SpliT9

VIRTUAL REALITY
A novel split keyboard design for virtual reality.
Click here for the detailed research paper.

Introduction

For this project, we focused on intuitive and effective text keyboard inputs within virtual reality platforms. Currently there is no standardized method of keyboard input in virtual reality. More often than not, the current text input methods do not make use of 3D interaction techniques, and simply translate a 2D keyboard into the VR space. Solutions such as split keyboards, which deploy smaller floating keyboards within the virtual space, have been proposed,
but have not been studied extensively. Therefore, among the many limitations to consider in designing a new keyboard, the one that we chose to focus on is the natural instinct to use both hands while typing, specifically with split keyboards.

Role

This study was conducted by myself and a partner. We both shared the following roles:

User Researcher
UX Designer
Unity/Oculus Developer

Research

There were three research questions that we focused on in this study:
(1) Does the controller based split keyboard perform better than the hand tracking split keyboard?
(2) Are the affordances provided by the SpliT9 keyboard intuitive?
(3) Does the SpliT9 keyboard improve typing performance in VR? For example, if they prefer the iPad split keyboard, does that influence their choice of virtual keyboard?

Our hypothesis was that the SpliT9 keyboard would make typing faster, easier and more enjoyable to the user. By reviewing the existing literature (see research paper linked in title), we decided to focus on evaluating controller based and hand tracking based split keyboards, and propose the SpliT9 keyboard based on the strengths and weaknesses we discovered.

Following this, I conducted three qualitative interviews with people who had been cyberbullied in different ways: mass trolling, cyber stalking, and cyber harassment.
"I wish I could just give my phone away for a week to have someone scrub it clean."
"While we shift blame on to platforms to fix it, this is actually a human glitch."
Our user research was summarized into the following two personas:

Prototypes

Controller-based split keyboard
Hand tracking split keyboard
SpliT9 keyboard
In order to evaluate the two split keyboards, we had to first begin by building each one since open source options to test them were not available to us. Both systems were built in Unity using C#, and were built for the Oculus Quest. Once each system was built, we tested them between ourselves and conducted a heuristic evaluation due to the lack of participants available at the time due to the ongoing coronavirus pandemic.

Testing & Results

For each users interaction with each keyboard, we measured their Words Per Minute (WPM) and Uncorrected Error Rate (UER). After each user completed the tasks, we also conducted a brief qualitative study to understand their experience better. We asked them a set of 5 questions to go through their thought processes and emotions while using the three systems, and learn if there was a preference. Finally, each user completed a System Usability Scale survey for the SpliT9 keyboard.
For each keyboard, each participant was given the same set of 5 phrases to type out to measure their typing performance. Their typing speed in WPM and UER were calculated for each session afterward. The chart on the right shows the average WPM for each keyboard design. We saw that on average, the WPM was highest on the hand tracking split keyboard. Four out of six participants had the highest WPM with the hand tracking keyboard than the other two. It was clear that participants were able to move their hands in a faster manner as compared to moving step by step using the joysticks.
Average Words Per Minute
Average Uncorrected Error Rate
However, the Hand Tracking Keyboard also showed the highest average uncorrected error rate (chart on the left). When using the hand tracking keyboard, participants would often accidentally press keys more times than expected or trigger the wrong keys.

Both controller-based keyboards have a significantly less UER
than hand tracking keyboard. In particular, SpilT9 has overall the lowest UER while have the second highest average WPM rate.
To measure usability, we employed a ten-item System Usability Scale (SUS) questionnaire. We asked participants to give a score ranging from 1 to 10 for each question indicating how much they agreed or disagreed with a statement regarding the ease of use. Overall we can see from the results that the SpliT9 keyboard performed slightly above average for most
statements.

Takeaways

We received a lot of great feedback on this project from the participants we were able to recruit. It is important to note however, that we were limited in the recruitment process due to the ongoing lockdown across the world. We were able to recruit among our classmates and friends and family, all of which present different biases to some of our results. Moving forward it would be beneficial for us to test with a larger sample of users.

There are also a few more design considerations that we would like to implement. We would like to offer users the option
to re-position the keyboards as they see fit. We would also like to give users the option to easily capitalize letters as needed. We believe building in an auto-complete functionality would significantly increase our average WPM, since many people have grown accustomed to this technology while typing. Finally, we would like to introduce more audio and haptics into our design to create a more realistic experience.
Powered by Webflow
LINKEDINGITHUBTWITTER