Browser-Based Robotics

Robotics Simulator
Learning Path

An interactive, progressive introduction to robotics — from joint angles to full pick-and-place automation. No installation required. Runs entirely in your browser.

Learning Path — 6 Levels
01 — 02 Available
Robotic Arm — Kinematics
A 3-DOF planar arm. Drive it with joint sliders (FK) or click a target and watch the arm solve its own angles (IK).
  • Forward kinematics: joint angles → position
  • Inverse kinematics: position → joint angles
  • Workspace limits & singularities
  • Elbow-up vs. elbow-down solutions
  • Coordinate frames at each joint
Launch →
03 Available
Pick & Place
Objects on a table. The arm plans and executes a full pick-up sequence — approach, grasp, lift, transport, release.
  • Task decomposition into motion phases
  • State machine design (IDLE→GRIP→LIFT…)
  • Pre-grasp positioning & orientation
  • Gripper control
Launch →
04 Available
Mobile Robot Navigation
A differential-drive robot in a 2D environment. Click to set a goal — watch it plan a path and drive there, avoiding obstacles.
  • Differential-drive kinematics
  • Occupancy grid maps
  • A* / Dijkstra path planning
  • PID velocity controller
Launch →
05 Available
SLAM
Simultaneous Localization and Mapping. The robot starts with no map — it builds one in real time using simulated lidar while navigating.
  • Lidar sensor simulation
  • Occupancy grid mapping
  • Frontier-based exploration
  • Mapping with known poses
Launch →
06 Available
Mobile Manipulation
A mobile base with a robotic arm — navigate to a shelf, pick an item, deliver it. Combines every prior level into one coherent system. This is what Amazon, Figure, and 1X are building today.
  • Whole-body coordination (base + arm)
  • Task and motion planning
  • Real-world system integration
  • Industry context: warehouse robots
Launch →
Beyond the Learning Path
Lab Available
MuJoCo Lab — Real Physics
A Franka Emika Panda running physics-accurate simulation in the browser via WebAssembly. Point a Gemini vision model at the scene and watch it identify objects and direct the arm to pick them up.
  • MuJoCo physics engine via WebAssembly
  • Real robot model (Franka Panda) from MuJoCo Menagerie
  • Gemini vision layer for object detection
  • What production robotics tooling actually looks like
Launch →
Requires internet connection to load MuJoCo WASM from CDN. Gemini features require a Google AI API key.
Professional Tools

Where you'd go to work on real robots. These platforms are used in industry and research — all require local installation, and some require serious GPU hardware.

Isaac Sim · NVIDIA
GPU-accelerated platform built on Omniverse. Photorealistic sensor simulation, RL training via Isaac Lab, ROS 2 integration, and digital twin pipelines.
RTX 4080+ · Ubuntu 22.04 or Windows · Apache 2.0
Gazebo · Open Robotics
The standard open-source robotics simulator. Deep ROS 2 integration, large ecosystem of robot models, and the default testbed for most academic research.
Linux recommended · Free & open source
MuJoCo · DeepMind / Google
Physics-accurate simulation with exceptional contact modeling. The engine behind the Lab demo here — used heavily in robotics research and RL. Python and C APIs.
Cross-platform · Free & open source (Apache 2.0)
PyBullet · Erwin Coumans
Lightweight Python physics sim. Lower fidelity than MuJoCo but easy to set up, well-documented, and a common first step for RL experimentation on real hardware.
Cross-platform · pip install pybullet · zlib license