Gesto (EICS `19)

LogoTitle (1)

Objective

Gesto is a system that enables task automation for Android apps using gestures and voice commands. Using Gesto, a user can record a UI action sequence for an app, choose a gesture or a voice command to activate the UI action sequence, and later trigger the UI action sequence by the corresponding gesture/voice command. Gesto enables this for existing Android apps without requiring their source code or any help from their developers. In order to make such capability possible, Gesto combines bytecode instrumentation and UI action record-and-replay.

To evaluate the usefulness of Gesto, we develop four use cases using real apps downloaded from Google Play—Bing, Yelp, AVG Cleaner, and Spotify. For each of these apps, we map a gesture or a voice command to a sequence of UI actions. According to our measurement, Gesto incurs modest overhead for these apps in terms of memory usage, energy usage, and code size increase. We also evaluate our instrumentation capability and overhead by downloading 1,000 popular apps from Google Play and instrument them. Our result shows that Gesto is able to instrument 94.9% of the apps without any significant overhead.

Instrumentation

Screen Shot 2018-10-07 at 2.22.01 PM

Workflow

Screen Shot 2018-10-07 at 1.54.21 PM

Usage at App Development/Instrumentation Time: Gesto is a static bytecode instrumentation tool that injects new bytecode into an existing Android app and rewrites some parts of the app’s bytecode (more details on this in Section 3.5). It takes an existing Android app (a .apk file) as input, instruments the app, and produces another version of the app that is now capable of task automation using gestures and voice. A few ways are possible to use this tool: (i) a developer can run it on the developer’s machine to transform an app that the developer has written before releasing it to an online app market, (ii) a user can run it on the user’s machine to transform an app that the user downloads from an online app market such as Google Play, and (iii) a third party can provide a web service that runs the tool in the cloud.

Usage at Run Time: Once an app is Gesto-enabled and installed on an Android device, a user can automate custom tasks using gestures and voice commands. In order to create a UI task to be automated for an app, a user needs to (i) open the app, (ii) perform the “start recording” command (currently, a pre-registered gesture), (iii) perform a gesture or record a voice command that will be used as the trigger for replaying, (iv) record a sequence of UI actions to be automated, and (v) perform the “stop recording” command (another pre-registered gesture). In our current prototype, waving once near a device starts recording; waving twice near a device stops recording. Figure 1 depicts these run-time workflows.

After a user finishes mapping a sequence of UI actions to a gesture or voice command for an app, the user can simply perform the gesture or speak the voice command to trigger the action sequence when the app is open. We can further automate app opening by instrumenting the main app launcher of the user’s device using Gesto, since app launchers are regular Android apps. However, Gesto currently does not support cross-app task automation.

 

Use Case – Spotify App

Simplifying UI Interaction

A music player app is typically used to provide background music while a user performs other tasks. us, there is a high probability that they are occupied (such as when running or driving) and unable to interact with the app via the touch interface of the mobile device. The provided functions are set statically which means users cannot change their mapping or define custom mappings to other actions in an app. Further, they can enable just one action and not a complete sequence. Gesto instrumentation enables both dynamic mapping and mapping multiple UI actions with single gesture or word.

 

Use Case – Yelp App

Long Sequence of UI Actions

Today’s applications are content-rich with an involved, hierarchical UI. While this makes for more compelling applications, navigating to an element of interest requires executing a long sequence of UI actions. Some apps provide a bookmark feature which helps users to get to desired content that has been visited before. However, such bookmarks are limited to static content. Gesto allows a user to define a dynamic mapping to a sequence of UI events. We chose yelp to demonstrate this use case. Yelp provides information about local businesses along with customer reviews for these businesses.