Modules 00–02
Build the foundations
The math, Python and Linux you need, then how real robots are put together: frames, kinematics, dynamics and control.
Kinemata · the open course in Physical AI
Give AI a body.
Go from zero to expert in Physical AI and humanoid robotics: robot fundamentals, ROS 2, simulation, perception, robot learning and vision-language-action models. Rivet, your AI tutor, has read every lesson.
- modules
- 11
- lessons
- 52
- weeks
- 16
Zero to expert, layer by layer
Eleven modules in the order a robot needs them: a body you understand, a nervous system, senses and movement, and finally the judgement to learn and act on what it is told.
Modules 03–04
Wire the nervous system
ROS 2 from first node to full robot description, then simulated worlds in Gazebo and Isaac Sim to practise safely.
Modules 05–06
See and move
Cameras, LiDAR, state estimation and SLAM, then planning, Nav2 navigation, MoveIt 2 manipulation and grasping.
Modules 07–10
Learn, walk and reason
Imitation and reinforcement learning, humanoid balance and locomotion, vision-language-action models, and the capstone.
Sixteen weeks, from sensor to speech
Follow it as a semester or dip in by topic. Each block opens with clear objectives and ends with something that runs.
Open week 1- Weeks 1–2
Start here + foundations
Set up your workstation; Python, linear algebra, probability and the Linux toolbelt.
- Weeks 3–4
Robot fundamentals
Frames, kinematics, dynamics and PID control on a simulated arm.
- Weeks 5–6
ROS 2
Nodes, topics, actions, TF2 and URDF: a whole robot that talks to itself.
- Week 7
Simulation
Gazebo worlds, simulated sensors, Isaac Sim, and the sim-to-real gap.
- Weeks 8–9
Perception
Images, point clouds, detection, Kalman filters and SLAM.
- Week 10
Motion
Path planning, Nav2, MoveIt 2 and grasping.
- Weeks 11–12
Robot learning
Imitation learning, reinforcement learning and training in simulation.
- Week 13
Humanoids
Balance, walking, whole-body control and dexterous hands.
- Week 14
Foundation models
LLM planners, vision-language grounding and VLA models.
- Weeks 15–16
Capstone & career
Ship a voice-commanded robot assistant and turn it into a portfolio.
Tools that read along
Built into every page, so you never leave the lesson to get unstuck.
Rivet, a tutor that cites the course
Rivet answers from the lessons themselves, knows which page you are on, and links back to the sections it used. Select any text in a lesson to ask about it.
Example
When should I use a service instead of a topic?
Rivet
Use a service for a request that needs one reply, like “reset the arm”. Use a topic for a stream, like camera frames or joint states.
1Nodes, topics and services with rclpyRead it in Urdu
Translate any page from the Tools menu. Technical terms stay intact.
ROS 2 روبوٹ کا اعصابی نظام ہے۔
“ROS 2 is the nervous system of a robot.”
Pitched at your level
Tell it your software and hardware background once; Personalize rewrites a page for you.
Your first lesson takes an evening.
Start with how the course works and set up your workstation. No robot needed.