iLearnSTEM Humanoid Robotics

Humanoids, from the controller up.

A verified, simulation-first curriculum for humanoid robotics — every claim is executed before it's taught. Derive the physics, run it in your browser, tune it until it holds.

Start the Control Systems trackBrowse the catalog

live inverted pendulum under PID — shove the bob with your cursor. holding it up is the CS-1 capstone.

tracks
5
tracks
lessons authored
81
lessons authored
runnable lessons machine-verified
81
runnable lessons machine-verified
open for enrollment
2
open for enrollment

How it's built.

θ̈ = (g/L)·sinθ + u

Physics first

Every controller starts from the differential equation it tames. You derive the model, predict its behavior, then watch the prediction hold — steady states, time constants, margins.

import numpy as np

Simulation you run

Every concept executes in your browser — NumPy in Pyodide, no installs. You change the numbers, break the loop, and tune it back. Not videos of simulations: the simulations.

81 runnable lessons · 81 PASS

Machine-verified content

Every runnable block is executed by a harness before it publishes, and its printed output is asserted against the prose that quotes it. If a lesson claims 16.3% overshoot, a machine checked that it prints 16.3%.

Every concept runs.

These aren't screenshots — they're the lesson widgets themselves, running now.

loading simulation…

Drag Kp and watch it ring. Gravity is the opponent; the I-term is the closer.

CS-1 Classical Control · m2 PID in anger →
loading simulation…

Deterministic isn't predictable — the passive dynamics a balancing humanoid has to tame.

CS-3 Humanoid Balance & Locomotion · m1 ZMP & LIPM →
loading simulation…

A belief with error bars, updated live. Estimation is where CS-2 meets the sensors.

CS-2 State-Space & Optimal Control · m4 Estimation →

Real code, really graded.

Python (NumPy) executes in your browser for the control courses. C++ compiles on our servers and comes back with GCC's honest opinion.

Python · runs in your browser (Pyodide)

diff_drive.pyRuns in your browser
import math

# Two wheel speeds in, one motion out — the diff-drive forward map.
b, r = 0.3, 0.05          # wheelbase, wheel radius (m)
wl, wr = 8.0, 10.0        # wheel speeds (rad/s) — try changing these

v = r * (wl + wr) / 2     # forward speed
w = r * (wr - wl) / b     # turn rate
print(f"v = {v:.2f} m/s, \u03c9 = {w:.2f} rad/s")
print(f"turn radius R = v/\u03c9 = {v / w:.2f} m")

Press Run — this executes right here, no account needed. Every Python block in every lesson is editable and re-runnable, just like this one.

C++ · compiled on our servers (GCC)

clamp_cmd.cppCompiled on the server
#include <algorithm>
#include <iostream>

int main() {
    double v, v_max = 0.25;            // m/s — the robot's contract
    while (std::cin >> v)
        std::cout << std::clamp(v, -v_max, v_max) << '\n';
}

✓ Accepted — 3/3 test cases (1 hidden)

g++ (GCC 14.1.0) · cpu 0.002 s · 1.1 MB

A replay of a real graded run. Inside the course, C++ blocks compile on our Judge0 servers — graded exercises included, with hidden test cases like a real interview.

Five tracks. One robot.

Each track is a guided study plan ending in a certificate. Cards show the true state — derived from the plan data, updated the moment a track opens.

  1. 01

    Mechanics for Humanoids

    Planned

    Kinematics, dynamics, actuators, and the structural design of legged and armed humanoids — the physical plant every controller acts on.

  2. 02

    Control Systems for Humanoids

    Enroll now

    From PID on a single joint to whole-body balance and safe physical human–robot interaction: classical control, state-space and optimal control, humanoid locomotion, and compliance control — every controller written and tuned by you.

    4 courses · 57 lessons · every runnable machine-verified

    1. Phase 1 — Foundations of control
    2. Phase 2 — Humanoid control
    3. Phase 3 — Capstone
      Start this track
    4. 03

      Electronics-lite for Humanoids

      Planned

      Just enough electronics to read a humanoid: power and battery systems, motor drivers, sensor front-ends and buses — building on the fundamentals taught on electronics.ilearnstem.ca.

      Prerequisite: Electronics fundamentals on electronics.ilearnstem.ca — taught there, not duplicated here.

    5. 04

      AI & Reinforcement Learning for Humanoids

      Planned

      Learned locomotion and manipulation: from imitation and RL fundamentals to sim-to-real policies for legged humanoids — and the judgment to know when a model beats a policy.

    6. 05

      Wearable Robotics

      Enroll now

      Exoskeletons and assistive wearables: apply humanoid control to a device worn by a person — biomechanics of assistance, exoskeleton design and control, and a capstone device, design-track or build-track.

      2 courses · 24 lessons · every runnable machine-verified

      1. Phase 1 — Biomechanics
      2. Phase 2 — Design & control
      3. Phase 3 — Capstone
        Start this track

        Gate: earn the Control Systems for Humanoids certificate first.

      The pendulum is waiting.

      Start the Control Systems trackBrowse the catalog