Robotics in the browser

Drive, compare and program
more than 40 robots

From a two-wheeled balancer to the Unitree G1, the TurtleBot 4 and the xArm: every robot runs in your browser on the same obstacle course. See what each motor does, drive with the keyboard, then program it from Python or ROS 2, or let it learn by itself.

Tutorials

Step by step, from your first script to reinforcement learning and ROS 2. The ROS 2 tutorials run entirely in the browser; for the others, Getting started sets up the API on your own PC.

Robots

Open a robot and drive it with the buttons or W/A/S/D. Every simulator has the same obstacle course, so you can see who gets over what; every robot can also be programmed through the API.

Drive types One generic robot for every way to drive

Each simulator shows which motors there are, the PWM and direction every H-bridge gets, and how the robot moves as a result. Drive with the buttons or with W/A/S/D.

Legs and snakes Servos, inverse kinematics and balance

Legs instead of wheels, or both. The snake has neither: it moves only by changing its shape and letting friction push back.

Snake robot

4 to 12 Dynamixel XL430 in a chain

4–12 servospitch + yaw joints3 gaitsservo sliders
Open simulator →

Robot arms Joints, a gripper and a camera at the wrist

Arms on a table with blocks to pick and stack. The xArm and OpenMANIPULATOR-X can also be a digital twin of the real arm.

Robot dogs Real quadrupeds, driven with their real interface

At their real size, with the commands the real robot accepts: WebRTC or unitree_sdk2, the AgiBot SDK, xgolib or OpenCat serial tokens. Every button shows the line that does the same on the real robot.

Humanoids and bipeds Real two-legged robots

State machines, walking and gestures, through WebRTC, unitree_sdk2, the Booster SDK or the LimX WebSocket API.

Unitree H2

Full-size 1.82 m humanoid, programmed with unitree_sdk2's high-level API

31 jointsunitree_sdk2FSMarm actions

Booster K1

RoboCup kid-size humanoid with the Booster SDK: modes, dances and kicks

22 jointsBooster SDKRoboCupdanceskicks

Wheeled robots Real ROS 2 and education robots

Mobile robots with LiDAR, cameras and odometry. Most are driven the ROS 2 way, with a terminal on the page for the ros2 command line; the Jetson kits take Python.

Leo Rover

Outdoor ROS 2 rover with rocker suspension and four hub motors

ROS 2rocker suspension4 hub motorswide-angle cameraLiDAR add-on

JetBot

NVIDIA's Jetson Nano robot kit, driven with the jetbot Python library

Jetson NanoPython2 motorscamera

JetRacer

Jetson Nano RC car: NvidiaRacecar steering and throttle in Python

Jetson NanoPythoncar steeringcamera

Who gets over what?

Every robot drove straight into each obstacle of the course with its default settings. In the corridor every robot drives by itself: robots that move sideways slide round the corner, robots that turn on the spot turn there, car-like robots steer through the bend. Some do better after tuning, such as the humanoid legs with a higher step.

✓ gets over it✗ stuck: hover to see why▶ click a picture to watch the hardest obstacle it takes

Robot1. 1 cm cable2. 3 cm threshold3. 6 cm step4. 12 cm curb5. Stairs 3 × 7 cm6. 15° ramp7. 30° ramp8. Gaps 5 + 12 cm9. Gravel10. 60 cm corridor + turnScore
✓✓✗✗✗✓✗✓✓✓6/10
✓✓✗✗✗✓✗✓✓✗5/10
✓✗✗✗✗✓✗✗✓✓4/10
✓✗✗✗✗✓✗✗✗✓3/10
✓✗✗✗✗✓✗✗✗✓3/10
✓✗✗✗✗✓✗✗✓✓4/10
✓✓✓✗✓✓✓✓✓✓9/10
✓✓✗✗✗✓✓✗✓✓6/10
✓✗✗✗✗✓✗✗✗✓3/10
4WD
✓✗✗✗✗✓✓✗✓✓5/10
✓✓✓✗✓✓✓✓✓✓9/10
✓✗✗✗✗✗✗✗✗✓2/10
✓✓✓✓✓✓✓✗✓✓9/10
✓✓✗✗✗✓✓✗✓✓6/10
✓✓✓✓✓✓✓✓✓✗9/10
✓✓✓✓✓✓✓✓✓✓10/10
✓✗✗✗✗✓✗✓✓✓5/10
✗✗✗✗✗✗✗✓✗✓2/10
✓✗✗✗✗✗✗✓✗✓3/10
✓✗✗✗✗✓✗✓✓✓5/10
✓✓✓✓✓✓✓✗✓✓9/10
✓✓✓✗✓✓✗✓✓✓8/10
✓✗✗✗✗✗✗✗✗✓2/10
✓✓✓✓✓✓✓✓✓✗9/10
✓✓✓✗✓✓✗✓✓✓8/10
✓✗✗✗✗✓✗✗✓✓4/10
✓✓✓✗✓✓✓✓✓✓9/10
✓✗✗✗✗✓✗✗✓✓4/10
✓✗✗✗✗✓✗✗✓✗3/10
✓✗✗✗✗✓✗✗✗✓3/10
✓✓✗✗✗✓✗✗✓✓5/10
✗✗✗✗✗✓✗✗✗✓2/10
✓✗✗✗✗✓✗✗✗✓3/10
✓✗✗✗✗✓✗✗✓✓4/10
✓✓✓✓✓✓✓✓✓✗9/10
✓✓✓✓✓✓✓✓✓✗9/10
✓✓✓✓✓✓✓✓✓✓10/10
✓✓✓✓✓✓✗✓✓✓9/10
✓✓✗✗✗✓✗✓✓✓6/10
✓✓✓✓✓✓✓✓✓✓10/10
✓✓✗✗✗✓✓✓✓✓7/10
✓✓✓✓✓✓✓✓✓✗9/10
✓✓✓✗✓✓✓✓✓✗8/10
✓✓✗✗✗✗✗✓✓✗4/10
✓✓✗✗✗✓✗✓✓✗5/10