02 / INTERACTReach beyond
Reach beyond
movement.
Program the 4-DoF arm alongside the legs and wheels. Turn a motion sequence into a hands-on experiment.
XGO-mini2SW | AI Wheeled-Legged Robot Dog with LiDAR, Powered by Raspberry Pi CM5
$1,124.25
Regular price
$1,499.00
Sale price
$1,124.25
Skip to product information
XGO-mini2SW | AI Wheeled-Legged Robot Dog with LiDAR, Powered by Raspberry Pi CM5
$1,124.25
Regular price
$1,499.00
Sale price
$1,124.25
Now in Pre-sale — Ships within 45 Days
Due to the complexities of international shipping, returns are not accepted once your order has shipped. If you experience a product quality issue, we will arrange a replacement shipment or an exchange.
Learn, code, and build with XGO-mini2SW, a wheeled-legged robot dog equipped with LiDAR and designed for AI robotics exploration. Powered by Raspberry Pi CM5, it brings together reinforcement learning, multimodal AI, and open-source development in one versatile platform. Explore AI-assisted coding, develop your own applications, and bring your robotics ideas to life.
SMALL FOOTPRINT. BIG AMBITIONS.
Your ideas deserve
to get up and move.
Go from lines of code to movement you can see.
XGO-mini2SW with LiDAR brings mobility, manipulation, and AI together.
12 joints
Articulated leg control
4 wheels
Powered rolling motion
4 DoF
Integrated robotic arm
CM5
Raspberry Pi · 4GB RAM
01 / MOVEBuilt to roll.
Built to roll.
Ready to step.
Wheels bring efficient movement on flat surfaces. Articulated legs bring flexibility in posture and motion. See it in motion across four everyday settings.
Meet the hardware ↗
03 / ENGINEERA closer look
A closer look
at every turn.
Integrated BLDC wheel modules with FOC control put powered mobility at the end of each leg.
OPEN REINFORCEMENT LEARNING
Explore how
movement is learned.
Give your curiosity a training ground. With luwu_mjlab, you can develop and evaluate motion policies in GPU-accelerated MuJoCo simulation.
Explore the open toolkit01 / WALKING
Find your stride.
Explore flat-ground walking with a Mini2SW-specific task. Tune rewards for velocity tracking, gait, and smooth actions.
Walking simulation
02 / RECOVERY
Learn to get back up.
Investigate recovery from varied starting orientations. The Mini2SW recovery task lets you experiment with how a robot returns to an upright posture.
Recovery simulation
Simulation examples from the official Luwu toolkit. Collapse a demo to hide its animation.
Train
Run parallel simulation environments and learn motion policies with PPO.
Evaluate
Review training progress in TensorBoard and replay saved checkpoints.
Export
Export policies to ONNX with metadata for further integration and hardware development.
MORE THAN MOVEMENT
Say hello
to possibility.
“Hi Luka.” Two words.
A new way to interact.
Ask about what’s in front of the camera. Trigger supported actions with your voice. Customize its personality, voice, and memory through Luwu-OS.
Cloud AI requires internet access and a compatible API key. Provider charges may apply. Features vary with software and service configuration.
FROM FIRST SCRIPT TO NEXT EXPERIMENT
Made for the
“what if?” in you.
Start with an idea. Build a behavior.
Keep going with an open development platform.
BlocklyPythonC++ROS 2
THE DETAILS THAT MATTER
Small robot.
Serious hardware.
A compact platform with room for ambitious projects.
280 × 150 × 180 mm Base platform reference
| Main computer | Raspberry Pi CM5 · 4GB RAM |
|---|---|
| Motion controller | ESP32 |
| Actuators | 20 total 12 leg servos + 4 arm servos + 4 powered wheels |
| Robotic arm | 4 degrees of freedom |
| Wheel drive | 8.4V BLDC motors · FOC control |
| Leg servos | 6V · 4.5 kg·cm Steel gears, magnetic encoders, TTL serial control |
| Frame | 1.5mm aluminum alloy |
| Weight | 1,150g |
| Battery | 18650 · 2S · 3,800mAh |
| Operating time | Approximately 2 hours* |
| Charging port | DC barrel connector |
*Operating time varies with use. Dimensions and weight are base-platform references. Confirm final LiDAR-equipped dimensions, weight, and shipping configuration when ordering.
Who is XGO-mini2SW with LiDAR for?
It is a platform for makers, educators, and developers exploring robotics. Graphical programming offers a starting point, while ROS 2 and reinforcement learning suit users ready for a more technical setup.
Can I talk to it in English?
Yes. The Luwu-OS documentation includes English interaction with the wake phrase “Hi Luka.” Cloud conversation requires a compatible API key and internet access.
Does it train AI models onboard?
The toolkit trains motion policies in a compatible NVIDIA GPU environment, including the documented Colab route. Evaluate and export your policy before adapting it to the physical robot.
Is LiDAR included?
This page features the XGO-mini2SW with LiDAR configuration, shown with a top-mounted LiDAR unit. Confirm the LiDAR model and complete package contents when ordering.
What can I train with the open toolkit?
The repository includes dedicated Mini2SW tasks for flat-ground walking and recovery. Use them as starting points to study learned motion and develop your own experiments.
Can I customize the training?
Yes. The task configurations expose rewards and observations, while training settings let you adjust your experiments. Save checkpoints, compare runs, and export policies to ONNX.
Where do I find setup instructions?
The official knowledge base covers startup, network setup, programming, firmware maintenance, and advanced development. Use the links on this page to explore the documentation.
THE NEXT STEP STARTS HERE
Your guide to
getting started.
Setup guides, programming tutorials, and hardware references.
Everything you need for your next experiment.
Set up. Explore. Start building.