# google-deepmind/android_env

RL research on Android devices.

Repository: https://github.com/google-deepmind/android_env
Canonical: https://ross.abutalabs.com/products/android_env
Language: Python
License: Apache-2.0
License Family: permissive
Topics: reinforcement-learning, android
Last push: 2026-08-25T11:18:05+00:00

## Health v2 (maintenance only)
Score: 85/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 58, longevity 100
- inputs: {"age_days": 1960, "days_push": 8, "days_rel": 64, "gap_med": 433, "n_releases_24m": 2}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1240, forks 119 (observed 2026-08-28T04:04:06.025378+00:00)

## What it is
AndroidEnv is a Python library from DeepMind that exposes an Android device (real or emulated) as a Reinforcement Learning environment. Agents interact via a universal touchscreen action interface and receive pixel observations and task-defined rewards.

## Use cases
- train RL agents to control an Android device
- define custom RL tasks on top of Android apps
- research GUI/screen-based agent learning
- benchmark agents on real-world mobile tasks
- set up a virtual Android device for RL experiments
- train agents to play Android games via touch events

## When to choose
- you need an RL environment grounded in a real mobile OS
- you want to train agents that interact via touch gestures
- you need pixel-based observations with configurable task rewards
- you're doing DeepMind-style RL research on Android

## When to avoid
- you need a simple desktop or web RL environment
- you want lightweight unit testing of mobile apps rather than RL training
- you can't run an Android emulator or device in your setup
- you need iOS or cross-platform mobile RL

## Facets
- artifact type: library
- maturity: active
- function: reinforcement-learning, simulation, machine-learning
- domain: reinforcement-learning, android-tools, artificial-intelligence, machine-learning
- platform: python
- tags: rl-environment, gym-environment, android-emulator, deepmind, agent-training, touchscreen-control, android, linux, macos

## Member repositories
- google-deepmind/android_env (main) score 85

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:06.025378+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T05:09:52.317299+00:00, confidence not recorded.
  - readme: https://github.com/google-deepmind/android_env (fetched 2026-08-28T04:04:06.025378+00:00, sha 4867ac0f097e)
- Data as of 2026-08-30T08:39:29.467469+00:00.