We’re often asked about virtual reality (VR) with integrated eye tracking and have, only recently, started recommending it to customers. News from the Consumer Electronics Show (CES), tests on different VR headsets (Head Mounted Displays) and our own mobile heatmap technology now provide researchers with the option to include eye tracking on projects.
The promise of eye-tracking in VR has, in reality, been around for a long time with a number of manufacturers claiming to have cracked it. The problem, however, is reliability and the value of the data output. Our Chief Scientist, Dr Alastair Goode, first used eye tracking in the 1990s and the key problem, historically, was the sensitivity of the equipment; namely the movement of the camera compared to the eye it was tracking. A slight nudge would corrupt any meaningful measurement. The only solution was to – literally – fix the head to the camera. Respondents used a bite bar fixed to a camera rig which was bolted to a lab bench. Add to that a screw which clamped their lower jaw to the bite bar and it looked like a Dickensian experiment.
Whilst technology has advanced in the last three decades, the issue (solved by the bite bar) still remains. Integrating eye tracking into a Head Mounted Display (HMD) virtual reality headset is a challenge. As soon as you calibrate the eye tracker to start taking measurements, a slight movement of the HMD on the head renders all your measures redundant. Measuring eye movements is actually quite tricky with errors in the measurements causing ‘drift’. The hack is to tightly fix the HMD to the respondent however this feels a bit like the old days of bite bars.
The other challenge with eye tracking integration with HMDs is the eye tracking cameras ideally need to be at 90 degrees to the eyes – exactly where the virtual reality display lens needs to be. Most systems use pupil tracking (an infra-red light shone into the eyes) which makes the pupil look like a black circle to the camera. This black circle is recognised by clever software that follows movement and, by knowing where it started, calibrates an estimate of where the eye is pointing. When cameras are placed at an acute angle, as per the Fove headset, our tests have shown that it stops tracking when you look up as the eye tracking camera can’t see the pupil. The acute angle leads to an increase in ‘drift’ so it can only really track the eye (when it does) for short periods of time.
Tobii have solved this problem by using artificial intelligence (AI) to model the actual eye balls. They take multiple measures from each eye and the eye ball is modelled by software so that, if the camera is moved, it can quickly understand the movement, adjust and carry on taking measures. Our team have tested all eye tracking headsets ‘in anger’ and Tobii is one of the few that stands up against deliberate ploys to confuse the eye trackers. Tobii also power the recently launched HTC Vive Pro Eye with eye tracking directly integrated into one of the market’s most successful HMDs. This shows that software and hardware manufacturers see significant market potential for VR eye tracking as there are seven fundamental reasons it will change VR in a number of sectors.
Integrating eye tracking into VR is now a reliable option for researchers who want additional data when using VR as a consumer research tool. It’s a compelling offer for the market research industry, albeit one which significantly increases the cost of conducting qual projects. The price of entry will decrease over time however, in the meantime, our mobile VR heatmaps (using head movement as a proxy for what respondents are looking at) deliver an immediate and cost effective solution for both qual and quant studies.
Recent Comments