Skip to content
buildrobots.io

Path 2 of 3 · Q · 4 steps

Quadruped

Four legs and a robot that can fall. Actuators, gaits, state estimation, then sim-to-real walking.

  1. 01

    Start with a servo quadruped

    Servo calibration, gait tables, a serial skill protocol.

    Done whenYour robot walks, turns and stands up from a script you edited.

    Read / watch / run

    • Petoi documentation

      Petoi · Docs · free · docs.petoi.com

      Assemble, calibrate and program the Bittle and Nybble servo quadrupeds (OpenCat firmware).

    • OpenCat firmware

      Petoi · Repo · free · github.com/PetoiCamp/OpenCat

      Open quadruped firmware for Arduino and ESP32 boards: gaits, skills, serial protocol.

    • Pupper independent study

      Stanford · Course · free · pupper-independent-study.readthedocs.io/en/latest

      A self-paced course: build Stanford Pupper, then learn kinematics and gaits on it.

    Practise on · computed

    Buildable robots of this class, easiest first by the difficulty index; robots with a build package listed first.

  2. 02

    Understand the motor

    BLDC drivers, encoders, current limits, torque control.

    Done whenOne brushless joint holds position, then follows a torque command, on a bench supply with a current limit set.

    Read / watch / run

    • ODrive getting started

      ODrive Robotics · Docs · free · docs.odriverobotics.com/v/latest/guides/getting-started.html

      Wiring, configuring and calibrating a BLDC motor driver: the core of every quasi-direct-drive leg.

    • moteus brushless controller

      mjbots · Repo · free · github.com/mjbots/moteus

      Open firmware and docs for a small high-rate BLDC controller used in many legged robots.

    • Open robot actuator hardware

      Open Dynamic Robot Initiative · Repo · free · github.com/open-dynamic-robot-initiative/open_robot_actuator

      An open, printable brushless actuator module and leg: CAD, BOM and assembly.

    BenchOne brushless motor, a driver board and a current-limited bench supply.
  3. 03

    Gaits and balance

    Leg kinematics, gait scheduling, IMU state estimation.

    Done whenYou can explain why the robot tips in a trot and change a gait parameter to fix it.

    Read / watch / run

    • StanfordQuadruped code

      Stanford Student Robotics · Repo · free · github.com/stanfordroboticsclub/StanfordQuadruped

      The original Pupper controller: inverse kinematics, gait scheduler, joystick control.

    • Pupper v3 documentation

      Stanford Student Robotics · Docs · free · pupper-v3-documentation.readthedocs.io/en/latest

      Build and program a brushless quadruped with ROS 2 and a learned walking policy.

    • Underactuated Robotics

      Russ Tedrake, MIT · Book · free · underactuated.mit.edu

      Walking, balance and control of robots that can fall over: the theory behind legged locomotion.

    BenchThe robot you built in the build step of this path.
  4. 04

    Learn to walk in sim

    Train a locomotion policy in parallel simulation, check it sim-to-sim.

    Done whenA trained policy walks in MuJoCo on a model that was not the training sim.

    Read / watch / run

    • MuJoCo Playground

      Google DeepMind · Tool · free · playground.mujoco.org

      GPU-accelerated RL environments for quadrupeds, humanoids and hands, with sim-to-real examples.

    • legged_gym

      ETH Zurich Robotic Systems Lab · Repo · free · github.com/leggedrobotics/legged_gym

      The reference RL setup for learning to walk in parallel sim, used by many open policies.

    • Isaac Lab documentation

      NVIDIA · Docs · free · isaac-sim.github.io/IsaacLab/main/index.html

      GPU robot-learning framework: environments, RL and imitation workflows, sim-to-real.

    Published policies on humanoidrobots.training
    BenchA PC with a GPU for training; the robot from step 1 only to test afterwards.

Resource links checked 2026-09-25.