We started this book with a simple question: what is a Node in ROS 2? Nineteen chapters later, you now know how a complete industrial robot — from concept to real hardware — is built. Before looking at the final architecture, let's review the path once more:
Each chapter built on the previous one: without TF2 you couldn't understand RViz; without Gazebo and ros2_control, Nav2 wouldn't have made sense; without Odometry and SLAM, Navigation2 would have had nothing to work with; without Lifecycle Nodes, Nav2's architecture would have stayed incomplete; and without this final chapter, all of this would have remained mere theory rather than a real, safe product.
And this core architecture is surrounded by the supporting layers below, which we built across the middle and final chapters:
| Input | Feeds into |
|---|---|
| LiDAR | SLAM / AMCL / Costmap |
| IMU | robot_localization |
| Camera | Perception / Visual SLAM (if present) |
| Safety MCU | Watchdog / E-Stop / Brake |
| Fleet Manager | Multi-robot missions |
| Docker + CI/CD | Repeatable deployment |
| Layer | Related Chapter | Responsibility |
|---|---|---|
| 1. Identity and basic communication | Chapters 1-3 | Node, Topic, Service, Action, Parameter |
| 2. Body and space | Chapters 4-7 | URDF/Xacro, TF2, RViz, Gazebo |
| 3. Control and motion | Chapters 8-9 | ros2_control, Odometry, EKF |
| 4. Environment awareness and path planning | Chapters 10-13 | SLAM, Nav2, Behavior Tree/Lifecycle, MoveIt 2 |
| 5. Scalability | Chapter 14 | Multi-Robot, Fleet Manager |
| 6. Software delivery | Chapter 15 | Docker, CI/CD |
| 7. Hardware and industrial safety | Chapters 16-19 | Jetson/RPi/ESP32, CAN/EtherCAT, Sim-to-Real, QoS/DDS/Diagnostics/Safety |
On a real team, it's rare for one person to build all these layers. A software engineer focuses on layers 4 and 5, a mechanical engineer on layer 2, an electronics engineer on layer 7, and a DevOps engineer on layer 6. This book deliberately took you through layers 1 to 7 so you can hold a shared conversation with each of these roles — exactly what's expected of a good robotics engineer or product manager.
Now let's see all these concepts play out in a real, complete scenario:
| Step | Event | Related Chapter |
|---|---|---|
| 1 | ARCHO powers on; Bringup runs in order: power, MCU, drivers, sensors, Hardware Interface | 16, 18 |
| 2 | Lifecycle Manager activates Map Server, AMCL, Planner, and Controller in sequence | 12 |
| 3 | Fleet Manager assigns the mission "go to shelf 24" to ARCHO | 14 |
| 4 | The Behavior Tree hands the mission off to the Planner and Controller | 11, 12 |
| 5 | A worker crosses in front of ARCHO; the Local Controller corrects the path | 11 |
| 6 | A forklift blocks the aisle; the Behavior Tree requests replanning | 12 |
| 7 | ARCHO reaches the shelf; the MoveIt arm picks up the box using IK and a collision-free path | 13 |
| 8 | Throughout, the Safety Supervisor filters commands and Diagnostics reports system health | 19 |
| 9 | Battery runs low; the Behavior Tree cancels the mission and sends ARCHO to the charger | 12 |
| 10 | rosbag2 has recorded this entire scenario for later analysis | 18 |
This book was a complete and coherent roadmap, but real industrial robotics always runs deeper than any single book. A few natural directions to keep learning:
| Area | Why it's worth learning |
|---|---|
| Advanced Perception (Point Cloud, Visual SLAM, Object Detection) | For robots that need to recognize unknown objects |
| Docking and automatic charging | Without this, a fleet isn't truly autonomous |
| Cybersecurity (SROS2) | Once ARCHO connects to the corporate network, security is no longer optional |
| OTA (over-the-air updates) | For managing dozens or hundreds of ARCHOs without a physical trip |
| Precise time synchronization | For coordinating multiple sensors and multiple robots with millisecond accuracy |
| Industrial observability (Prometheus/Grafana) | For managing a large fleet at real-world scale |
If you want to keep going down this path, the most logical next step is to actually build a small version of ARCHO yourself — even with a simple two-wheeled robot and a Raspberry Pi — and implement Chapters 1 through 19 on top of it. Nothing replaces actually building the real thing.
A real industrial robot is a collection of correct decisions at every layer — from correctly naming a TF frame to correctly designing a hardware safety layer; and ARCHO showed that ROS 2 is exactly the framework that turns these decisions into a coherent, reliable, and scalable system.
Project ARCHO grew from a simple idea ("Archo the warehouse robot") into a complete industrial architecture: a body engineered with Xacro, alert senses from TF2 and SLAM, a decision-making brain with Nav2 and Behavior Tree, smart hands with MoveIt, a resilient network with CAN, and a guardian spirit that never allows danger.