Hello Martina D'Agostini - if you interested in event-related changes in pupil diameter over time, I'd hesitate with Pupil Labs. They are low cost (which is inciting), but the cameras are not fixed. So if the participant moves the unit on their face, the pupil will be out of position and could cause errors in diameter measurement. Is there a reason you want to use a wearable system versus a screen based system? What is the task your participants will be doing?

We have been very happy with the Pupil Labs for use in naturalistic tasks (mainly driving, both real cars and simulated on a screen setup). We especially like the versatility for using the same system and much of the same workflow for lab and field experiments for maximum comparability. We've also trialled the system in driving, walking with the laptop in a backpack.

We have the high speed binocular which can record eye images at up to 120 Hz, though in practice we’ve used 30 Hz or 60 Hz which is fine for our applications. If you want to do very precise saccade timing fixation durations you may want to use 120 Hz. The higher you want to go, the more powerful data collection laptop you need, though. We have not used the stereo world camera tracker.

A recent paper found Pupil to compare favorably to the Tobii and SMI in terms of calibration accuracy with a traditional 9-dot calibration

https://www.biorxiv.org/content/biorxiv/early/2018/06/28/299925.full.pdf

The plus points for us are good signal quality, and that it is fairly easy to use. But the most important thing for us compared to most eye tracker manufacturers is that you have access to the raw eye images for post processing - and everything is open source code. So there are no black boxes you cannot open if your team has people with good Python coding skills. T

We use a custom head strap to reduce slippage of the headset, 3D print files available at

https://zenodo.org/record/1246953#.XAVVmC10f-Y

But it pays to recalibrate often!

We've also developed our own fixation/pursuit & saccade event detection algorithm that works well with field data https://www.nature.com/articles/s41598-017-17983-x

The data is usually very clean. Usually the only preprocessing step

before event detection is to drop samples where the confidence is bad (under 0.8 or so).

The cost of the system is not high - they use off-the shelf components and do not have extortionate pricing / closed-source “support”.

For field experiments, one more tip: https://figshare.com/articles/Supplementary_Movie_1_full_video_mp4/4498466

Here you'll see our field calibration routine, with the target on a tripod and the subject is asked to move their head to different poses. It works well, both in terms of participants understanding instructions and the calibration producing a good signal. The calibration is during the first four minutes of the video. The driver is moving his head according to the experimenter's instructions and looking at the "target" on a tripod some 10m in front of the vehicle with different head poses.

One challenging thing is that as with any head-mounted system you get xy positions calibrated in reference to the head, not the world or a screen. For tracking the head relative to a monitor to project gaze ot the screen tthere is a readily available marker system. In the wild its a bit more challenging.

Hope this helps,

0tt0

thanks for all the answers!

I'm working with Pupil Labs eye-tracker to measure pupil diameter in a experiment that the volunteer must watch videos for 12 minutes in a light-controlled environment. I must say that this equipment has good and bad points (as any other):

It is portable and easy to use in any experiment, besides the fact that I'm using fNIRS at the same experiment and the infra-red signal of the headset doesn't act so much (it depends on the layout of the probes of fNIRS too) as noise to the infra-red signal of the fNIRS.

One problem that is more related to the preprocessing of the data is that the pupil_positions.csv file (the one that contains the data of pupil diameter and confidence) doesn´t show you in what timestamps the equipment didn´t sample the data for any reason. You could use the data about the recorded FPS time series to know when the data wasn't sampled, but this information isn't exported to the csv file (at least I didn't find a way to do this - if someone knows how, I would be pleased to know haha).

For me, the Pupil Labs eye-tracker is a good equipment to make experiments in a lab environment in case you don't want to expend so much time preparing the headset and if you want to minimize the noise generated by the interactions of signals from different equipments being used together.

I have pupil and i have done a lot of projects in field of sport, it works well in visual search behavior and vision in action such as targeting skills.

Hope it works well for you,

Ayoub

Sorry but I don't have an answer. Though I have a serious question.

It is obvious that we create perceptual images of latent action trajectory shapes prior to the actual execution of a motoric action. How is eye-tracking-gear ever capable of recording those latent perceptual images?

Bruno Costa Caforio it seems like the information you are looking for is available in the raw data as described in the doc https://docs.pupil-labs.com/developer/core/overview/#pupil-datum-format

Absolutely not. Look at Tobii - only real game in town.

Michael Brown Your answer isn't very specific.

Nobody questions that Tobii is the current market leader, but given the pricing, I'm not very eager to go for it.

For me (and probably everyone else here) it would be much more helpful if you would flesh out in which aspects you consider Tobii to stand out against this competitor, apart from the market leader position.

Sorry but I don't have an answer. Though I have a serious question.

It is obvious that we create perceptual images of latent action trajectory shapes prior to the actual execution of a motoric action. How is eye-tracking-gear ever capable of recording those latent perceptual images?

Experiment Findings The scientific proof in relationship to the prior constructi...

Article Boy, are those eyes active!; Why a pro golf player lies on h...

Article Do the perception processes within the putt in mini-golf dif...

Thank you for to all the info in this thread. i will be getting mine soon and will this will most certainly help.