Home Eye tracking The EyeTribe stops, and that stinks

The EyeTribe stops, and that stinks

,

This post first appeared on pygaze.org, in December 2016.

TL;DR

This morning, the EyeTribe announced via an email to their customers that they would stop the development of their products. The particular reason is rather vague (“we’ve decided to go in a different direction with our technology“), and researchers across the board are not happy. The EyeTribe was the only real option for cheap eye tracking: It was great for demonstrations, for pupillometry and fixation control, it had a very elegant API, and the hardware was great for how much you paid for it. Best of all: It didn’t come with the restrictive licenses that almost all of the EyeTribe’s competitors use to milk their customers for more money. I, for one, am sad about the loss of this great company.

It might be good to point out that I have absolutely no official connection with the EyeTribe. I did produce unofficial software libraries for Python and Matlab, and did an independent assessment of the EyeTribe tracker’s suitability for scientific research.

The letter

This is the letter that EyeTribe sent to their customers this morning at 10:45 (UTC +0:00).

An Update From The Eye Tribe

Thank you for supporting The Eye Tribe and ordering the world’s first truly affordable eye tracker. It is customers like you that have helped us get to where we are today.

Unfortunately, we’ve decided to go in a different direction with our technology and will stop development of our products. We thought you should hear this news directly from us. We thank you for the time you’ve spent in discussions.

-The Eye Tribe Team

It is remarkable that the EyeTribe indicates that it has decided to go in a different direction than eye tracking. Although this might just be marketing-speak for “We didn’t get enough support from investors, so we are forced to quit our great enterprise”, it does leave us to wonder about those new directions. Is there too little money in eye tracking? Rumours are buzzing that Tobii (which has a bigger focus on commercial applications than other eye-tracking manufacturers) also fails to sell their hardware to the commercial market. (Other companies like SR Research and SensoMotoric Instruments seem to focus more on research and education applications.)

Another remarkable thing is that the email thanks customers for “the time you’ve spent in discussions“. This might just refer to the active community, who were supported by the EyeTribe and each other on the discussion forum. In my mind, one would phrase this as “support” rather than “time in discussions”. Does the EyeTribe’s choice of words reveal that they were a bit tired of their demanding clients, perhaps? Obviously, this is wild speculation, but it did leave me wondering.

Response in the scientific community

We’re sad. Very sad. And some company is pointing out that their stuff will still work with existing EyeTribe models.

Now what?

So, what do we do now? The EyeTribe won’t be shipping any new products, including their much-anticipated new $200 model. So where do we turn for cheap alternatives?

Tobii EyeX

One potential alternative is Tobii’s EyeX. At $150, it seems a perfect alternative. However, it’s license prevents data logging. Yes, you read that right, you’re not allowed to use the EyeX for storing any data. This makes it completely useless for a whole lot of purposes, including scientific publications. Tobii is said to be working on an additional license, for which researchers will probably have to pay dearly. So Tobii is off the table.

GazePoint GP3

GazePoint is a smaller company, but their GP3 tracker seems to be relatively popular* among researchers. At $495 $795 (price up-to-date as of May 2017), it’s still relatively inexpensive, but it’s considerably more than the $100 model that EyeTribe produced.

* I’m inferring this from the occasional requests we get to support it in PyGaze and OpenSesame. Unfortunately, our developers don’t have access to a GP3, so we can’t really develop and test any code. If someone feels charitable: I’m open to add support if someone were to send me a GP3!

UPDATE (2017-05-01): Someone did actually send me one, and now the GP3 is supported in PyGaze!

Webcams?


Part of why eye-tracking manufacturers are having trouble to sell their products to regular consumers, seems to be a lack of practical applications: With the exception of virtual reality, gaming doesn’t really seem to be improved by gaze interaction, and guiding your computer’s mouse cursor with your eyes isn’t as practical as it might seem at first. In addition, I think big players in tech might hesitate to invest time and money in hardware that might soon be obsolete. Although experts used to be quite sceptical about eye tracking with ordinary webcams, there have been promising developments in the area (MIT used the very contemporary approach of throwing neural networks at the problem). Companies like xlabs are already claiming that they can harness this upcoming technology in a meaningful way.

Does this mean we can use webcam tracking in research? Well, not quite. It’s hard to implement predictive models, as my webcam eye-tracking software nicely illustrates. It’s also hard to implement machine-learning approaches like the referenced MIT paper. Regular old researchers will likely be cheaper of investing some money in an eye tracker, than investing their time (and thus money!) in developing a webcam application.

Conclusion

EyeTribe, please come back!

Leave a Reply

Your email address will not be published. Required fields are marked *