abstreet/docs/roadmap.md

# Roadmap

A/B Street has been under active development since June 2018. That's a long time
-- what work is happening now and how can you contribute?

## Current blockers before a "playable" release

- The traffic model handles complex map geometry poorly. Permanent traffic jams
  unrealistically form, preventing a full day's worth of traffic from running on
  many maps.
- Better scoring, so people understand how their changes affect different agents
- Challenge/story modes structured like a game. We have ideas, but need game
  design help to flesh these out.

## Ongoing work

If I had resources to hire a team, this is roughly how I'd organize different
roles. If you're interested in helping, these aren't strictly defined positions,
just ideas of related tasks.

### Map data / GIS

Support more cities:

- generalize the import pipeline (mostly done)
- manage if everybody downloads all cities or not
- write docs/tools to help people add new cities without programming experience
- add support for non-OpenStreetMap input: GeoJSON for parking in Perth, other
  trip demand sources, etc
- fix bugs for driving on the left side of the road

Improve the quality of map geometry derived from OpenStreetMap:

- try new algorithms to generate intersection polygons
- make the map editor easy to use and handle evolving OSM data
- use ML and lidar/satellite data to get extremely accurate curb / planter /
  sidewalk geometry

Build tools and organize community mapping:

- organize an effort to map how traffic signals are timed (partly started)
- divide and track work for distributed mapathons

Bring in new data to understand more about cities:

- PM2.5 pollution
- Tax / land value (is there inequitable access to transit?)

### Simulation / modeling

Totally new areas:

- light rail
- shared bike/pedestrian paths
- ridesharing
- micromobility (scooters, floating bikeshare)
- more score functions (elevation gain, biking safety)
- generating trip demand / activity models from scratch or modifying existing
  ones

Improve existing models:

- overtaking / lane-changing
- pedestrian crowds
- instant vehicle acceleration
- pedestrians walking on road shoulders (some streets have no sidewalks)
- buses: transfers, proper schedules, multiple buses per route

### UI and data visualization

We've got a UX designer, but implementing all of the new designs takes time.
Also:

- minimap camera controls are notoriously hard to get right
- refactor and clean up the GUI library for other Rust users
- lots of data viz design / implementation needed

### Game design

- the tutorial mode needs attention
- many ideas for challenge/story modes, but playtesting, tuning, and game design
  needed

### Web

A/B Street runs on the web via WASM and WebGL; just waiting on vector text
support. That unlocks:

- Share community proposals online, discuss them, vote, etc
- All of the infrastructure to do this

## Contributing for non-programmers

I've heard many people want to help with something other than programming or
design. The best ideas are to start mapping, especially since most work is
directly through OpenStreetMap:

- sidewalks, crosswalks, on-street parking
- traffic signal timing (needs more planning / tooling)
- fixing geometry problems with the map editor (needs tooling)

Playtesting by attempting to implement real proposals would also be helpful, to
expose where it's awkward for A/B Street to edit the map and to write up
problems encountered.

I'd like to give more specific guidance / tooling, but that in itself takes time
to organize.

## Long-term vision

Longer term, I'd like to take lots of the work in generating and interacting
with high-detail OpenStreetMap-based maps and generalize it, possibly as a new
OSM viewer/editor.

More generally, I'd like to see how simulation can help individuals understand
and explore other policy decisions related to cities. Domains I'm vaguely
interested in, but not at all knowledgable about, include land-use / zoning,
housing, and supply chains. In late March 2020, a new collaborator started a
pandemic model using the existing simulation of people occupying shared spaces.
What are other domains could benefit from the rich agent-based model we're
building?