged Robot
High-Performance Wheel-Leg Experimental Platform
Integrates wheel-legged mobility, reinforcement learning, multimodal AI, and open-source development.
AI learns and adapts in real time, moving beyond hard-coded movement.
Smooth wheels plus agile 12-DoF legs overcome any terrain.
Raspberry Pi CM5 with Blockly, Python, C++, and ROS.
See, listen, understand, and act in the real world.
XGO-mini2SW Series is more than a hardware upgrade — it represents a shift from traditional hard-coded movements to AI-driven autonomous learning.
Our open-source Sim-to-Real pipeline simplifies reinforcement learning deployment, allowing developers to move from simulation to real-world robots with minimal setup and tuning.
Powered by reinforcement learning, XGO can adapt to changing environments, maintain balance, and recover from unexpected disturbances in real time.
Join a growing robotics community. Explore advanced toolchains, build, train, and share custom RL skills through open-source collaboration.
Describe the behavior you imagine. AI understands XGO-mini2SW’s real codebase, writes and debugs the software, and deploys it directly to the robot—turning intent into motion.
Describe your idea.
AI builds and deploys.
See it move for real.
Exceptional Mobility
Enjoy the speed and efficiency of wheels on smooth surfaces while retaining the agility of 12-DoF legs to overcome obstacles, climb slopes, and navigate complex terrain.
A compact all-in-one hub-wheel module integrates the motor, drivetrain, and wheel into a single efficient unit.
Switch between wheeled and legged configurations in minutes with a simple screw-based design.

No waiting. No awkward pauses. XGO-mini2SW listens, understands, and responds in near real time.

Powered by an HD camera and vision-language models, it can recognize objects, understand scenes, and describe what it sees in real time.
From simple voice commands to real-world tasks, XGO-mini2SW can autonomously navigate, interact with objects, and execute actions using its mobility system and 4-DoF robotic arm.
Create a personalized AI companion with expressive animations, unique voice styles, and long-term memory of your preferences and past conversations.
Powered by Raspberry Pi CM5, XGO-mini2SW provides a fully open-source platform with support for Blockly, Python, C++, and ROS. Build, customize, and deploy your own robotics and AI projects with ease.