Front Matter
Reader’s Guide
Everything you need before the first chapter: who the book is for, how it is built, and the fastest path for your situation. Skip nothing in here if you are new to the field; skim it if you already know your way around. You do not have to read all 43 chapters: the book has a canonical sequence, and this guide gives you the faster paths through it.
Who this book is for
You do not need to be a roboticist. It assumes you are comfortable with software and basic math, and that you would rather understand a thing than memorize it. It is written for three readers in particular:
- Developers building localization or navigation stacks who keep hitting the same wall because the map underneath is wrong.
- GIS and surveying professionals moving into high-precision, low-latency work where a few meters of error is no longer acceptable.
- Engineers working with Autoware (or any autonomy stack) who need to debug a map, not just consume one.
How to read it
A robot moving through a place has to answer four questions: Where am I? What is around me? Where can I go? What am I allowed to do? The whole book is organized around giving you the mental model to answer those questions, and then the tools to build the map that encodes the answers.
To keep it concrete, every concept lands on the same place: Harbor Yard, a small logistics depot with a row of loading docks, a one-way loop, a gate, and a stop line a human driver long ago decided to ignore. Harbor Yard is small enough to hold in your head and complicated enough to break a naive map. When a format or a rule finally clicks, it usually clicks there.
One structural choice worth knowing up front: the tool comes late. Veer Studio, the practical laboratory, sits in Part VI, deliberately after you understand the domain. You should know what you are doing before you worry about which button to click.
Two stretches of the book will tempt you to skim. Do not. Part II, the coordinate systems, looks like a math detour; it is the reason real maps land shifted, rotated, or a hundred meters from where they belong, and you cannot debug a map you cannot locate. Part V’s regulatory elements look like data entry; they are the map’s entire legal soul. The mistake you are about to make is almost always hiding in one of those two places.
How the book is organized
Front matter frames the book and back matter closes it. This guide and the Preface come before the first chapter; the Glossary, the Index, and the Colophon come after the last. The main body moves from the conceptual to the practical in eight Parts.
| I | The Big Picture. What an HD map is, why ordinary maps fail robots, the many representations of one place, and the pipeline from physical world to autonomous machine. |
| II | Coordinate Systems. Frames, reference systems, and geographic and projected coordinates: the foundation underneath everything, and the reason a map can land shifted, rotated, or a hundred meters from where it belongs. |
| III | Scanning Your Site. Turning physical reality into data: LiDAR, point clouds, capturing a site, SLAM, and localization. |
| IV | From Measurements to Meaning. Turning measurements into meaning: from geometry to semantics, through OpenStreetMap and GIS fundamentals, to why the same object lives in several representations at once. |
| V | Rules of the Road. Where the autonomy map begins: Lanelet2 primitives and regulatory elements, routing and topology, intersections, traffic rules, and what separates a drawing of a road from a road a robot can plan through. |
| VI | Building HD Maps With Veer Studio. The practical workflow from end to end: capture, align, draw, annotate, connect, validate, export. |
| VII | From Map to Autonomy. The map becomes executable: how Autoware loads it, localizes and perceives against it, plans a path through it, and how to debug it when the robot does not behave. |
| VIII | Building Autonomous Places. Beyond roads: warehouses, campuses, farms, and controlled sites, multiple robots sharing one place, and the thesis that a map can become spatial infrastructure. |
Reading paths
Different readers need different paths. Pick the one that fits where you are.
New to everything →
Start at Part I and read Parts I through IV in order. These build the mental model from the ground up. Do not skip the Reader’s Guide and the Preface, and expect to reread the first two chapters after Part III: they mean more once the machinery is in view.
You will learn: how robots represent places, how coordinates and sensors fit together, and the shape of the whole field before the tools appear.
I know robotics but not HD maps →
Jump to Part II for coordinates and reference frames, then Part VI to see the full mapping workflow. Come back to Parts III through V when a specific concept needs the depth.
You will learn: how the map layer fits into an autonomy stack, and the full mapping workflow in Part VI.
I use Autoware and want to build maps with Veer Studio →
Part V grounds you in Lanelet2 primitives, then Part VI walks through the full mapping workflow in Veer Studio. Part VII is your debugging manual when the robot misbehaves.
You will learn: how a map is constructed, validated, and consumed, from Lanelet2 primitives through the Veer Studio workflow.
I am looking for one term →
The Index lists the terms this book considers noteworthy, each linked to the chapters that explain it. It exists for the moment when you know the word but not the way to it. The Glossaryholds the definitions those chapters lean on, so a word you meet mid-book is one click from its canonical meaning.
You will learn: where any term is defined and which chapters explain it.