← all builds

From-Scratch Build · Technical Writing

Robot-Arm Operator's Handbook

Getting a six-axis robot arm to do anything is mostly a documentation problem — drivers, errors, frameworks and APIs scattered across a dozen places. This build is an eleven-lesson handbook I wrote from scratch that takes a complete beginner from "it won't install" to driving two arms at once.

ROSMoveItRViz GazeboPython API6-axis arm

What it is

The missing manual

A modern robot arm comes with enormous capability and almost no on-ramp. The official material is thorough but scattered, and a newcomer drowns before the arm ever moves. This handbook is the path I wished existed: a single, ordered course that assumes nothing.

It's structured as eleven progressive lessons, each a self-contained, step-by-step guide. By the end, a reader has installed the software, survived the common errors, set up a workstation, and learned to move the arm through simulation, code, recordings, a joystick, vision and even a second arm.

The core idea I wanted to learn: writing good documentation is engineering. Sequencing the lessons so each one only depends on what came before — and putting "Errors" second, right after Installation — is a design decision that saves a beginner hours of frustration.

The toolchain it teaches

What a reader meets along the way

The handbook is a tour of the robot-arm software ecosystem. Here is what each major tool is for.

middleware

ROS

The messaging backbone that ties drivers, planners and tools together. Installing and understanding it is the first hurdle the handbook clears.

planning

MoveIt

Motion planning. The lessons walk through both Python and C++ demos for getting the arm to plan and execute paths.

visualisation

RViz

The 3D viewer for seeing the arm's state, frames and planned motions before committing them to hardware.

simulation

Gazebo

A physics simulator so the arm can be tested in software, with no risk to the real hardware.

scripting

Python API

The functions that let a reader write their own code to command the arm directly.

perception

Vision package

A lesson on the computer-vision package for perception-driven tasks like locating and acting on objects.

The curriculum

Eleven lessons, in order

The sequence is the whole point — each step assumes only what the previous ones taught.

  1. Installation & errors covered

    Full and remote (Raspberry Pi) install, then a guide to the usual errors and how to fix them.

  2. Setup & general knowledge covered

    Preparing the workstation, first commands, and the background needed before anything complex.

  3. RViz & MoveIt covered

    Visualising state in RViz, then planning and executing motion with MoveIt in Python and C++.

  4. Python API covered

    Learning the library functions to write your own scripts that move the arm.

  5. Record, play & joystick covered

    Guiding the arm by hand with torque off, recording the motion, replaying it — and driving it live with a joystick.

  6. Perception, Gazebo & multiple arms covered

    Vision-driven tasks, simulation in Gazebo, and finally coordinating two arms at once.

How to use it

A path, not a reference

The handbook is meant to be read in order the first time and dipped into afterwards:

In my rebuild the focus was the on-ramp: nailing the installation and error lessons so a reader gets a moving arm on day one and never loses momentum.

Reflection

What writing it taught me