AndroidEnv - The Android Learning Environment
AndroidEnv is a Python library that exposes an Android device as a Reinforcement Learning (RL) environment. The library provides a flexible platform for defining custom tasks on top of the Android Operating System, including any Android application. Agents interact with the device through a universal action interface - the touchscreen - by sending localized touch and lift events to the system. The library processes these events and returns pixel observations and rewards as provided by specific task definitions. For example, rewards might be given for events such as successfully scrolling down a page, sending an email, or achieving some score in a game, depending on the research purpose and how the user configures the task.
Index
- Environment details
- Running AndroidEnv
- Setting up a virtual Android device
- Defining a task in AndroidEnv
- Example tasks available for download
Environment features
There are a number of aspects that make AndroidEnv a challenging yet suitable environment for Reinforcement Learning research:
-
Allowing agents to interact with a system used daily by billions of users around the world, AndroidEnv offers a platform for RL agents to navigate, learn tasks and have direct impact in real-world contexts. The environment wraps a simulated Android device, which runs independently from the environment, completely unaltered, and works in exactly the same way as the devices that humans use, exposing exactly the same features and services.
-
The platform offers a virtually infinite range of possible tasks, all sharing a common action interface. The library facilitates the design of Reinforcement Learning tasks for any existing or custom built Android application. For example, it exposes the broad world of Android games, ranging from card games, puzzle games, time reactive games, all requiring a diverse set of action combinations and interaction types.
-
The environment runs on top of a real-time simulation of an Android device. In other words, the environment dynamics does not wait for the agent to deliberate, and the speed of the simulation cannot be increased.
-
The observation is a collection of RGB values corresponding to the displayed pixels on the screen. The exact screen resolution depends on the simulated device, but in general it will be considered relatively large in an RL context. However, users have the option of downsampling each observation.
-
The learning environment has an interesting, complex action space unique to the touchscreen interface of Android.
- The raw, hybrid action space consists of a continuous tuple signifying the action location, and a discrete signal determining whether the agent wants to touch the screen or lift its virtual finger.
- Raw actions are highly composable: the Android UI and most applications were designed so that they could be intuitively navigated via common touchscreen gestures such as tapping, scrolling, swiping, pinching, drag & drop etc. This is still the case in AndroidEnv: to trigger meaningful changes in the environment, the agent often has to perform carefully timed and positioned sequences of raw actions. For example, in order to navigate to the next image in a photo gallery, the agent would have to perform a swipe, touching the screen multiple times, gradually shifting the actions' positions to the right. Thus, in most contexts raw actions do not trigger changes in the state of the environment unless correctly chained together to make up a human gesture.
- The action interface is closely related to the observation space, as meaningful touch and lift events are often either co-localized or strongly correlated to the location or movement of salient objects in the observation. For example, the position of a button on the screen aligns with the location of the actions that trigger the button press.
- The library provides tools for flexibly altering the action interface if needed for particular studies, such as discretization or hard-coding gesture skills. Still, we believe that the real challenge remains in devising agents that are capable of dealing with a large suite of diverse tasks, through acting and learning in the complex unifying action interface.
Getting started
Installation
The easiest way to get AndroidEnv is from the latest AndroidEnv codebase from git's main branch:
$ git clone https://github.com/deepmind/android_env/
$ cd android_env
$ python3 setup.py install
Create a simulator
Before running the environment, you will need access to an emulated Android device. For instructions on creating a virtual Android device, see the Emulator guide.
Define a task
Then, you will want to define what the agent's task is. At this point, the agent will be able to communicate with the emulated device, but it will not yet have an objective, or access to signals such as rewards or RL episode ends. Learn how to define an RL task of your own, or use one of the existing task definitions for training.
Load and run
To find out how to run and train agents on AndroidEnv, see these detailed instructions. Here you can also find example scripts demonstrating how to run a random agent, an acme agent, or a human agent on AndroidEnv.
About
This library is developed and maintained by DeepMind.
You can find the technical report on Arxiv.
If you use AndroidEnv in your research, you can cite the paper using the following BibTeX:
@article{ToyamaEtAl2021AndroidEnv,
title = {{AndroidEnv}: A Reinforcement Learning Platform for Android},
author = {Daniel Kenji Toyama and Philippe Hamel and Anita Gergely and
Gheorghe Comanici and Amelia Glaese and Zafarali Ahmed and Tyler
Jackson and Shibl Mourad and Doina Precup},
year = {2021},
eprint = {2105.13231},
archivePrefix = {arXiv},
primaryClass = {cs.LG},
volume = {abs/2105.13231},
url = {http://arxiv.org/abs/2105.13231},
}
Disclaimer: This is not an official Google product.
886 Dec 27, 2022
67 Nov 29, 2022
2 Jun 24, 2022
5 Jan 01, 2023
6.2k Jan 01, 2023
1 Dec 06, 2021
7 Dec 31, 2022
1 Feb 01, 2022
3 Sep 07, 2022
29 Dec 12, 2022
61 Jan 06, 2023
10 Dec 28, 2022
3 Aug 03, 2022
14 May 29, 2022
3 Dec 09, 2022
9 Dec 31, 2021
2 Jul 25, 2022
1 Nov 17, 2021
1 Dec 13, 2021
1 Dec 15, 2021