Robotics Docs

Robotics, from first principles

Learn robotics in 10 short books.

The books come in 3 parts. The foundation builds the ground a robot arm stands on, from the maths and the camera to the frameworks and ROS. The techniques and models then explain the methods themselves, both the ones somebody wrote down and the ones fitted to examples. The last part takes one problem on a real table and follows it the whole way down. Every chapter comes from the robotics-basics docs and keeps their reading order.

3 parts10 books80 chapters354 sections120 h 25 min of reading
Start with Book 1

Robotics Foundation

Start here. These five books build the ground every robot arm stands on: the Python and the maths, how a camera turns the world into numbers, the simulators and frameworks the field is built on, ROS, the software the parts talk through, and how a neural network works, from a single neuron worked out by hand up to the models that drive an arm today.

Book 1

Robot Arm Basics

Python, maths, frames, kinematics and arm types

Start here if you know nothing about robots. The Python and NumPy that robot code is written in, the angles, vectors and matrices an arm needs, frames and transforms, forward and inverse kinematics, and the common kinds of robot arm, with the six-joint arm in detail.

  1. Python and NumPy
  2. Angles, Vectors and Matrices
  3. Frames, Position and Transforms
  4. Forward and Inverse Kinematics
  5. Joints and Types of Arm
5 chapters · 10 sections · 2 h 50 min
0%
Book 2

Perception

Cameras, depth and finding objects

How a camera turns the world into pixels, how to turn pixels back into points, and how a robot finds an object and measures its size and pose, with the techniques, models and licences compared.

  1. Cameras
  2. Object Perception
2 chapters · 14 sections · 6 h 55 min
0%
Book 3

Frameworks & Manipulation

MuJoCo, Gazebo, MoveIt and training arms

The tools and simulators the field is built on, MuJoCo and Gazebo among them, and what you do with them: gripping, moving an arm, programming and training one or two arms, worked case studies, and what changed at the frontier in 2026.

  1. Tools and Libraries
  2. Gripping
  3. Arm Movement
  4. Programming and Training One Arm
  5. Two-Arm Training
  6. and 3 more
8 chapters · 41 sections · 19 h 45 min
0%
Book 4

ROS and RViz

The robot software stack and its 3D viewer

What ROS is and how its programs talk to each other, one small program per idea, then a camera, an arm and the two together. It ends with RViz, the 3D viewer that shows you what the robot thinks is happening.

  1. ROS Basics
  2. RViz and the 3D View
2 chapters · 6 sections · 1 h 15 min
0%
Book 5

Neural Networks and AI Models

How a model works and how it is trained, from nothing

For a reader who knows some maths, some science and some programming, and nothing at all about machine learning. It starts with why some jobs cannot be written as rules, works one neuron out by hand, and builds up through layers, loss, gradient descent and backpropagation to the transformer, which nearly every model in use now is made of. It then explains each family in turn: pretraining and fine-tuning, diffusion and flow matching, models that see, language and vision-language models, learning from outcomes, the policies that move an arm, and world models. It ends with what it takes to run a model on a real robot and judge honestly whether it works.

  1. What Learning From Data Means
  2. Inside a Neural Network
  3. How Training Works
  4. Making Training Work
  5. Turning the World Into Numbers
  6. and 9 more
14 chapters · 59 sections · 24 h 15 min
0%

Techniques and Models

The methods themselves, each one explained on its own. The first book holds the techniques somebody wrote down, and the second holds the models that were fitted to examples instead. Both say where a method is the right tool and where it is not.

Book 6

Programming Techniques

The algorithms robot arm software is built from

The written, language-independent techniques behind seeing, planning and moving: camera geometry, pose from points and calibration, matching and registration, least squares, RANSAC and filters, masks, clustering and 3D maps, sampling-based planning, inverse kinematics and MPC, PID, dynamics and safety monitoring, and state machines and behaviour trees. Each chapter puts the most used techniques first. Each page says where a technique is used on an arm, where it fails, and which library already does it.

  1. What Techniques Are
  2. Geometry and Cameras
  3. Searching and Matching
  4. Fitting and Estimation
  5. Image and Point Cloud Processing
  6. and 3 more
8 chapters · 45 sections · 17 h 35 min
0%
Book 7

Learned Models

Every kind of learned model a robot arm uses, in plain words

What a model is, how it learns and where its data comes from, for a reader who has never met one. Then classical machine learning, from regression and decision trees to Gaussian processes and movement primitives, and seven families of neural network models: models that see, that work in 3D, that choose a grasp, that move the arm, that understand words, that predict what happens next, and that make sense of touch and force. It ends with fine-tuning, running, testing and trusting a model on a real arm.

  1. What Models Are
  2. Classical Machine Learning
  3. Seeing Models
  4. 3D Models
  5. Grasp Models
  6. and 5 more
10 chapters · 62 sections · 33 h
0%

Robotics by Example

Three problems from one work cell, each followed the whole way down. Instead of explaining a method and then showing a use for it, these books start with a table, a camera and a job to do, build six quite different answers to the same job, and measure all six against one another with the same examiner. The first book is about seeing, the second is about moving, and the third removes the help the second was given.

Book 8

Seeing the Glasses

One perception problem, answered six ways and measured

Several glasses of one kind stand on a table, and the arm has to work out which pixels belong to which glass. This book follows that single question the whole way down: what the camera can and cannot see, why a glass standing behind another is in no picture at all, and six quite different ways to answer it. One is a written rule with nothing fitted to anything, one is a small network trained in this cell from nothing, and four start from weights somebody else fitted to photographs of the everyday world. All six are built, and all six are scored by the same examiner, so the differences between them are measured rather than argued. The programs are in the repository, and each solution names the folder it was built in.

  1. The Cell Everything Happens In
  2. The Problem
  3. The Examiner
  4. The Six Solutions
  5. Solution 1: Rules on the Table
  6. and 7 more
12 chapters · 47 sections · 5 h 30 min
0%
Book 9

Pushing the Glasses Apart

One manipulation problem, answered six ways and measured

The glasses stand too close together for the gripper to reach between them, so the arm has to drag them apart without knocking any of them over. This book answers that six ways, from a single fixed nudge with nothing learned at all, through a model that ranks pushes the geometry proposed, to a borrowed robot foundation model fine-tuned on this cell. All six are built and scored by one examiner, and the result is worth the reading: nothing learned beats the written rule here, the one learned method that wins is the one that learned the quantity the geometry gets wrong, and every method that learns something topples at least one glass where the geometry topples none.

  1. The Problem
  2. The Seven Solutions
  3. Solution 1: One Fixed Nudge
  4. Solution 2: Geometry Generates, a Model Ranks
  5. Solution 3: Imitation From Demonstrations
  6. and 5 more
10 chapters · 48 sections · 6 h 55 min
0%
Book 10

Clearing the Table

The same pushing problem, with nobody helping the arm

The task of the pushing book with two helps removed. The camera above the table is gone, so the arm carries its own camera and has to travel to see. Nothing counts its pushes or tells it when to stop, so it must say itself when the table is done. The same six answers are run again, with fifteen arm minutes for a queue of fifty tables. This book covers only what changed: why looking became most of the work, why every learned answer at first broke glasses or froze, the shared arithmetic that made them safe, and how a learned world model came to lead.

  1. The Problem
  2. What Every Solution Had to Add
  3. Solution 1: One Fixed Nudge
  4. Solution 2: Geometry Generates, a Model Ranks
  5. Solution 3: Imitation From Demonstrations
  6. and 4 more
9 chapters · 22 sections · 2 h 30 min
0%

How the books are organised

1

Parts and books

Each book covers one broad area, and the books are grouped into parts. Read them in order, or go straight to the one you need.

2

Chapters

Each chapter is one topic from the docs, such as ROS, cameras or gripping, and keeps its original order.

3

Sections

Each section is one document. The site marks the sections you have opened, so you can pick up where you stopped.