🚁 DroneRL

DroneRL is a reinforcement learning environment in which agents can be trained for optimal package delivery. This environment is implemented in both PyTorch and JAX.

AMLD 2025 Workshop: Training models in JAX: JIT all the things!

Resources for the participants of the AMLD 2025 workshop.

🥇 Public Leaderboard on AIcrowd

Part 1: Jax Intro

• Slides
• Notebook

Part 2: RL Crash Course

• Slides
• Notebook

Part 3: DroneRL in JAX

• Slides
• Notebook

Credits

Part of this work was supported by the EPFL Extension School and AIcrowd.

This more optimized version of DroneRL was implemented by @MasterScrat and @mar-muel, but over the years many people have contributed:

• @spMohanty
• @pacm
• @metataro

License

• 16ShipCollection by master484 under Public Domain / CC0
• Inconsolata-Bold by Raph Levien under Open Font License
• Press Start 2P by Cody Boisclair under SIL Open Font License 1.1