Eye-tracking study paves way for gaze-based phone interaction

Human-computer interaction specialists from universities in Scotland, Germany and Portugal are investigating how ‘gaze-based’ interaction could be integrated into future generations of technology.

Researchers investigated how three forms of gaze interaction work while users are walking or sitting, and which methods users prefer in both situations. The results are set to be presented as a paper at the ACM Conference on Human Factors in Computing Systems later this month.

The team believes the paper could help shape the user experience of future mobile devices, likely to embrace eye-tracking technology as front-facing cameras become more sophisticated.

The paper is based on an evaluation of 24 study participants’ experiences with using different eye-based interaction methods whilst seated at a desk and then walking around a room.

According to the team, participants used the methods to select specific targets from a grid of white, circular shapes on a mobile phone screen each time one of the targets turned from white to black.

During the study, participants were asked to select different numbers of onscreen targets. The numbers of targets varied between two and 32, and were counterbalanced between participants to minimise the influence of extraneous factors, such as practice or fatigue, on the experimental results.

The three methods the participants were tasked with using were Dwell Time, Pursuits and Gaze Gestures. Dwell Time lets users select items by fixating their gaze on a target for 800 milliseconds. In Pursuits, users follow a small object orbiting around the target to select it. Gaze Gestures uses a multi-stage process where users look off-screen either to the left or right to narrow down the number of targets until they reach the one they want to select.

MORE IN ELECTRONICS

Researchers found that, when seated, participants preferred to use Pursuits. That method was also faster – on average, it took just 1.36 seconds for users to select a target, compared to 2.33 seconds with Dwell and 5.17 seconds using Gaze Gestures.

When on the move, users preferred Dwell Time. At an average of 2.76 seconds it was slightly slower than Pursuits at 2.14 seconds but faster than Gaze Gestures at 6.68 seconds.

Dr Mohamed Khamis, of Glasgow University’s School of Computing Science, supervised the research and co-authored the paper. He said: “Eye tracking has been well-studied in recent years across a range of user environments, but most of those where either the user, the camera, or both are stationary. 

“As smartphone camera technology has advanced, it’s become much more practical for eye-tracking to be implemented in those devices despite the challenges of both the device and the user moving at the same time.”

Dr Khamis said the eye-tracking methods could make it easier for people with mobility issues to use smartphones, as well as expanding the number of situations where devices can be used.

Omar Namnakani, a PhD student at the School of Computing Science and first author of the paper, added that the paper suggests guidelines for how the different methods can be used in different situations.

“Where users are sitting and there are less than nine targets on screen, Pursuits seems to be the best method to use based on our research. However, Pursuits can be tiring to use when there are more targets. In those cases, Dwell Time is the best option both when sitting and moving around,” he said.

“However, despite users’ preferences in our study and the slower speed of the input, Gaze Gestures was the most accurate method of selection when users were both seated and moving. When accuracy is more important than speed, that should be the preferred method of selection.”

The researchers plan to continue their collaboration to examine new methods of eye interaction which could improve on the three techniques explored. They are also keen to explore the implications of eye-tracking technology for users’ digital privacy.