Accessibility: Touch Access
Now that there are Chromebooks with touch screens, it is important that we have an interface for users with disabilities to be able to interact with a touch screen. It's based on the same ideas used on iOS VoiceOver and Android TalkBack, but with a few Chrome-specific differences.
Currently iOS and Android provide mobile touch accessibility by allowing users to touch the screen to explore the content currently showing. The user can hear spoken feedback about whatever is under their finger. They may then perform certain gestures to send a click action to the last explored target. There are other gestures to adjust settings such as volume or reading speed, to navigate between items on the page, to enter passthrough mode (where touch events are sent through directly as if accessibility was off), and to perform many other actions.
Touch accessibility on ChromeOS essentially works like this:
Users may use a finger to explore the screen by touch, hearing spoken feedback about the item under their finger. They may then perform certain gestures such as tapping and swiping to send a click to the last explored target, entering passthrough mode (where touch events are sent through directly rather than being intercepted by the accessibility system), adjusting volume, changing tabs, or scrolling.
The feature turns on when spoken feedback is enabled. Currently there's no way to turn touch accessibility on or off independently; for now we're assuming that any users who want spoken feedback on probably want this functionality too, but we could make it a preference later.
The majority of the code exists in the touch exploration controller. When spoken feedback is enabled, an instance of this class is created and all touch events inputted by the user are rewritten through RewriteEvent. Sometimes these events are discarded, sometimes they are released with a delay, and sometimes they change the user's current state (e.g. from single finger pressed state to touch exploration state).
## Navigating the Screen
- When the user is in the touch exploration state, simulated mouse moves are dispatched whenever the finger moves. These mouse moves causes the screen reader (ChromeVox) to read the name of the object the user is touching.
- Touch exploration state is entered by holding a finger down for more than 300 ms, or by moving a finger on the screen for longer (grace period ends) or slower (leave the grace area at a low velocity) than it takes to perform a gesture.
- Once in touch exploration mode, all movements are translated into
synthesized mouse move events. This causes hover states and events
to be triggered. (e.g. a menu that opens on a mouse hover would open
with touch exploration)
- The last touch exploration location is stored and updated after every artificial mouse move.
- Touch exploration mode is left when the touch exploration finger is lifted or when other fingers are added.
- Currently, there are three different ways of clicking on the item that was most recently touched through touch exploration.
* If the user is in touch exploration mode, they can click on the last touch exploration location by lifting their finger and quickly placing it on the screen again within the double tap timeout (300 ms).
* If the user is in touch exploration mode, they can click without lifting their touch exploration finger by tapping anywhere else on the screen with a second finger. * Details: * After a successful split tap has been completed, the user will be back in touch exploration. * If either finger moves too far from its original location, no click goes through. * If the touch exploration finger is lifted first, a click still goes through when the split tap finger (the second finger) is lifted.
* If the user double-taps, the second tap is translated as a click to the location of the last successful touch exploration - that allows the user to explore anywhere on the screen, hear its description, then double-tap anywhere to activate it. * Details: * The double tap must be completed 300 ms after the touch exploration is released or else it is considered to be a single tap. * In terms of timing, the second tap must also be completed within 300 ms of the first tap.
- Gestures can be used to send high-level accessibility commands.
* The user can perform swipe gestures in one of the four cardinal directions which are then interpreted and used to control the UI. * A gesture will be registered if: * the user places one or more fingers on the screen and moves them all in one of the cardinal directions * the distance moved is far enough (outside the "slop region") to determine that the movement was deliberate * all fingers are released after the gesture is performed * the entire gesture is performed within the grace period (300ms) * If the grace period expires before all fingers are lifted: * If only a single finger was placed down, then touch exploration mode is entered. * Otherwise, the system will wait for all fingers to be lifted before processing any more touch events. * Details: * One finger swipes are used for navigation on the screen. For example, a swipe right would correspond to the keyboard shortcut **Shift+Search+Right** (**ChromeVox+Right**). * The default for this is **up/down:** go to next/previous group, and **left/right**: go to next/previous object (for instance, bullet points), but this can be changed in ChromeVox settings. * Two finger swipes are mapped as: * **up**: go to top * **down**: read from here * **left**: browser back * **right**: browser forward * Three finger swipes are mapped as: * **up**: page down (scroll by a set amount) * note that up is page down because with touch it's like you're literally dragging the page up - which lets you see down the page * **down**: page up * **left**/**right**: scrolls through the tabs * Four finger swipes are mapped as: * **up**: home page * **down**: refresh * **left**/**right**: decrease/increase brightness (for low vision users) * Note that all of these mappings are not final and will probably be shifted to best help users. Ideally there will eventually be a menu for users to pick their own mappings
* Slide gestures performed on the edge of the screen can change settings continuously. * Currently, sliding a finger along the right side of the will change the volume. * Volume control along the edge of the screen is directly proportional to where the user's finger is located on the screen. * The top right corner of the screen automatically sets the volume to 100% and the bottom right corner of the screen automatically sets the volume to 0% once the user has moved past slop. * This could be applied to other settings in the future (brightness, granularity, navigation of headings). * Details: * If the user places a finger on the edge of the screen and moves their finger past slop, a slide gesture is performed. * The user can then slide one finger along an edge of the screen and continuously control a setting. * Once the user enters this state, the boundaries that define an edge expand so that the user can now adjust the setting within a slightly bigger width along the side of the screen. * If the user exits this area without lifting their finger, they will not be able to perform any actions, however if they keep their finger down and return to the "hot edge," then they can still adjust the setting. * In order to perform other touch accessibility movements, the user must lift their finger.
- Passing finger events as if accessibility is off is useful for smooth scrolling, drag and drop, and pinching. There are currently two implementations for passthrough.
- Passing finger events through is also useful for anything which implements custom gestures, as mentioned in the "Why A New Set Of Features" section.
- An earcon will always sound when the user enters passthrough mode, allowing the user to know that passthrough has begun.
* If the user double taps and holds for the passthrough time-out, all following events from that finger is passed through until the finger is released. * This can be useful if a certain element requires a gesture to operate, allowing the user to touch explore to that element, then perform the gesture without needing to precisely locate the element again. * Details * These events are passed through with an offset such that the first touch is offset to be at the location of the last touch exploration location, and every following event is offset by the same amount. * If any other fingers are added or removed, they are ignored. * Once the passthrough finger is released, passthrough stops and the user is reset to no fingers down state.
* When a user holds a finger in a bottom corner of the screen, any other fingers placed on the screen are passed through. * Details: * The user needs to place their finger on the corner of the screen for 700 ms (passthrough time-out) before passthrough will begin. * An earcon will play to indicate that passthrough has been activated. * Once activated, all the events from any subsequent fingers placed on the screen are passed through as long as the finger in the corner stays in the corner and is not released. * Once the finger in the corner has been released, all the following events are ignored until all the fingers have been released.
- If the user taps and releases their finger, after 300 ms from the initial touch, a single mouse move is fired. This can be used to explore an item on the screen without entering touch exploration state.
- Earcons (short sound notifications) are fired whenever the user comes close to moving off the screen or moves onto the screen from the edge. Many touch devices have a smooth bezel around the edge of the screen, making it practically impossible to tell where the touch screen starts by only touch. This earcon helps the user be more aware of whether or not their finger is on the touch part of the screen.
To determine what would be the best direction for new ChromeOS touch accessibility features, there were a few things we needed to consider. First of all, people who were already accustomed to using touch accessibility on mobile had opinions on what functions they believed to be the most useful. Second of all, ChromeOS accessibility was new and only had a few features implemented when we arrived. We surveyed Google employees that used/were interested in accessibility features on mobile, and we asked for their preferences and ideas for the design for ChromeOS.
What we discovered:
- Most commonly used gestures: read next/previous item, read from here, jump to top, scroll by page, go back
- When scrolling, users tended to prefer to go by a page then to scroll smoothly
- Being able to use multiple fingers for gestures was considered to be much simpler and more intuitive than more complicated single finger gestures - especially on a bigger laptop screen
- Passthrough mode was rarely used and users were frustrated by Android's emphasis on it, but it was felt to be necessary - especially when encountering “web apps in the wild” on a Chromebook (apps could be developed needing fancy swipes or multi finger touches, and there is no "internet police" to stop them)
note: the sample size was fairly small, and we hope that we'll be able to collect more user feedback when the first version is released.
Therefore, we have decided to implement the most useful features borrowed from Android and iOS as well as features completely unique to ChromeOS that took advantage of its own capabilities. For example, corner passthrough and side slide gestures wouldn't have worked as well on a tablet or mobile device, but take advantage of the Chromebook's bigger screen to give the user better accessibility.
This is a graph of the possible states the user can be in, and all of the user input and timer time-outs that causes the touch exploration controller to change the user's state. Occasionally touch presses and releases are dispatched to simulate clicks, and these dispatched events are also shown in this chart.
This chart was generated using https://www.gliffy.com
Here are some ideas that didn't make it into version one, but would be awesome to implement soon:
- Menu for accessibility settings (detail of navigation by word/line/paragraph, rate of speech, pitch of speech, etc.)
- Drag and drop gesture to move an object or copy/paste while being able to navigate the screen between locations to find the drop point
- Pinch/zoom - for low vision users (currently possible through corner passthrough, but perhaps a three finger hold could zoom in)
- Enabling ChromeVox from anywhere - shortcut gesture recognized with ChromeVox turned off
- Custom gestures - user chooses gesture mappings