Why drawtonomy — a whiteboard built for driving scenarios

drawtonomy is a whiteboard built specifically for driving scenarios. Most teams sketch these diagrams today in generic drawing tools or slide decks — those work fine for general shapes, but they don’t know what a lane is, so the geometry has to be redrawn whenever the road turns, the intersection grows a leg, or a crosswalk needs to line up with the road.

This page explains the design choices that follow from leading with “whiteboard for driving scenarios” rather than “tool that exports to a simulator”.

The problem it’s built around

Most actual autonomous-driving communication happens through diagrams: in papers, design reviews, planning meetings, incident write-ups, classrooms, and slide decks. The diagram is the artefact people look at, argue about, and remember.

Generic drawing tools at that level only give you generic shapes. A lane is a rectangle you redraw every time the road turns; a crosswalk is a stack of rectangles you keep aligning by hand; an intersection is a half hour of fiddling. Worse, the moment the road geometry changes — and on AV work it changes constantly — you start over.

drawtonomy exists to make that loop fast. The building blocks the domain actually has — lanes, intersections, crosswalks, traffic lights, road markings, vehicles, pedestrians — are first-class shapes, so the figure stays correct as you iterate.

Where drawtonomy sits

Driving-scenario work happens at a few different levels:

Diagrams. Papers, slides, whiteboard sketches, design-doc figures, classroom material. Fast and easy in principle, but in a generic tool the road geometry has to be rebuilt every time something moves.
Authoring tools. OpenSCENARIO editors, road-network editors, CAD-style packages. Precise, slow, expensive to learn.
Simulators. esmini, CARLA, in-house tools. Run the scenario, produce data.

drawtonomy lives at level 1, and crosses into level 2 when you need to: import a Lanelet2 map, sketch changes, export OpenDRIVE/OpenSCENARIO, hand the result to esmini.

Design priorities

Whiteboard-first

The point of comparison is a quick whiteboard or slide-deck sketch, not a CAD tool. That sets the bar for friction: open a URL, draw, share. No install, no account, no project file format. Anything that would make drawtonomy feel heavier than a quick sketch gets cut.

Topology-aware

A road is not a bag of polylines. drawtonomy models lane connections (Next / Previous / Left / Right) so that moving a boundary updates neighbouring lanes automatically. Two lanes that share a boundary share the same boundary points — drag once, both move. See Lane connection model.

Driving-domain templates

Vehicles (sedan, bus, truck, motorcycle…), pedestrians (walking, simple), traffic lights for vehicles and pedestrians, crosswalks, road markings, signs, intersection templates. They’re built-in shapes rather than generic-rectangle approximations. Custom SVG templates can be added by PR.

Editable on the way out as well as in

Every output format drawtonomy produces preserves enough state to be re-edited. drawtonomy.svg is the lossless canonical form: a regular SVG that previews everywhere (browsers, GitHub, slide decks, paper figures) and reopens in drawtonomy with every connection and overlap relationship intact. Nothing is trapped in a format you can’t read back.

Headless when needed

The exporter and parser code is part of @drawtonomy/sdk and runs without the editor. CI pipelines, browser extensions, and AI tools can generate and validate scenes programmatically.

Bridges to the rest of the workflow

Once you have a diagram, you usually want to do something with it. drawtonomy ships several bridges so the figure doesn’t stay locked inside the editor:

drawtonomy.svg — the default. Embed in papers, slides, Markdown docs; reopen later to keep editing.
Lanelet2 round-trip — open a Lanelet2 OSM map (including Autoware sample maps), edit, export back. Useful for sketching changes against an existing HD map.
ASAM export — OpenDRIVE 1.8 + OpenSCENARIO 1.3, optionally bundled as an esmini-ready zip.
AI Scene Generator — describe a scenario in natural language, or paste OpenSCENARIO XML, and get an editable canvas to start refining from.

These bridges are useful, but the diagram itself is the reason drawtonomy exists. A figure in drawtonomy is already valuable as a figure; these formats let it flow into the next stage of the workflow when needed.

What drawtonomy is not

Not a simulator. It doesn’t run scenarios. Export to esmini, CARLA, or your own tool for that.
Not a CAD tool. It doesn’t enforce engineering accuracy (clothoid splines, banking, elevation). Geometry is straightforward 2D.
Not a real-time collaboration suite. It’s a single-user editor. Save, share, reopen.