Free courses, lecture series, books and exercises from robotics educators and project maintainers, arranged as a learning path: foundations first, then the software stack, specialisms and research.
Learning material is free; equipment and compute may cost extra. Levels are our guidance. Activities have not been tested by us. How we select resources
No matching resources
Try a broader search or reset the filters.
Robotics foundations 11
Kinematics, mathematics and introductory courses that the rest builds on.
Teaches kinematics in moving frames, force-momentum and work-energy methods, Lagrange's equations, linearization and vibration of mechanical systems.
Beginner
Video lessons
Read or watch
First activity & requirements
Try first
Watch 1. History of Dynamics; Motion in Moving Reference Frames, answering the concept questions in Problem Set 1 as the unit page instructs.
What you need
Suggested: single-variable calculus and introductory physics.
What is free
The OCW course page lists lecture notes, lecture videos, problem sets with solutions, exams with solutions and problem-solving videos. License: Creative Commons BY-NC-SA.
Keep in mind
Fall 2011 undergraduate course, 39 videos including recitations. General mechanics, not robot-specific.
Asokan T, Balaraman Ravindran and Krishna Vasudevan / IIT Madras (NPTEL)
Covers transforms, DH parameters, forward and inverse kinematics, Jacobians, electric motor drives, Bayes, Kalman and particle filters, and path planning.
Beginner
Video lessons
Read or watch
First activity & requirements
Try first
Watch video #8, DH Algorithm, then video #9 on DH Algorithm examples in the same playlist.
What you need
Suggested: undergraduate linear algebra, basic mechanics and basic circuits. No software is needed to follow the videos.
What is free
The playlist shows 43 free videos on the NPTEL-NOC IITM YouTube channel.
Keep in mind
Course page names Prof. Asokan T, Prof. Balaraman Ravindran and Prof. Krishna Vasudevan, IIT Madras. Includes two probability tutorial videos. Playlist last updated Jan 16, 2025.
Study OpenUSD stage structure, composition and reusable assets as suggested background for USD-based robot simulation asset workflows.
Beginner
Course
Computer
First activity & requirements
Try first
Start the Stage lesson and inspect its Python examples for creating and saving a USD stage.
What you need
The introductory Stage lesson includes Python and Jupyter notebook examples. The later Creating Composition Arcs exercises specify usdview, Python, downloaded exercise content and a Visual Studio Code workspace. These are module-specific requirements, not a curriculum-wide minimum.
What is free
NVIDIA describes the learning path as free, and the Stage lesson is publicly readable with substantive examples. The provider mentions a certification exam; its access terms were not established in this inspection.
Keep in mind
The inspected lessons progress from stage fundamentals to composition and asset workflows. This is suggested background for USD-based robotics simulation, not a robotics course or certification.
Tamim Asfour / KIT High Performance Humanoid Technologies (H2T)
Covers kinematic and dynamic robot modelling, control, trajectory and motion planning, grasp planning, image processing and programming by demonstration.
Intermediate
Video lessons
Read or watch
German
First activity & requirements
Try first
Watch 'Robotik 1 - Kapitel 2 - Kinematik', then the exercise session 'Robotik 1 - Übung 2 - Kinematik'.
What you need
No prerequisites stated on the course page. Suggested: German for these recordings, linear algebra and basic mechanics.
What is free
The 28 recordings, lectures and Übung sessions, are free on the H2T YouTube channel. The course page links an ILIAS course for the current term.
Keep in mind
Recorded winter semester 2020 with German titles. The course page now lists English as language of instruction for WS 26/27, with lecturers Tamim Asfour and Katja Mombaur.
Structure robot task logic as behavior trees with BehaviorTree.CPP: custom action nodes, XML trees, ports, subtrees, scripting, loggers and ROS 2 wrappers.
Intermediate
Tutorial
Computer
First activity & requirements
Try first
Work through 01. Your first Behavior Tree: write the ApproachObject SyncActionNode, register nodes with BehaviorTreeFactory, load my_tree.xml and tick the tree.
What you need
Suggested: C++ and a CMake build setup. ROS 2 is needed only for the BehaviorTree.ROS2 wrappers on the integration page.
What is free
Documentation pages are public. Tutorial 11 connects to the Groot2 editor.
Keep in mind
Docs for version 4.8, with 11 basic tutorials. The ROS 2 page recommends a central coordinator node built with BT.CPP that calls ROS 2 actions and services.
Understand how ros2_control connects controllers to hardware interfaces through a mobile-robot simulation lesson.
Intermediate
Tutorial
Computer
First activity & requirements
Try first
Trace the controller-manager diagram and identify the hardware interface and controller in the example.
What you need
ROS 2, a compatible Gazebo setup and familiarity with robot description files.
What is free
The written article is publicly accessible; real-robot continuation requires hardware.
Keep in mind
The inspected lesson mixes gazebo_ros2_control and gz_ros2_control names. A matching runnable configuration remains unresolved; use the conceptual explanation before attempting setup.
Original resource Access checked 2026-09-30 · English · Source reviewed
A structured NVIDIA course sequence covering a first simulated robot, URDF assets, synthetic perception data, software-in-the-loop and hardware-in-the-loop.
Beginner
Course
GPU / workstation
First activity & requirements
Try first
Start Building Your First Robot in Isaac Sim, then add control and sensor data in the first module.
What you need
The lessons are readable in a browser. Running the exercises requires Isaac Sim; the exact release-specific local hardware requirements were not verified in this course-page inspection. Check the matching Isaac Sim requirements before installation.
What is free
The lessons are freely readable; running Isaac Sim may require capable local hardware or paid cloud compute.
Keep in mind
The modules proceed from a first simulated robot through URDF import, synthetic perception data and software- or hardware-in-the-loop.
Use supplied assets, an AI coding agent and Omniverse libraries to render an OpenUSD scene, then prepare it for a physics drop test.
Beginner
Course
Computer
First activity & requirements
Try first
Open the chosen course Overview to get its environment and materials, then read Get Started before initializing the coding agent.
What you need
The inspected bootcamp minimum is Windows 11 or Ubuntu 22.04/24.04, 16 GB RAM, an RTX 3070 or better and 250 GB free storage. The forthcoming Isaac Sim module has higher and Linux-specific requirements.
What is free
NVIDIA calls the self-paced lessons free; agent account, local hardware and optional cloud costs are separate.
Keep in mind
The local developer setup and course-specific Brev setup instructions are marked as coming soon; use the current course Overview environment and materials. This is simulation tooling, not a complete robot build.
Build small Newton physics simulations, control a Franka arm and train a cube-lift policy with Newton inside Isaac Lab.
Intermediate
Course
GPU / workstation
First activity & requirements
Try first
Read Getting Started and run the first Newton simulation before moving to the Isaac Lab module.
What you need
Tested Newton Fundamentals setup: Linux x86-64, Python 3.10 or 3.11, Newton 1.5.x, RTX GPU, CUDA 12.8 user-space libraries and NVIDIA driver 550 or newer. The Isaac Lab module uses v3.0.0-beta2.
What is free
The course and lesson scripts are public. Optional Brev GPU instances are billed while running, including idle time.
Keep in mind
The versions are tested course configurations, not a general compatibility promise. Prior Python comfort is expected.
Learn to assemble, validate and optimize a factory digital twin using Omniverse and OpenUSD assets.
Intermediate
Course
Computer
First activity & requirements
Try first
Begin Getting Started With Digital Twin Scene Assembly and inspect the factory asset library.
What you need
The course recommends familiarity with 3D concepts and basic Python. Application access and rendering requirements should be checked before attempting the exercise.
What is free
The written lessons are public; access to the course application and compute may have separate conditions.
Keep in mind
Included as an adjacent prerequisite for robotics simulation; it teaches facility-scale scenes rather than a robot task.
Provides SOFA simulation tutorials for soft robots, from a first-steps scene to cable and pneumatic grippers, a trunk and a tripod.
Intermediate
Tutorial
Computer
First activity & requirements
Try first
Open firststeps-tuto.py from the FirstSteps tutorial in your runSofa application and follow its steps, then move on to the CableGripper tutorial.
What you need
SOFA with the SoftRobots plugin; the plugin site has download and build pages for Linux, Windows and macOS. Suggested: Python basics.
What is free
Tutorials are in the plugin's examples/tutorials folder on GitHub. The plugin site lists the tutorials and has an overview video of the CableGripper tutorial.
Keep in mind
The tutorials folder lists nine tutorials, including CircularRobot, DiamondRobot, SoftArmGripper and SoftFingerDesign. The plugin site is run by the INRIA Lille DEFROST team.
Begin Best Practices for Robotics and OpenUSD by studying the asset structure and UR10e URDF import steps.
What you need
Prior ROS, Isaac Sim and USD familiarity is helpful. Running examples requires matching simulation software and a capable system; a GRID lesson uses a third-party service.
What is free
The lesson pages are public. Software, compute and any third-party GRID service terms need separate checks before hands-on use.
Keep in mind
This is a collection of separate advanced lessons rather than one validated end-to-end robot build.
Original resource Access checked 2026-10-13 · English · Source reviewed
Learn to create simulation scenes and reinforcement-learning environments with Isaac Lab's worked examples.
Advanced
Tutorial
GPU / workstation
First activity & requirements
Try first
Start with an empty scene, then add a rigid object before constructing a learning task.
What you need
The inspected guide specifies Ubuntu 22.04 or Windows 11, 32 GB RAM and 16 GB GPU VRAM. Follow its NVIDIA driver requirements and release-matched Python; use the recommended Isaac Sim pip plus Isaac Lab source installation to include example scripts.
What is free
How-to material and example source are public. Hardware, cloud compute and component license terms are separate.
Keep in mind
The main documentation tracks development and may differ from stable releases. Follow one release's installation and tutorials together.
Understand why robot vision transforms an image before extracting information for a task.
Beginner
Video lessons
Read or watch
First activity & requirements
Try first
Read the lesson transcript and identify a use for contrast adjustment, reduced resolution and edge highlighting.
What you need
The provider suggests high-school mathematics, basic optics and some programming familiarity. Reading this introductory transcript needs no robot or MATLAB installation.
What is free
The written lesson transcript is publicly accessible without signing in.
Keep in mind
This is a conceptual introductory lesson with no accompanying code. Video playback was not tested; sign-in is required for comments and ratings.
Original resource Access checked 2026-10-01 · English · Source reviewed
Learn image formation, structure from motion, stereo, graphical models, optical flow, shape-from-X and recognition, with a strong emphasis on 3D vision.
Intermediate
Video lessons
Read or watch
First activity & requirements
Try first
Watch Lecture 3.2 (Structure-from-Motion: Two-frame Structure-from-Motion) with the L03 Structure-from-Motion slides listed on the course page.
What you need
Provider lists basic CS skills, basic Python and PyTorch, linear algebra, probability and information theory, and experience with deep learning.
What is free
Videos are free on YouTube; the course page lists slides and problems per week. It says participants must register via its Quiz Server; download access was not checked here.
Keep in mind
46 videos; playlist last updated May 16, 2023. Course ML-4360 has 6 assignments, some using PyTorch. The course page says the class will not be taught in SS 2026.
Anthony Opipari and Chad Jenkins / University of Michigan
Build deep learning skills for robot perception, from linear classifiers and CNNs to object detection and segmentation, with a robot-object dataset.
Intermediate
Video lessons
Computer
First activity & requirements
Try first
Watch DeepRob Lecture 2 - Image Classification, then do Project 1: download and unzip P1.zip, upload the extracted P1 directory to Google Drive, and implement the KNN classifier on PROPS in Colab.
What you need
Suggested: Python, linear algebra and calculus. Project 1 setup uses Google Drive and Google Colab, which need a Google account.
What is free
The 22 videos are free on YouTube. The Project 1 page offers P1.zip starter code. The homepage links a forum, autograder and Gradescope.
Keep in mind
ROB 498-002 and 599-009, Winter 2023. 22 videos: Lectures 1 to 13 plus Discussions 1 to 9. Project 1 also covers SVM and Softmax classifiers.
Build intuition for noisy sensors and state estimation through an executable Python book.
Intermediate
Book / notes
Computer
First activity & requirements
Try first
Start with the introductory notebook and vary the noise in a worked filtering example.
What you need
Python/Jupyter and the book's dependencies; basic algebra and probability help.
What is free
Notebook text and source are publicly available; hosted notebook providers may impose limits.
Keep in mind
The author prioritizes teaching over efficiency and warns that early implementations may be numerically unstable. Use the repository's environment instructions.
Original resource Access checked 2026-09-30 · English · Source reviewed
Learn 3D vision geometry: rigid motion, projection, correspondences, two-view and multi-view reconstruction, bundle adjustment and direct visual SLAM.
Advanced
Video lessons
Read or watch
First activity & requirements
Try first
Watch MVG - Lecture 06: Perspective Projection, then do Exercise 4: Perspective Projection from the course page and compare with Solution 4.
What you need
Suggested: linear algebra and calculus; the course itself opens with two lectures on mathematical background in linear algebra.
What is free
Videos are free on YouTube. The course page lists slides, 10 theoretical exercise sheets with solutions and 5 practical exercise packages; announcements use Moodle.
Keep in mind
Summer Semester 2025 (IN2228), lecturer Daniel Cremers. 21 videos for 17 scheduled lecture slots, some split into parts; playlist last updated Jul 7, 2025.
Investigate visual-inertial estimation by setting up the research software and its evaluation dependencies.
Advanced
Research reproduction
Computer
First activity & requirements
Try first
Read the installation and dataset guidance, then plan one recorded-data evaluation with a fixed configuration.
What you need
C++ build environment, Eigen, OpenCV and Ceres; ROS support depends on the documented platform.
What is free
Public setup and evaluation documentation; dataset storage and local computation are separate.
Keep in mind
The guide documents Ubuntu 20.04 with ROS Noetic or ROS 2 Galactic among its older pairings. These are provider-reported configurations; newer environments are unverified here.
Original resource Access checked 2026-09-30 · English · Source reviewed
Reproduce visual or visual-inertial SLAM examples using recorded EuRoC or TUM-VI data.
Advanced
Research reproduction
Computer
First activity & requirements
Try first
Build the documented environment and run one supplied dataset sequence before comparing configurations.
What you need
The provider lists Ubuntu 16.04/18.04, C++11 or C++0x, Pangolin, OpenCV and Eigen3. Trajectory evaluation also needs Python and NumPy; these are provider-tested configurations, not tested by us.
What is free
Source and reproduction instructions are public; consult dataset providers for their terms.
Keep in mind
The README identifies the V1.0 research release and legacy dependencies. No accuracy or successful build claim is made here.
Original resource Access checked 2026-09-30 · English · Source reviewed
Implement graph search, informed search, dynamic programming, Dubins paths and collision checking as Python exercises with local test cases and reports.
Intermediate
Exercises
Computer
First activity & requirements
Try first
After the Preliminaries and Hello-world pages, implement BFS, DFS and Iterative Deepening in exercises/ex02/algo.py, then compare your output with the local test reports in out/.
What you need
Graph search lists Preliminaries and Hello-world as prerequisites. The repository includes a dev container configuration. Suggested: Python and data structures.
What is free
Exercise pages are public and the GitHub repository is public under GPL-3.0. Local test cases are not graded; the page says final evaluation uses private test cases.
Keep in mind
Fall 2026 edition. Graph search also covers Wavefront Planning; previous-year exercises include Optimization, Driving Games and Spaceship.
Learn manipulator motion planning in Python with Pinocchio: kinematics, collision checking, inverse kinematics, RRT and PRM planning, and trajectory generation.
Intermediate
Tutorial
Computer
First activity & requirements
Try first
Install pyroboplan following the repository Setup section, run the 2-DOF RRT example, then the RRT example for the full Franka robot arm described on the Motion Planning page.
What you need
Python; the library uses the Pinocchio Python bindings and the Coal library for collision checking. Suggested: basic linear algebra and robot kinematics.
What is free
Documentation is public. The GitHub repository shows an MIT license and install commands for PyPI or from source.
Keep in mind
The docs say it is educational and not designed for performance, and point to RoboPlan for serious work. All current planners in pyroboplan are offline planners.
Explore forward and inverse kinematics, Jacobians and robot models in Python.
Intermediate
Exercises
Computer
First activity & requirements
Try first
Load a supplied arm model and compare forward kinematics with an inverse-kinematics solution.
What you need
Python 3.10 or later. On Windows, the README says pip collision checking is unavailable because coal has no Windows wheels and documents a conda alternative; neither route was tested by us.
What is free
Public examples and code; printed companion books are separate.
Keep in mind
Windows Swift support is unresolved: the installation section says other features work normally, while troubleshooting says Swift is unsupported and offers a hotfix. The numerical kinematics activity avoids that optional visualization path.
Original resource Access checked 2026-09-30 · English · Source reviewed
Use cuRobo from Python for GPU-based forward kinematics, collision checking, inverse kinematics, collision-free motion generation and MPC for robot arms.
Advanced
Tutorial
GPU / workstation
First activity & requirements
Try first
After installing, follow Using in Python: compute forward kinematics for the Franka with CudaRobotModel, then plan a trajectory with MotionGen for the UR5e example.
What you need
NVIDIA GPU (Volta or newer, 4 GB VRAM or more), Ubuntu 20.04 or 22.04, Python 3.8 to 3.10, PyTorch. Isaac Sim is optional but used in the robot configuration tutorial.
What is free
Documentation pages are public. Isaac Sim is not required for the Python API.
Keep in mind
The MPC section cites 500 Hz on an RTX 4090 and marks MPC experimental, without safety guarantees and able to collide; the provider advises testing in simulation first. Configuring a New Robot uses Isaac Sim's Lula Robot Description Editor to build collision spheres.
Introduces classical control: transfer functions, Laplace transforms, stability, Routh-Hurwitz, Bode plots, root locus, PID and Nyquist.
Beginner
Video lessons
Read or watch
First activity & requirements
Try first
Watch Control Systems Lectures, Closed Loop Control, then PID Control, A brief introduction from the same playlist.
What you need
Suggested: algebra and basic calculus.
What is free
The public playlist on the instructor's own YouTube channel lists 26 videos. Slides and assignments were not identified on the inspected pages.
Keep in mind
The first video was published 2012-09-15; the others about 13 to 14 years ago. Covers classical control fundamentals, including time- and frequency-domain topics.
Explains state-space control: stability, controllability, pole placement, LQR, observability, Kalman filtering, LQG, robust control and loop shaping.
Intermediate
Video lessons
Read or watch
First activity & requirements
Try first
Watch Inverted Pendulum on a Cart, then Linear Quadratic Regulator (LQR) Control for the Inverted Pendulum on a Cart from the playlist.
What you need
Suggested: linear algebra and differential equations. The playlist description says examples are applied in Matlab.
What is free
The public playlist on the instructor's YouTube channel lists 39 videos. The book site's Chapter 8 page links the videos by section, and the playlist description links chapters at databookuw.com.
Keep in mind
Videos were published about 8 to 9 years ago. The playlist description calls it a rapid, non-exhaustive overview that follows Chapter 8 of Data-Driven Science and Engineering.
Pranav A. Bhounsule / University of Illinois Chicago
Builds legged robot models step by step: Euler-Lagrange dynamics, hybrid systems, passive walkers, Raibert hopping, trajectory optimization and 3D bipeds.
Intermediate
Video lessons
Computer
First activity & requirements
Try first
Watch Legged Robotics 8a: Passive Dynamic Walker (part 1 - derivation), then use Lec 8 Code and the Passive Dynamic Walker PDF from Module 3 on the course page.
What you need
Course page: MATLAB 2021 with symbolic and optimization toolbox and CoppeliaSim v4.1.0 (free for education, Lua) were used. MATLAB licensing must be checked separately. Suggested: dynamics and linear algebra.
What is free
The 48 playlist videos are free. The course page links lecture notes PDFs, per-lecture code, all notes zipped and all code zipped.
Keep in mind
Playlist has 48 videos, including three MATLAB videos labelled Fall 2020. Course page last updated November 26, 2022.
Practises nonlinear control of mobile robots: feedback linearization, sliding modes, model-free control, vehicle control, potential fields and path planning.
Intermediate
Exercises
Computer
French
First activity & requirements
Try first
With Python installed, take crank.py from robmoocpy.zip plus roblib.py, solve Exercice 1 (Manivelle), then compare with the lesson's correction video.
What you need
Stated: some notions of MATLAB or Python and elementary mathematics at engineering-school entry level; MATLAB or Python must be installed.
What is free
The page describes the MOOC as free and open to all. Registration is not required to follow the videos, only for the diploma; statements and starter programs are listed as files.
Keep in mind
Listed as the 11th edition, 2027: lessons A to F with 37 exercises (33 of 43 points for the diploma). Lesson D has no lecture video. The main course is in French; the book page mentions correction videos in French or English, but English coverage of this exercise was not established.
Marco Pavone and Daniele Gammelli / Stanford University
Teaches calculus of variations, indirect and direct trajectory optimization, dynamic programming, LQR, reachability, MPC, imitation learning and RL.
Advanced
Video lessons
Computer
First activity & requirements
Try first
Watch Lecture 8, LQR-Style Algorithms, then work Homework 2 (sequential convex programming, dynamic programming, LQR) from the course homework page.
What you need
Suggested: linear algebra, ODEs, optimization and Python. Course recitations cover automatic differentiation with JAX and convex optimization with CVXPY.
What is free
The 19 Spring 2026 lecture videos are on the Stanford Online YouTube channel; the course site links lecture slides and code. The site says recordings and recitations are on Canvas.
Keep in mind
Spring 2026 offering, 19 videos. The course schedule also lists an optional guest lecture, an optional extra lecture on world models, and HW0 to HW4.
Covers MDPs, LQR and iLQR, optimization-based control, motion planning, Kalman and particle filters, POMDPs, imitation learning, RL and sim2real.
Advanced
Video lessons
Computer
First activity & requirements
Try first
Watch Lecture 5 LQR, then work through Problem Set 2: LQR, iLQR, DDP, Feedback Linearization (hw2.zip on the course page).
What you need
Course page prerequisites: mathematical proofs, probability, algorithms, linear algebra and ability to implement algorithmic ideas in code (CS188, EE126, EE127).
What is free
The course page links slides and a YouTube recording for most lectures, and five problem set zips (hw1.zip to hw5.zip) plus a midterm study handout.
Keep in mind
Fall 2019 offering; the playlist has 24 videos and no Lecture 16. The course page lists the Ike and Waymo industry sessions without slides or video links.
Covers marine craft kinematics, kinetics, hydrodynamics, guidance, navigation and motion control for ships, USVs and AUVs, with Python vehicle models.
Advanced
Book / notes
Computer
First activity & requirements
Try first
Read the Chapter 2 Kinematics slides, then install the Python Vehicle Simulator with pip install -e followed by the cloned repository path, as the README shows, then run python3 main.py from src/python_vehicle_simulator to try the remus100 or otter vehicle.
What you need
Python 3 with numpy, matplotlib and pytest (installed by the README pip command). Optional MSS m-files also run in GNU Octave. Suggested: dynamics and control basics.
What is free
Chapter slide PDFs are downloadable from the FossenHandbook repository, and the simulator code is on GitHub. The textbook itself is published by Wiley (2nd edition, 2021).
Keep in mind
Slides for chapters 1 to 16, last modified 2021 to 2026; Ch5 and Ch6 are marked 3rd edition and appendices A to D list no slides. Used in NTNU course TTK4190.
Teaches nonlinear control: Lyapunov theory, sliding and robust control, adaptive and adaptive robot control, feedback linearization, contraction analysis.
Advanced
Video lessons
Read or watch
First activity & requirements
Try first
Watch Lecture 2, Basic Lyapunov Theory, from the MIT index. The index pairs it with Sections 3.3 to 3.4.2 of Applied Nonlinear Control, a textbook whose access was not checked.
What you need
Suggested: linear systems and ODEs. Textbook for lectures 1 to 13: Applied Nonlinear Control, Slotine and Li, Prentice-Hall 1991; papers R1 to R5 for lectures 14 to 20.
What is free
The MIT index page lists 20 lecture recordings with durations and readings. Lecture 1 is on the YouTube channel Lectures on Nonlinear Systems, MIT 2.152.
Keep in mind
The MIT index page is the starting point for the series. The Lecture 1 video is titled 2013_09_05_2.152 and was published 2013-09-09.
Michael Triantafyllou and Franz Hover / MIT OpenCourseWare
Explains equations of motion, hydrodynamic coefficients, control surfaces, stability and automatic control for surface and underwater vehicles.
Advanced
Book / notes
Read or watch
First activity & requirements
Try first
Download the Maneuvering and Control of Marine Vehicles textbook (full text of the course lecture notes) and read it alongside the syllabus and calendar.
What you need
Listed as a graduate-level course. Suggested: dynamics, fluid mechanics and linear control background.
What is free
The open textbook PDF (2 MB) is free to download from MIT OpenCourseWare, which also lists syllabus, calendar and assignments pages.
Keep in mind
Fall 2004 course by Prof. Michael Triantafyllou, originally numbered 13.49. Topics include Kalman filtering, optimal control and applications to autopilots, ocean towing and ROVs.
Covers dynamics, optimization, Pontryagin, LQR, convex MPC, trajectory optimization, DDP, 3D rotations, legged robots, LQG and Kalman filters.
Advanced
Video lessons
Computer
First activity & requirements
Try first
Watch playlist item 8, The Linear Quadratic Regulator Three Ways (its title says 2024 Lecture 8), then try Assignment 2 (LQR, TVLQR, and Convex MPC) from the homeworks page.
What you need
Suggested: strong linear algebra, ODEs and programming. Assignment 0 is a Julia warm-up and homeworks are completed as Jupyter notebooks.
What is free
The course lectures page lists a video and slides for each lecture; the homeworks page lists Assignments 0 to 4. Zoom links are via Piazza and Canvas; submissions go to Gradescope.
Keep in mind
Playlist Optimal Control 2025: 24 videos, last updated Apr 18, 2025; items 8 and 13 are titled 2024. The lectures page also lists 2024 and 2023 editions. Channel: MIT Robotic Exploration Lab.
Watch 'Denavit-Hartenberg Convention', then 'Academic Examples: Denavit-Hartenberg Method' in the same playlist.
What you need
Suggested: Spanish and basic linear algebra; CoppeliaSim and ABB RobotStudio to repeat the simulation and RAPID videos.
What is free
54 videos are free on the instructor's YouTube channel. UPV RiuNet lists his video 'Sistemas robotizados. Matriz de rotación' as a Polimedia learning object.
Keep in mind
Playlist description says these are theory videos used in UPV robotics subjects. Most run 2 to 22 minutes; includes CoppeliaSim and meArm tutorials. Titles mix Spanish and English.
Teaches how to describe object motion in space, forward and inverse kinematics of multi-joint arms, and trajectory planning, ending with a worked example.
Stated audience: learners with a science or engineering background who know basic mechanics and linear algebra. Mandarin comprehension.
What is free
All 7 lectures are on NTU OCW and its YouTube channel. The OCW page says Coursera's free mode now previews only the first unit; the full Coursera course is paid.
Keep in mind
OCW page gives teaching date May 2017 and the Department of Mechanical Engineering; YouTube uploads show as 4 months old. Reuse requires permission from the rights holder.
Jacques Gangloff / Université de Strasbourg, Télécom Physique Strasbourg
Teaches manipulator robotics: positioning and kinematics, Denavit-Hartenberg modelling, direct and inverse Jacobians, Euler-Lagrange dynamics and control.
Intermediate
Video lessons
Read or watch
French
First activity & requirements
Try first
Watch 'Cours de Robotique : séance 1 sur 9', then answer the Google Forms quiz linked at the end of the chapter; the videos correct the quizzes.
What you need
French. Stated prerequisites: engineering mathematics, with basic linear algebra, 3D geometry and basic rigid-body mechanics.
What is free
The course page links PDF slides, lab work (travaux pratiques) and robotics exam papers with solutions. The 9 sessions are on the instructor's YouTube channel.
Keep in mind
Nine 1h45 sessions plus one 4h lab, level 3e année d'école d'ingénieurs / M2 (bac+5). The 32-video playlist continues with other courses after séance 9.
Choose between an agent-led simulated healthcare robotics workflow and a code-first robotic ultrasound course with anatomy, teleoperation and policy training.
Intermediate
Course
GPU / workstation
First activity & requirements
Try first
Open the healthcare path and choose the agentic workflow or the independent ultrasound deep dive.
What you need
For the agentic course, the inspected setup lists Ubuntu 22.04 or 24.04 on x86_64 or aarch64, an NVIDIA GPU with driver 580.65.06 or newer for Isaac Sim 5.1.0 and CUDA 13, and about 30 GB free disk. The separate ultrasound course may have different requirements.
What is free
The overview and agentic-course lessons are publicly readable. The separate ultrasound course is linked from the overview; its substantive lessons were not inspected in this pass.
Keep in mind
This is training material for simulation and research workflows, not evidence of clinical validation or medical safety.
Follow a 15-tutorial path from rigging a wheeled robot to setting up a UR10e with a 2F-140 gripper and running a pick-and-place sequence in Isaac Sim.
Intermediate
Tutorial
GPU / workstation
First activity & requirements
Try first
Do Tutorial 9: Pick and Place Example, running its four standalone scripts from gripper control to the full pick-and-place with cuMotion RMPflow or PINK IK.
What you need
Requirements page minimum: GeForce RTX 4080 with 16 GB VRAM, 32 GB RAM, Ubuntu 22.04/24.04 or Windows 11. The series recommends starting with Setup a Wheeled Robot.
What is free
Documentation pages are public. Tutorial 9 runs with packaged UR10e and Robotiq assets, so no custom USD, URDF or XRDF is required.
Keep in mind
Manipulator tutorials are 6 to 9; Tutorial 9 is listed as 30 minutes. PINK wheels are Linux x86_64 only, so Windows users run the cuMotion example.
Domenico Prattichizzo, Monica Malvezzi and Maria Pozzi / University of Siena (SIRSLab)
Explains grasp modeling and control: friction cones, the grasp matrix, hand and object kinematics, contact models, grasp properties and SynGrasp.
Intermediate
Video lessons
Read or watch
First activity & requirements
Try first
Watch Lecture 1.3: The friction cone and Lecture 1.4: The Grasp Matrix, then Lecture 2.3: Contact models.
What you need
The course announcement says Unit 4 introduces the SynGrasp MATLAB Toolbox. Suggested: linear algebra and rigid-body kinematics.
What is free
The 16 playlist videos are free. The playlist links a course page on sirslab.dii.unisi.it; slides and exercises for most units were not identified on the inspected pages.
Keep in mind
Four units taught by Domenico Prattichizzo, Monica Malvezzi and Maria Pozzi, posted October 2018. Unit 4 (SynGrasp) has one video in the playlist.
Follow a vial-to-rack SO-101 manipulation workflow through calibration, demonstration data, GR00T training, simulation evaluation and real robot tests.
Intermediate
Course
Physical robot
First activity & requirements
Try first
Read How to Take This Course and the theory sections before deciding whether to assemble the physical workspace.
What you need
The theory module requires no equipment. Full participation uses an SO-101 arm, workspace parts and GPU compute. NVIDIA reports testing on Ubuntu 24.04 with an RTX 5090 Laptop edition and 64 GB RAM, or an RTX PRO 6000 Blackwell Workstation Edition and 125 GB RAM; these are tested configurations, not published minimums.
What is free
Self-paced lessons are public; the full hands-on exercise requires purchased or otherwise sourced hardware and compute.
Keep in mind
The workflow is for a specific vial-to-rack setup. The page offers pretrained checkpoints and datasets as one route; the complete run has not been tested by this catalog.
David Dudás / BME MOGI (Budapest University of Technology and Economics)
Build a 4 axis arm in URDF, drive it with gz_ros2_control and a trajectory controller, write custom IK, then use MoveIt 2 with it, OpenMANIPULATOR-X and UR3e.
Intermediate
Tutorial
Computer
First activity & requirements
Try first
Clone the Simple-arm starter-branch into your colcon workspace, rebuild and source it, launch spawn_robot.launch.py, then follow Building the robotic arm to add shoulder, elbow, wrist and gripper.
What you need
ROS 2 Jazzy with Gazebo Harmonic, plus apt packages named in the lesson (controller-manager, gz-ros2-control, joint-trajectory-controller, MoveIt). Suggested: earlier course weeks.
What is free
Both lessons are public GitHub READMEs with starter branches to clone.
Keep in mind
With the 4 DoF arm, MoveIt cannot set every TCP orientation; the lesson covers position_only_ik and adding fake roll and yaw joints. Week 11-12 covers OpenMANIPULATOR-X and UR3e with MoveIt 2.
楊谷洋 Gu-Yang Yang / National Yang Ming Chiao Tung University OCW
Covers spatial representation, forward and inverse kinematics, differential relationships, motion planning, dynamics, control and compliance, VR telerobotics.
Intermediate
Video lessons
Read or watch
Mandarin Chinese
First activity & requirements
Try first
Watch 單元三 Kinematics and Inverse Kinematics (1/4), then continue with parts 2 to 4 of that unit.
What you need
Suggested: Mandarin comprehension, linear algebra and rigid-body dynamics.
What is free
26 videos are free on the NYCU OCW YouTube channel.
Keep in mind
Playlist last updated Jun 28, 2017; unit titles mix Chinese and English. Slides, homework and licence terms were not identified in the inspected excerpt.
Original resource Access checked 2026-10-05 · Mandarin Chinese · Source reviewed
Prototype robot behaviors in a 2D mobile robot simulator: build worlds, navigate, pick objects, add path planners and design task and motion planners.
Beginner
Tutorial
Computer
First activity & requirements
Try first
Clone the repository and complete Local Setup, run python3 examples/demo.py from the pyrobosim subfolder, enter bedroom desk in the Goal query box, click Navigate, then Pick.
What you need
The provider lists tested pairings: Python 3.10 on Ubuntu 22.04, 3.12 on 24.04 and 3.14 on 26.04, optionally with ROS 2. PDDLStream is an optional setup prompt for task and motion planning.
What is free
Documentation is public and the project invites open-source contributions. pip install is available but omits ROS 2 and task and motion planning features.
Keep in mind
Version 5.0.1 docs. The demo also runs as a local browser interface. Worlds can be exported to Gazebo for 3D testing.
Explore modular activities combining robotics theory, code and experiments around small autonomous vehicles.
Intermediate
Exercises
Computer
First activity & requirements
Try first
Choose a learning experience and check whether it uses simulation, physical hardware or both.
What you need
Python/Docker setup; account and hardware requirements depend on the learning experience.
What is free
Public learning documentation; some platform features require accounts and hardware is purchased separately.
Keep in mind
Activities include solutions; exercises do not. The manual warns that MOOC activities still use daffy code and documentation, even though this page belongs to the ente manual.
Original resource Access checked 2026-09-30 · English · Source reviewed
Introduces software and algorithms for autonomous marine vehicles, with MOOS-IvP labs on MOOS, helm autonomy, multiple vehicles and inter-vehicle messaging.
Intermediate
Course
Computer
First activity & requirements
Try first
After building the MOOS-IvP tree in Lab 1 (Machine Setup), do Lab 2, Introduction to MOOS: launch the MOOSDB, scope it with uXMS, poke it with uPokeDB and script pokes with uTimerScript.
What you need
Listed prerequisite: permission of instructor. Lab 2 assumes a MOOS-IvP tree built on your computer (macOS and Linux examples). Suggested: C++ and command-line basics.
What is free
The course home page and lab pages are readable on the course site, and Lab 2 offers a PDF version and links the matching lecture PDF. The home page also links a 2017 class video on YouTube.
Keep in mind
Spring 2026 course page. Labs run from 01 Machine Setup to 19 Auto Competitions; Lab 12 is Operating MIT Herons, and the description mentions field work with small vehicles on the Charles River.
Learn autonomous racing through lectures and labs on braking, reactive driving, mapping and planning.
Intermediate
Course
Computer
First activity & requirements
Try first
Start with the ROS 2 introduction lab and simulator before considering a racecar build.
What you need
The inspected Lab 1 uses ROS 2 Foxy on native Ubuntu or Docker for package, node and launch-file exercises. Later simulator activities and physical racing labs have additional requirements.
What is free
Public course-kit material; a physical car and in-person instruction are separate.
Keep in mind
The current course-kit title is RoboRacer, formerly associated with F1TENTH. Follow the version used by each lab.
Douglas G. Macharet / DCC, Universidade Federal de Minas Gerais
Teaches mobile robot kinematics and control, bug and potential-field planning, PRM/RRT, occupancy grids, Kalman, Markov and particle-filter localization.
Intermediate
Video lessons
Computer
Portuguese
First activity & requirements
Try first
Watch Aula 03 - Ferramental (Python e CoppeliaSim), then open the Ferramental notebook and Pioneer scene listed in the course repository.
What you need
Suggested: Portuguese, Python and basic probability. Jupyter and CoppeliaSim are needed to run the repository notebooks and scenes.
What is free
The repository lists a video and slides for each topic, plus Jupyter notebooks and CoppeliaSim scenes for many topics.
Keep in mind
23 videos, last updated Feb 16, 2022; most titles are Portuguese, some appear in English. Several notebooks have ZeroMQ variants.
Implement Monte Carlo Localization for a racecar, then grid-based or sampling-based path planning and pure pursuit, tested in a 2D racecar simulator with ROS 2.
Intermediate
Exercises
Computer
First activity & requirements
Try first
Set up the course Docker container, then do Lab 5 Part B: read the instructions notebook, implement the motion and sensor models, run the provided unit tests, and test MCL in simulation.
What you need
Git and Docker for the racecar container. Suggested: Python with NumPy (the lab advises NumPy for speed), ROS 2 basics, probability. Part C uses the physical racecar.
What is free
Lab handouts, starter packages and unit tests are public on GitHub. The handouts list Gradescope submissions, Canvas reports and TA checkoffs as course deliverables.
Keep in mind
Handouts include the term's due dates and grading weights. Lab 6 Part C integrates planning and pure pursuit in racecar_simulator before deploying on the physical racecar.
Learn the TurtleBot3 simulation workflow before moving to a physical mobile robot.
Intermediate
Tutorial
Computer
First activity & requirements
Try first
Select the instructions for your ROS distribution and launch the corresponding Gazebo world.
What you need
A supported ROS/Gazebo computer; no physical TurtleBot required for simulation.
What is free
Public simulation instructions and associated packages.
Keep in mind
Choose a single ROS/Gazebo version. The e-Manual announces migration to ROBOTIS Docs; the original guide was accessible at the check, but its replacement needs rechecking.
Original resource Access checked 2026-09-30 · English · Source reviewed
David Dudás / BME MOGI (Budapest University of Technology and Economics)
Learn mapping with SLAM Toolbox, localization with AMCL, Nav2 goal and waypoint navigation, and exploration for a simulated robot in ROS 2 Jazzy and Gazebo.
Intermediate
Tutorial
Computer
First activity & requirements
Try first
Clone the starter-branch into your colcon workspace, rebuild and source it, test spawn_robot.launch.py, then follow SLAM toolbox: install slam_toolbox and create mapping.launch.py.
What you need
Ubuntu 24.04 and ROS 2 Jazzy per the course intro (WSL2 or a VM also listed), Gazebo, and apt packages from the lesson. Suggested: the earlier course weeks.
What is free
The lesson and course intro are public GitHub READMEs. The intro describes final-project teams, deadlines and live submission for the course.
Keep in mind
Week 7-8 of the BME MOGI ROS 2 course. The starter branch supplies the robot package and configs; you still need the course environment and create the launch files during the lesson.
Asokan Thondiyath and Santhakumar Mohan / IIT Madras and IIT Palakkad (NPTEL)
Teaches wheeled robot kinematic and dynamic models, sensors and error modelling, Markov and Kalman localisation, SLAM, graph-search planning and motion control.
Intermediate
Video lessons
Read or watch
First activity & requirements
Try first
Watch mod01lec03 - Introduction to Mobile Robot Kinematics, then mod01lec04 - Degree of Maneuverability and Types of Wheels.
What you need
Suggested: engineering mathematics and basic dynamics. No software is needed to follow the videos.
What is free
The playlist shows 48 free videos on the NPTEL-NOC IITM YouTube channel.
Keep in mind
Course page names Prof. Asokan Thondiyath and Prof. Santhakumar Mohan (IITM, IIT Palakkad). An introduction plus mod01lec01 to mod08lec47; several cover kinematic and dynamic simulation.
Explains probabilistic robotics chapter by chapter: particle and Kalman filters, KLD sampling, FastSLAM, graph-based SLAM, MDPs, Q-learning, Sarsa and POMDPs.
Intermediate
Video lessons
Computer
Japanese
First activity & requirements
Try first
Watch 詳解確率ロボティクス第五章（前半、パーティクルフィルタ）, then open the section_particle_filter notebooks in LNPR_BOOK_CODES.
What you need
Suggested: Japanese, Python with Jupyter, and basic probability.
What is free
22 videos are free on the author's YouTube channel. Companion code notebooks are on GitHub. The videos explain the book 詳解確率ロボティクス linked in the playlist.
Keep in mind
Playlist last updated Aug 21, 2020; most videos were streamed. The code README warns some notebooks may fail with newer Python module versions.
Run the Autoware planning simulator on a sample map and try lane driving, parking, pull out and pull over, lane change, avoidance and crosswalk scenarios.
Advanced
Tutorial
Computer
First activity & requirements
Try first
Download the sample map with the demo_artifacts ansible role and check the ML model artifacts, launch planning_simulator.launch.xml, set initial and goal poses in RViz, then click Auto.
What you need
Autoware installed per its installation page (Debian packages are offered for Autoware Core only; check which install provides this demo). The installation page lists 8 CPU cores and 16 GB RAM; an optional NVIDIA GPU (4 GB) is not needed for basic functions; the page also links guidance for running detection without CUDA.
What is free
Documentation pages are public. The sample map is copyright TIER IV and downloaded through Autoware's setup tooling.
Keep in mind
The default maximum velocity is limited to 15 km/h; the page explains raising max_vel. Advanced scenarios include placing dummy objects and traffic light recognition.
Study state estimation for mobile robots: Kalman and particle filters, matrix Lie groups, invariant Kalman filtering, mapping, registration and SLAM.
Advanced
Video lessons
Computer
First activity & requirements
Try first
Watch Lecture 02-Bayes Filters-Kalman Filtering, then work Homework 2 (Estimation and Kalman Filtering) from the course's public GitHub repository.
What you need
Provider lists exposure to linear algebra, basic probability and statistics, estimation, matrix calculation and calculus, and familiarity with one programming language.
What is free
Lecture videos are free on YouTube. The course README points to its GitHub repository for slides, lecture notes and example code and lists seven homework assignments.
Keep in mind
Winter 2022 offering (NA 568/EECS 568/ROB 530). 29 videos: Lectures 01 to 20 plus guest and supplementary talks. The course uses MATLAB, Python, Julia and C++.
Survey self-driving: imitation learning, direct perception, RL, vehicle dynamics and control, odometry and SLAM, lane and object detection, and planning.
Advanced
Video lessons
Computer
First activity & requirements
Try first
Watch Lecture 7.1 (Odometry, SLAM and Localization: Visual Odometry), then work the E04 Exercises (L7) problems listed for that week on the course page.
What you need
Provider lists basic CS skills, basic Python and PyTorch, linear algebra, probability and information theory, and experience with deep learning.
What is free
Videos are free on YouTube; the course page links slides and problems per week. It says participants must register via its Quiz Server; download access was not checked here.
Keep in mind
48 videos; playlist last updated May 16, 2023. Course ML-4340 is run as a flipped classroom with 6 assignments; programming uses Python and PyTorch.
Learn SLAM from first principles: Kalman and information filter SLAM, grid maps, particle filters and FastSLAM, least-squares SLAM and SLAM front ends.
Advanced
Video lessons
Read or watch
First activity & requirements
Try first
Watch SLAM Course 04 (Extended Kalman Filter) and then SLAM Course 05 (EKF SLAM) from the playlist.
What you need
Suggested: probability, linear algebra and the Bayes filter (the playlist covers it in SLAM-Course 03 Bayes Filter).
What is free
The 22 playlist videos are free on YouTube. The playlist description points to lecture material on the Freiburg course page, which this record did not inspect.
Keep in mind
Lecture recordings from the winter 2013/14 SLAM course taught in Freiburg, per the playlist description. 22 videos; playlist last updated Apr 3, 2020.
Explore vehicle autopilot software through simulated sensor inputs and vehicle models.
Intermediate
Tutorial
Computer
First activity & requirements
Try first
Set up SITL, start one supported vehicle and inspect its command and telemetry workflow.
What you need
The guide documents native Linux and Windows environments, including a Windows Mission Planner simulation option. Follow the setup path for your chosen environment.
What is free
Public SITL documentation; no flight controller is needed for software-only simulation.
Keep in mind
ArduPilot covers aircraft, rovers and other vehicles. Choose the correct vehicle-specific example.
Original resource Access checked 2026-09-30 · English · Source reviewed
Learn localization, mapping and navigation for quadrocopters: visual motion estimation, filtering, bundle adjustment, SLAM, 3D reconstruction and planning.
Intermediate
Video lessons
Read or watch
First activity & requirements
Try first
Open the State estimation lecture (slides and video in the course schedule), then read Exercise Sheet 2 (Robot Localization) and its solution sheet.
What you need
Suggested: linear algebra and probability. The original exercises are typically in C++, start with a ROS introduction and ran on a Parrot Ardrone.
What is free
Videos are free on YouTube. The course page lists slide PDFs, exercise sheets, solution sheets, bag files and exercise source code.
Keep in mind
Summer Semester 2013, lecturer Jürgen Sturm. 13 videos (Lecture 11 in two parts); playlist last updated Apr 19, 2020. Exercises were built around ROS and a Parrot Ardrone.
Explore a ROS 2 stack and tutorials for coordinating small aerial robots.
Advanced
Tutorial
Physical robot
First activity & requirements
Try first
Begin with the setup and usage documentation, then inspect a tutorial's launch configuration.
What you need
ROS 2 and the selected Bitcraze flight hardware/positioning setup for physical experiments.
What is free
Public documentation and code; aircraft, radios and positioning equipment are separate.
Keep in mind
Match firmware and ROS versions. The provider reports intermittent unicast packet loss as a known firmware bug and marks advanced tutorials as under development.
Nitin J. Sanket and Chahat Deep Singh / University of Maryland PRG
Learn quadrotor autonomy: rigid body transforms, attitude estimation, Kalman filters, quadrotor dynamics and control, camera models, SfM and optical flow.
Advanced
Video lessons
Computer
First activity & requirements
Try first
Watch Class 2 Part 2 (IMU Basics, Attitude Estimation using CF and Madgwick), then do Project 1a, a Madgwick filter using the provided IMU and Vicon data files.
What you need
Provider lists Python 2.7 proficiency as a soft prerequisite and undergraduate linear algebra and calculus; quadrotor, ROS and Linux experience is a plus.
What is free
Videos are free on YouTube; the course page links slide PDFs and project pages. Assignments were submitted via ELMS and test sets released on Piazza.
Keep in mind
Fall 2019. 17 videos. Course software: Python 2.7 with ROS, OpenCV, NumPy, Scikit and Matplotlib on Ubuntu 16.04. Course projects were carried out on quadrotors in groups.
Study an aerial robotics simulator's installation, navigation and sensor-data workflow.
Advanced
Research reproduction
Computer
First activity & requirements
Try first
Follow the quickstart, then inspect how a simulated quadrotor exchanges data with the renderer.
What you need
C++/Python toolchain and the documented rendering/simulation dependencies.
What is free
Public instructions and code links; local hardware and component terms are separate.
Keep in mind
The inspected quickstart uses Python 3.6; its ROS route names older ROS/Gazebo pairings and protobuf 3.0.0. Compatibility with newer environments is unverified here.
Compare quadrotor control and reinforcement learning in a compact PyBullet simulation repository.
Advanced
Research reproduction
Computer
First activity & requirements
Try first
Run the PID example before comparing it with the single-drone hovering learning example.
What you need
The current instructions use Python 3.12 and warn that PyBullet must build locally beyond Python 3.10. Provider-tested platforms are listed; none were tested by us.
What is free
Public code and local simulation examples; no aircraft required.
Keep in mind
The current refactoring differs from the original IROS 2021 code. Use the paper branch when reproducing that publication.
Original resource Access checked 2026-09-30 · English · Source reviewed
Teaches step by step how to configure, program and operate the FTC Control System with Blocks op modes for a DC motor, a servo and sensors.
Beginner
Tutorial
Physical robot
First activity & requirements
Try first
After configuring your hardware and connecting to the Program & Manage server, do Creating Your First Op Mode, then modify it to control a DC motor and add telemetry.
What you need
Major parts: REV Control Hub and Driver Hub (officially supported; Android phones are unsupported alternatives), 12V battery, motor, servo, sensors, gamepad, laptop. Cables and adapters vary; see Required Materials.
What is free
The tutorial pages are on the official FIRST Tech Challenge Docs site. OnBot Java and Android Studio tutorials are listed alongside it.
Keep in mind
FTC Docs 0.3 documentation. Later pages cover controlling a servo, the color-distance sensor and the touch sensor.
Allison Okamura and Laura Blumenschein / Stanford University
Teaches students to design, make and test soft robots through weekly slides, six labs and a final project building a soft robot that benefits people.
Beginner
Build guide
Physical robot
First activity & requirements
Try first
Read Slides 1 and 2, then work through Lab 1, posted for Week 2 on the course page.
What you need
The course page states it requires high school physics (non-calculus-based) and a little familiarity with programming. Suggested: plan to buy lab materials.
What is free
Slides for weeks 1 to 8, six labs and the project description are linked from the course page. In-class attendance was required and auditors were not accommodated.
Keep in mind
Freshman Introductory Seminar by Allison Okamura, co-designed by Laura Blumenschein, with some materials shared by Elliot Hawkes (UC Santa Barbara). The page shows dates but no year.
Original resource Access checked 2026-10-05 · English · Source reviewed
Practice differential-drive control and sensor use through nine RedBot experiments, from motor tests to line following.
Beginner
Exercises
Physical robot
First activity & requirements
Try first
Read Experiment 1 and its wiring checks before attempting the basic motor test.
What you need
A matching RedBot with Shadow Chassis, Arduino IDE, SparkFun RedBot library, USB mini-B cable and four AA batteries. Some later experiments require the additional SIK sensors.
What is free
The complete experiment guide and example code are publicly readable. Robot parts and tools are separate.
Keep in mind
The guide marks both original kits retired. Confirm parts availability and match the chassis, wiring and kit variant before using this older guide; current IDE compatibility was not tested.
Original resource Access checked 2026-10-01 · English · Source reviewed
Learn basic robot assembly and Python control using a Raspberry Pi and a wheeled chassis.
Beginner
Build guide
Physical robot
First activity & requirements
Try first
Review the complete parts list, then follow the motor and chassis assembly steps.
What you need
The specified Raspberry Pi, motor controller, chassis, power and tools.
What is free
The written build guide is free; all physical components are separate.
Keep in mind
The inspected page lists the chassis as out of stock and several original parts as no longer stocked. Sourcing parts or checking substitutes is necessary before attempting this older guide.
Kevin Lynch, Nick Marchuk and Matt Elwin / Northwestern University
Explains PIC32 timers, interrupts, PWM and ADC, PID control, brushed DC motor physics, gearing and DC motor drive circuits.
Intermediate
Video lessons
Read or watch
First activity & requirements
Try first
Watch Intro to brushed permanent magnet DC motors, parts 1/2 and 2/2, then Driving DC motors, parts 1/3 to 3/3 (flyback diodes, PWM, H-bridges).
What you need
Sample code targets the PIC32MX795F512H on the NU32 development board. Suggested: basic C and circuits; the wiki page links a C crash course playlist.
What is free
The 56 videos are on YouTube, and sample code used in them can be downloaded from the NU32 page. The videos accompany the textbook Embedded Computing and Mechatronics.
Keep in mind
Short videos, mostly made with Northwestern's Lightboard; playlist last updated Sep 23, 2019. The NU32 Videos wiki page links a playlist for each chapter.
Learn the control workflow for a small robot arm through the manufacturer's setup and operation guides.
Intermediate
Build guide
Physical robot
First activity & requirements
Try first
Read the controller-launch and keyboard-teleoperation steps, checking their prerequisites first.
What you need
For physical ROS operation, the guide names a PC with a DYNAMIXEL Starter Set or OpenCR. Arm hardware is separate and the base plate is excluded.
What is free
The guide is free to read; arm hardware and power/control accessories are separate.
Keep in mind
Match the ROS distribution and hardware. The e-Manual announces migration to ROBOTIS Docs; this original guide was accessible at the check, but its replacement needs rechecking.
Learn how Poppy's Python interface connects a robot, reads sensors and sends motor commands, with a simulation option.
Intermediate
Tutorial
Computer
First activity & requirements
Try first
Read the Hello Poppy world quickstart and compare the real-robot and CoppeliaSim object-creation examples.
What you need
Python familiarity and Poppy libraries for your selected robot. Simulation requires local software setup; physical control requires a matching Poppy robot and connection drivers where applicable.
What is free
The upstream Python tutorial is publicly readable in its original GitHub repository. Hardware and simulator licensing are separate.
Keep in mind
Pinned documentation names Python 3.5+ and drops Python 2 and Anaconda recipes from pypot 4.0. It uses the legacy simulator='vrep' argument for CoppeliaSim; compatibility with current releases was not tested.
Teaches motors, mechanical modeling, manufacturing, sensors, kinematics and control while designing a ballbot that balances on a basketball.
Intermediate
Video lessons
Physical robot
First activity & requirements
Try first
Watch ROB 311 Fall 2022: Mechanical modeling, then read the course plan and labs in the ROB 311 GitHub repository (github.com/michiganrobotics/rob311).
What you need
Labs need a Raspberry Pi Pico, gearmotors, omni wheels, a basketball and other parts in the full bill of materials, plus a 3D printer, laser cutter and water jet.
What is free
The repository says all lecture videos are on YouTube, with lecture files, labs and a bill of materials on GitHub. The midterm is not publicly available.
Keep in mind
Fall 2022 first offering, 22 videos. Lab 1 (soldering) is not included in the repository and lab 5 is skipped.
Guides you through testing and configuring motor control with the Arduino SimpleFOC library, from sensor and driver checks to closed-loop torque control.
Intermediate
Tutorial
Physical robot
First activity & requirements
Try first
Do the four foundation steps in order: Test Your Sensor, Test Your Driver, Motor + Driver Test (Open-Loop), then Closed-Loop Control (Voltage-Based Torque).
What you need
Per the guide: SimpleFOC installed, familiarity with the library structure, and the hardware (motor, driver, position sensor, a board such as Arduino, STM32 or ESP).
What is free
Documentation is free to read, and the docs footer says SimpleFOC is distributed under an MIT license. Suggested: budget for the motor, driver and sensor.
Keep in mind
Each foundation step takes about 10 to 15 minutes. Optional advanced steps cover estimated and sensor-based current control; the theory corner covers FOC transforms, PID, low-pass filters and alignment.
Follow the practical sequence for assembling, configuring motors and calibrating an SO-101 arm.
Intermediate
Build guide
Physical robot
First activity & requirements
Try first
Read the motor-configuration and calibration steps before beginning assembly.
What you need
An SO-101 configuration, correct power supply, tools and a compatible host computer.
What is free
The guide is free; motors, printed parts, electronics, tools and other hardware are not included.
Keep in mind
The linked main documentation requires a source installation; the banner names v0.6.1 for stable pip users. Match the guide version and keep leader/follower configurations distinct.
Original resource Access checked 2026-09-30 · English · Source reviewed
Learn mechanical, electrical and software integration through JPL's six-wheel open rover build documentation.
Advanced
Build guide
Physical robot
First activity & requirements
Try first
Read the build roadmap and component list, noting the wiring-first sequence before mechanical assembly.
What you need
The provider estimates roughly $1,600 and at least 100 person-hours, plus tools including soldering equipment and a multimeter. These are upstream estimates, not current quotes.
What is free
Build documentation and source are public; the rover itself is not free.
Keep in mind
The project has multiple hardware generations. Follow one revision's parts and instructions rather than combining versions.
Original resource Access checked 2026-09-30 · English · Source reviewed
Start with an accessible introduction to robot learning and how it relates to classical robotics.
Beginner
Course
Read or watch
First activity & requirements
Try first
Read the introduction, then work through the released classical-robotics unit.
What you need
Basic Python; introductory algebra helps. A physical robot is not required for the reading sequence.
What is free
The released course units are free to read.
Keep in mind
At the source check, reinforcement learning, imitation learning and foundation-model units were marked Coming Soon. This is not a complete seven-unit course.
NVIDIA’s guided robot-learning path introduces Isaac Lab, trains a cartpole, adds a second robot and discusses sim-to-real transfer.
Intermediate
Course
GPU / workstation
First activity & requirements
Try first
Read the introduction, then use Train Your First Robot in Isaac Lab to configure and train the cartpole task.
What you need
Python and reinforcement-learning background are useful. The training lessons use Isaac Lab with Isaac Sim; the exact release-specific local hardware requirements were not verified in this course-page inspection.
What is free
Course pages are public; software execution requires suitable GPU hardware or cloud compute.
Keep in mind
The guided sequence covers introductory robot learning, cartpole training, a second robot and sim-to-real concepts.
Walk through NVIDIA’s G1 apple-to-plate reference workflow, from demonstration collection through GR00T 1.7 post-training, simulation evaluation and optional real deployment.
Intermediate
Course
GPU / workstation
First activity & requirements
Try first
Read the prerequisites and begin the independent simulation workflow with Isaac Lab-Arena setup.
What you need
Linux, Docker, intermediate Python, basic humanoid control knowledge and prior real-robot operation are assumed. Both paths use a supported teleoperation device and at least 48 GB NVIDIA GPU VRAM for GR00T 1.7 post-training. Simulation additionally needs an Isaac Sim-compatible RTX workstation; GPUs without RT Cores, such as A100 and H100, are unsupported. The real path needs a G1 with Dex3-1 hands and Jetson AGX Thor.
What is free
Course pages are public. The simulation and real workflows are independent; hardware and GPU or cloud compute are separate costs.
Keep in mind
NVIDIA estimates 3 to 6 hours for each independent simulation or real-robot workflow after prerequisites and hardware are available. The course targets a specific G1 configuration and apple-to-plate task.
Learn the workflow for collecting demonstrations, training a policy and evaluating it with LeRobot.
Intermediate
Tutorial
GPU / workstation
First activity & requirements
Try first
Read the imitation-learning walkthrough and identify its recording, training and evaluation stages.
What you need
Python; physical collection needs compatible hardware, while training needs suitable compute.
What is free
Documentation and library source are public. Robots, training hardware and cloud services are separate expenses.
Keep in mind
The linked main documentation requires installation from source; its banner points pip users to stable v0.6.1 at the recorded check. Match library, dataset and guide versions.
Teaches the foundations of reinforcement learning: MDPs, dynamic programming, model-free prediction and control, function approximation and policy gradients.
Intermediate
Video lessons
Computer
First activity & requirements
Try first
Watch RL Course by David Silver - Lecture 1: Introduction to Reinforcement Learning, then try the Easy21 assignment linked from the teaching page.
What you need
No prerequisites or system requirements are stated on the inspected pages. Suggested: probability, linear algebra and basic programming.
What is free
The 10 videos are free. The teaching page lists slides for 10 lectures, Easy21 and past exam questions, under CC-BY-NC 4.0 except some third-party content.
Keep in mind
UCL Advanced Topics 2015 (COMPM050/COMPGI13). The playlist description links an older UCL URL for materials; the current teaching page is on David Silver's WordPress site.
Teaches robot control and MDPs, imitation learning, RL, generative models, transformers, world models, generalist robot policies and embodied reasoning.
Intermediate
Video lessons
Computer
First activity & requirements
Try first
Watch Lecture 3: Imitation Learning, then follow the hw3_imitation_learning README setup and key configuration and do Exercise 1: MSE policy and Exercise 2: DAgger on the simulated SO-101 arm.
What you need
The hw3 README demonstrates a Python 3.12 environment with uv and permits other package managers; its evaluation runner assumes ARM macOS or Linux and suggests WSL, Docker or student PCs otherwise. Suggested: Python, PyTorch basics (hw1 is a PyTorch and Numpy tutorial), linear algebra and probability.
What is free
The course page links slides and a recording for each lecture, guest lecture recordings, and code for four tutorial exercises in the course GitHub repository.
Keep in mind
Spring 2026 offering (263-5911-00L); the playlist has 11 videos. Guest spotlight recordings are linked from the course page, not this playlist.
Covers imitation learning, policy gradients, actor-critic, Q-learning, offline RL, reward learning, multi-task and meta RL, exploration and RL for robots.
Advanced
Video lessons
Computer
First activity & requirements
Try first
Watch Lecture 2: Imitation Learning, then attempt Homework 1, released in the Imitation Learning week on the course page with a PDF, code and template.
What you need
Course page prerequisites: CS229 or equivalent, probability, multivariable calculus, linear algebra, some deep learning (assignments use PyTorch) and basic RL.
What is free
The 19 playlist videos are free. The course page links homework PDFs, code and templates. It says current-term recordings are on Canvas and assignments are submitted on Gradescope.
Keep in mind
Playlist videos are from Spring 2025 and include one tutorial session. The course page shows Spring 2026 homework and says the default project has been adjusted since Spring 2025.
Columbia / MIT / Toyota Research Institute authors
Inspect a published imitation-learning method through notebooks, experiment configurations, logs and checkpoints.
Advanced
Research reproduction
GPU / workstation
First activity & requirements
Try first
Start with the state-based notebook, then compare its configuration with a published experiment.
What you need
The introductory Colab notebook has hosted-runtime limits. Local simulation benchmark reproduction specifies Linux with an NVIDIA GPU; the macOS environment lacks full benchmark support.
What is free
Public reproduction code and notebooks; hosted runtimes and dataset terms remain separate.
Keep in mind
The repository includes both simulated and real-world workflows. Real-robot reproduction needs additional hardware and setup.
Original resource Access checked 2026-09-30 · English · Source reviewed
Explore accelerator-based robot-learning environments for locomotion and manipulation.
Advanced
Research reproduction
GPU / workstation
First activity & requirements
Try first
Load a provided environment and inspect its training configuration before starting a run.
What you need
Python 3.10+ and a supported GPU/backend environment for the documented training workflow.
What is free
Public environments and training examples; accelerator time is not supplied.
Keep in mind
The authors document GPU-precision effects on training stability and differences between the current and paper-reproduction training scripts. Use the relevant reproducibility notes.
Original resource Access checked 2026-09-30 · English · Source reviewed
Learn an imitation-learning experiment from dataset download through training and result inspection.
Advanced
Tutorial
Computer
First activity & requirements
Try first
Use the quickstart notebook or run the documented small experiment and inspect its logs.
What you need
Python and the documented environment. The example uses robosuite v1.5.1 for rollouts, or disables rollouts to run without it; larger visual-policy runs need more compute.
What is free
Public tutorial and code; Colab has its own account and resource limits.
Keep in mind
The tutorial introduces the workflow; reproducing published results requires matching datasets, configurations and compute.
Original resource Access checked 2026-09-30 · English · Source reviewed
Francesco Capuano, Caroline Pascal, Adil Zouitine, Thomas Wolf and Michel Aractingi
Connect classical control, reinforcement learning, imitation learning and generalist robot policies through a research tutorial with LeRobot examples.
Advanced
Book / notes
Read or watch
First activity & requirements
Try first
Read the classical-robotics and robot-reinforcement-learning sections, then compare their approaches to modeling and feedback.
What you need
Suggested background: Python, basic robotics and machine learning. Reading needs no robot; executing the paper's examples has additional library, data, hardware and compute requirements.
What is free
The full tutorial is publicly available as HTML and PDF on arXiv.
Keep in mind
This entry links the v1 arXiv text. It is a conceptual survey with practical examples, not an introductory programming course or a verified reproduction recipe; match code and dataset versions separately.
Original resource Access checked 2026-10-01 · English · Source reviewed
Learn how LeRobotDataset v3 stores episodes, video and metadata and exposes them for training.
Intermediate
Tutorial
Computer
First activity & requirements
Try first
Inspect a small dataset's metadata and load selected frames before building a training pipeline.
What you need
Python and a matching LeRobot release; the linked main guide requires a source installation. Dataset size determines storage and decoding needs.
What is free
Public format documentation and code examples; each dataset has its own access and reuse conditions.
Keep in mind
Release guidance is inconsistent: the main banner names stable v0.6.1 while the installation text calls v0.4.0 support forthcoming. Check documentation against your chosen release.
Original resource Access checked 2026-09-30 · English · Source reviewed
Inspect a recorded robot episode's camera streams, states and actions using LeRobot's dataset visualization tools.
Intermediate
Tutorial
Computer
First activity & requirements
Try first
Read the local visualization example for episode 0 of lerobot/pusht and identify the streams you would inspect in Rerun.
What you need
Python, a compatible LeRobot installation and enough storage for the selected dataset. Private or gated datasets require authentication; the example names a public dataset.
What is free
The visualization instructions are publicly readable; dataset access and reuse conditions vary. Local tools and an online viewer are linked.
Keep in mind
This activity focuses on inspecting episodes, separately from the existing entry on dataset structure and loading. The guide tracks the development documentation; viewer execution and hosted-service availability were not tested.
Original resource Access checked 2026-10-01 · English · Source reviewed
Construct a household simulation task by configuring a scene, objects, robot and task in BEHAVIOR/OmniGibson.
Advanced
Research reproduction
GPU / workstation
First activity & requirements
Try first
Read the installation requirements, then follow the quickstart's environment configuration.
What you need
The inspected requirements specify Ubuntu 22.04+ or Windows 10+, at least 32 GB RAM and 8 GB VRAM, with RTX 2070 as a minimum example. Follow the source installation instructions.
What is free
Public quickstart and code; simulator dependencies, asset terms and compute requirements are separate.
Keep in mind
At the recorded check, Docker installation was temporarily unavailable and PyPI installation was marked forthcoming. Simulation and task definitions are substantial; start with the quickstart before larger experiments.
New learning resources, build notes and parts updates by email.
How we use your email
We use your address only for Open Source Robotics emails. We do not sell the list. Your address, consent dates and subscription status are stored with Cloudflare; our mail server sends confirmation emails. Unconfirmed signups are removed after 30 days. Unsubscribed addresses are retained to prevent unwanted mail.
Newsletter links may count visits by issue and story, without storing your email or IP address with those counts.
Source access checked 2026-09-30 to 2026-10-13. This is a growing first edition, not every resource on the web.