Your Journey To Physical AI

Welcome to Back To Engineering! This is your roadmap from curiosity to building intelligent robots. Whether you're coming from software, AI/ML, or starting fresh, this tech tree organizes everything you need to learn robotics and physical AI in 2026.

How This Works

This isn't a traditional course - it's a choose-your-own-adventure learning path:

• Click any topic card to explore concepts, watch tutorials, and find resources
• Follow the arrows - they show prerequisites (what you need to know first)
• Star-shaped nodes are hands-on projects where theory meets reality
• Color-coded by discipline - Core #7aa2f7, Electronics #f5b942, Mechanical #93a8b4, Programming #5fbf82, Data Science #b48ce6, AI & ML #f26d6d
• Scroll to zoom and drag to pan around the graph
• Hover over nodes to see prerequisite connections glow

Start with Curiosity on the left and work your way right through the eras. Or jump to what interests you - the graph shows what knowledge you'll need.

The Philosophy

My approach mixes practical building with industry best practices. You'll start simple (blinking an LED, cardboard robot arms) and progressively add complexity (ROS 2, AI-driven manipulation, VLAs). Each layer teaches you why we use professional tools, not just how.

By the end, you'll understand the full stack: from servo control to vision-language-action models deployed on real hardware.

Featured Project

Check out the Robot Arm Project to see the full journey: 7 episodes from scratch-built cardboard prototypes to AI-powered LeRobot arms trained on custom datasets.

---

Robotics & AI Tech Tree

A static learning site: topics and hands-on projects arranged in a dependency tree. Built with MkDocs Material, Cytoscape.js, and one small Python script. Deploys to GitHub Pages on every push.

Local development

uv sync                                # install dependencies
uv run python scripts/build_graph.py   # regenerate docs/assets/graph.json
uv run mkdocs serve                    # live preview at http://127.0.0.1:8000

Adding content

1. Create a Markdown file in docs/topics/ or docs/projects/.

2. Fill in the frontmatter:

   ---
   id: slam                       # unique, stable, never change it
   title: SLAM
   type: topic                    # topic | project
   category: ai-ml                # drives node color on the tree
   tree_icon: map-2               # Tabler icon name (tabler.io/icons), optional
   prerequisites: [ros2-basics, linear-algebra]
   video: https://youtu.be/...    # optional, counted on the node card
   contributors:                  # optional, renders a "Contributed by" callout
     - username: github-user
       url: https://github.com/github-user
       pr: https://github.com/owner/repo/pull/123
   ---
   

3. Add the page to nav: in mkdocs.yml.

4. Run uv run python scripts/build_graph.py (CI also runs it on push).

5. Push. Done.

The build fails loudly if a prerequisite id has a typo, an id is duplicated, or the graph contains a cycle.

The tree page

docs/tree/index.html is a standalone full-page renderer (no MkDocs theme): ELK computes the layered layout and orthogonal edge routes, nodes are HTML cards, edges are one SVG layer, pan/zoom is custom. MkDocs copies it verbatim to /tree/. It reads assets/graph.json, so content changes flow through automatically; only design changes touch this file. Category colors live in the :root CSS variables at the top of the file.

Design notes

• Node ids are the stable identity of every topic/project. Progress tracking (phase 2) will store sets of these ids, first in localStorage, later per account. Never rename an id once published; add a new node instead.
• The graph renderer lives in docs/js/tech-tree.js and only activates on pages containing <div id="tech-tree">.
• Cytoscape and dagre load from CDNs (see extra_javascript in mkdocs.yml). Vendor them into docs/js/ if you prefer no CDN.

One-time GitHub Pages setup

Repo Settings -> Pages -> Source: "GitHub Actions". Then update site_url in mkdocs.yml to your real URL.

Analytics

The site supports Umami analytics without requiring credentials for local builds. Create a site in Umami, then add these GitHub repository variables under Settings -> Secrets and variables -> Actions -> Variables:

UMAMI_SCRIPT_URL=https://your-umami-host.example/script.js
UMAMI_WEBSITE_ID=your-website-id
UMAMI_DOMAINS=www.backtoengineering.com,backtoengineering.com

When those variables are present, the deploy workflow injects the Umami tracker into the built pages. The site also sends custom events for tree navigation, next-node clicks, external links, video interactions, and tree node detail opens.

License

This project uses two licenses:

• Code (site, tree renderer, build scripts): Apache License 2.0
• Content (the learning map, node and project descriptions, documentation): Creative Commons Attribution-NonCommercial 4.0 (CC BY-NC 4.0)

In short: you're free to use, adapt, and build on this project for non-commercial purposes with attribution. Commercial use of the content is not permitted. The curated links point to third-party resources, which remain under their own licenses.

First-time contributors sign a CLA on the PR (one click, you keep ownership). Why: the map stays free and open source forever; the CLA keeps future paid features like progress tracking possible without chasing every past contributor.