multi-agent (path finding) planning framework

Mapf is a (currently experimental) Rust library for multi-agent planning, with a focus on cooperative path finding.

This repo is just getting started, so please bear with us for a while as we work towards making useful tools.

This is being developed as part of the Open-RMF project which provides an open source framework for enabling interoperability between heterogeneous fleets of mobile robots, including cooperation across different platforms and vendors.

Helpful Links

• Rust Book

Install dependencies

Ubuntu / MacOS / Linux / Unix

We tend to always target the latest versions of Rust, so we recommend using the rustup tool: https://www.rust-lang.org/tools/install

$ curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

If you already have rustup then you can use this to bring your installation up to date:

$ rustup update

Windows

Follow instructions for installing rustup-init.exe here.

Run an example

From the root directory:

$ cargo run --release --example grid

This example will allow you to experiment with multiple agents cooperatively planning from their starts to their goals in a grid environment. Use left/right click to set the start/goal cells of each agent. Use shift+left/right click to add/remove occupancy from cells. The velocities and sizes of each agent can be customized independently, and you can save/load scenarios into yaml files.

Some premade scenarios can be found in the mapf-viz/scenarios folder. Load a scenario on startup by passing the scenario name as an executable argument, e.g.:

$ cargo run --release --example grid -- mapf-viz/scenarios/sizes.yaml

Note that the scenario filename to load at startup must come after a -- with a space on each side.