Sector · Science

Here the glasses are the instrument, and two studies say what they cost

Twelve peer-reviewed papers. Eye-tracking headgear on marmosets and ferrets, a 33-step protocol on a lab bench, a resuscitation drill on a research ship over Starlink, and two studies that measured the discomfort instead of selling around it.

Everywhere else on this site the glasses help someone work. In science they are the measuring device. Half of these papers are not about a display at all: they are about a camera and an eye tracker strapped to a head, human or animal, recording where the gaze went and what the hands did. That distinction matters when you read the results, so it is stated here rather than blurred.

Two labs put trackers on animals. UC San Diego built CEREBRO, a lightweight wireless head-mounted eye tracker for freely moving marmosets, and found gaze more predictable in free movement than head-fixed, with compensatory eye movements and vestibulo-ocular reflex stabilising foveal vision during locomotion; it ran in PNAS. Rochester tracked freely moving ferrets in a visual discrimination task and found that heading predicted the animal's choice independently and better than saccades alone: the ferret turns its head first and confirms with the eyes after. On people, KU Leuven sat participants down to a plate of spaghetti with a phone to the left, to the right, or absent, and measured gaze pulled consistently toward the phone across the whole meal. Texas at Austin put wireless trackers on parents and toddlers at home and found that joint attention counted for more when the child was holding the named object.

The instrument in the field

At Zhejiang Sci-Tech and the China National Rice Research Institute the glasses are simply a 50 MP camera you do not have to hold: researchers photographed whitefly-infested vegetable leaves in the field and greenhouse and trained a modified YOLOv11 to count them, reaching 86.20% recall and 84.25% precision. The same group counted ten categories of planthopper off monitoring plates with a Cascade-RCNN model at 83.43% recall. Aboard the CSIC research vessel Odon de Buen, crew ran CPR and tourniquet drills wearing RealWear glasses linked over Starlink to an instructor who was not on the ship, and the paper concludes it is feasible to train a crew on site without a specialist present. At Wellesley College, eleven undergraduates wore Google Glass at least two hours a day in real biosafety level 1 labs, stepping through 33-step, 20-step and 13-step bench protocols by voice while their hands held pipettes.

What the two honest studies measured

The University of Hong Kong took three headsets to an archaeological dig on Armenia's Ararat plain in 2023 and reported that heat, dust and strong sun prevented continuous use of all three: weight discomfort, short battery, a small digital workspace, poor screen visibility in direct sunlight and a narrow field of view. The authors call their own result a proof of concept and a vision of the future, not routine field practice.

Marquette University fitted twelve utility line workers and thirteen manhole workers with a HoloLens, a RealWear HMT-1, or nothing, and measured surface electromyography on eight neck muscles plus blink rate. Neck muscle activity showed no significant difference with or without the glasses, which is good news. Blink rate fell by 8 to 11 blinks per minute more with the HoloLens than with the HMT-1, which the authors flag as an eye fatigue and dry eye risk over long procedures. That is the kind of number a vendor never publishes, and it ran in Human Factors.

Cited cases

Who
Richard W. Marklin (Marquette University), Ashley M. Toll (Milwaukee Tool Co.), Eric H. Bauman (Electric Power Research Institute, EPRI), John J. Simmins (Alfred University), John F. LaDisa and Robert Cooper (Marquette University)
What they did
Twelve line and substation workers and thirteen manhole workers at electric utilities carried out their normal field tasks wearing a Microsoft HoloLens, a RealWear HMT-1, or no glasses at all, while the researchers recorded surface electromyography (EMG) from eight neck muscles and blink rate to measure muscle load and eye fatigue.
What came out of it
There was no significant difference in neck EMG activity between wearing and not wearing the AR glasses, but blink rate fell by 8 to 11 blinks per minute more with the HoloLens than with the HMT-1, a finding the authors flag as a risk of eye fatigue or dry eye for anyone wearing the device through long procedures. Published in the journal Human Factors.
doi.org
Who
Peter J. Cobb (Faculty of Arts, University of Hong Kong) and Hayk Azizbekyan (Institute of Archaeology and Ethnography, National Academy of Sciences of Armenia), with the Ararat Plain Southeast Archaeological Project (APSAP)
What they did
During APSAP's 2023 excavation season the researchers field-tested Vuzix Blade 2 AR glasses side by side with Microsoft HoloLens 2 and Meta Quest Pro headsets, in four experiments: replacing phone-based data collection with hands-free voice and gesture interaction, overlaying 3D models of already excavated remains on site for direct stratigraphic comparison in the trench, guiding excavation down to a flat surface using a depth guidance system on the HoloLens 2, and digitally manipulating ceramic restoration.
What came out of it
Published in the Journal of Computer Applications in Archaeology (volume 7, issue 1, p.370-387, Nov 2024): the authors report that heat, dust and strong sun at the site imposed limits that prevented continuous use of all three devices, among them discomfort from weight, short battery life, a small digital workspace, poor screen visibility in direct sunlight and a narrow field of view; the article itself treats the outcome as a proof of concept and a vision of the future, not as routine field adoption.
doi.org
Who
Grace Hu, Lily Chen, Johanna Okerlund and Orit Shaer (HCI Lab, Wellesley College), in partnership with the Cidar Lab, a synthetic biology lab at Boston University
What they did
Eleven undergraduate researchers wore Google Glass for at least two hours a day in real biology labs (biosafety level 1), running a purpose-built app that displayed bench protocols step by step (a 33-step miniprep, a 13-step ligation, a 20-step transformation) navigable by voice, hand gesture or touch, while their hands were busy with reagents, pipettes and glassware.
What came out of it
Published in CHI'15 Extended Abstracts (ACM, Seoul): users reported that seeing the protocol in their field of view increased mobility and confidence at the bench ("I felt that I was able to do the procedure smoothly and confidently thanks to Glass", in one participant's words), and voice commands worked well even with gloves on; the reported limitations were a 2 to 5 hour battery, gesture recognition sensitive to lighting and awkward touch navigation with gloves, leaving average user satisfaction only moderate by the end of the study.
doi.org
Who
Carlos Albarran Morillo and John F. Suarez-Perez (Politecnico di Torino), with Micaela Demichela (Politecnico di Torino), Monica Andrea Camargo Salinas (Tecnologico de Monterrey) and Nasli Yuceti Miranda Arandia (Universidad de America, Bogota)
What they did
They fitted workers on a pharmaceutical packaging line with Tobii Pro Glasses 3 eye trackers across day and night shifts, cross-referencing the gaze data with self-report questionnaires and lighting measurements to quantify visual fatigue over the course of a shift.
What came out of it
They identified a progressive increase in visual fatigue during night shifts, with physically demanding tasks under poor lighting as the worst case; published in the journal Sensors (MDPI) as the basis for an ergonomic intervention framework.
doi.org
Who
Silvia Aranda-Garcia (GRAFAIS Research Group, INEFC-Universitat de Barcelona), Roberto Barcala-Furelos (Universidade de Vigo, REMOSS group), Jordi Sorribas (Spanish National Research Council, CSIC), Roberto Gonzalez, Marc Darne and Antonio Rodriguez-Nunez (University of Santiago de Compostela / SICR
What they did
Crew and scientific technicians aboard the oceanographic research vessel Odon de Buen used RealWear smart glasses connected by Starlink satellite to the Remote Eye platform to run two onboard emergency simulations (basic life support with a CPR manikin, and haemorrhage control with two types of tourniquet on a simulator arm), guided by an instructor who was not physically on the ship.
What came out of it
Proof of concept published in Resuscitation Plus: participants rated the remote, just-in-time training as useful, realistic and motivating, and the authors conclude it is feasible to train the crew on site without a specialist physically present, while keeping effective real-time communication.
doi.org
Who
Shilong Zhao, Jun Lyu, Shuhua Liu, Zelin Feng, Heping Ling, Jiabao Jiao, Zhaoxin Ni, Baojun Yang, Qing Yao and Ju Luo (Zhejiang Sci-Tech University and the State Key Laboratory of Rice Biology, China National Rice Research Institute)
What they did
Researchers used SUPERHEXA AR glasses (from Chinese manufacturer Superhexa Century Technology), with a 50 MP camera, to photograph whitefly-infested vegetable leaves directly in the field and greenhouse between May and October of 2023 and 2024, and trained an improved YOLOv11 model to segment the leaves and count the pests automatically from those images.
What came out of it
The system reached 86.20% average recall and 84.25% average precision (mAP50 of 91.60%) in counting whiteflies automatically; published in Frontiers in Plant Science.
doi.org
Who
Haiyuan Sheng, Qing Yao, Ju Luo, Yongjian Liu, Xiangfu Chen, Zhongru Ye, Tiezhuang Zhao, Heping Ling, Jian Tang and Shuhua Liu (State Key Laboratory of Rice Biology, China National Rice Research Institute, with collaborators)
What they did
The same rice pest research group used AR glasses with an HD camera to photograph planthoppers caught on white monitoring plates during field surveys, building a Cascade-RCNN based model ("Cascade-RCNN-PH") to detect and count ten categories of planthopper automatically from the images.
What came out of it
They reported 83.43% average recall and 83.56% average precision in planthopper detection; published in Computers and Electronics in Agriculture. The authors note that image capture by AR glasses could extend to other pests, such as aphids and whiteflies.
doi.org
Who
Vikram Pal Singh, Jingwen Li, Kana Dawson and Cory T. Miller (Cortical Systems and Behavior Lab, Department of Psychology, University of California San Diego), with Jude F. Mitchell (Department of Brain and Cognitive Science, University of Rochester)
What they did
They developed CEREBRO, a lightweight wireless head-mounted eye tracking system for freely moving marmosets in a naturalistic environment, using an infrared camera and a neural network to track the pupil and record the animal's gaze during locomotion.
What came out of it
Published in PNAS: gaze was more predictable in freely moving animals than when head-fixed, and the study identified compensatory eye movements and a vestibulo-ocular reflex stabilising foveal vision during movement.
doi.org
Who
Silei Zhu and Marc Mancarella (Del Monte Institute for Neuroscience, Center for Visual Science, Department of Neuroscience, University of Rochester), with Tingan Zhu (Columbia University), Abdelrahman Sharafeldin (Georgia Institute of Technology) and Farran Briggs (University of Rochester / National
What they did
They used head-mounted eye trackers on freely moving ferrets during a visual discrimination task, recording heading direction and eye saccades at the same time to understand how the animal combines the two movements when making a perceptual decision.
What came out of it
Published in eNeuro: heading direction predicted the ferret's choice independently and better than saccades alone, showing that the animal orients first with a fast head turn and only then confirms with a saccade to the target.
doi.org
Who
Simon Ladouce and Celine R. Gillebert (Brain and Cognition, Leuven Brain Institute, KU Leuven)
What they did
Participants wore mobile eye tracking glasses during a real meal (spaghetti), with a phone placed to their left, to their right, or absent from the table, and the researchers used computer vision on the egocentric video to measure where and how consistently gaze was pulled toward the phone during genuine everyday behaviour, outside the laboratory.
What came out of it
Published in Scientific Reports: the phone's position shifted gaze toward it consistently across the whole meal, showing that wearable eye tracking glasses capture spatial attention bias in natural, not simulated, behaviour.
doi.org
Who
Sara E. Schroer, Ryan E. Peters and Chen Yu (Department of Psychology, University of Texas at Austin)
What they did
Parent and toddler pairs (English-speaking and Spanish-speaking families) wore lightweight wireless head-mounted eye trackers during ordinary activities at home, such as playing with objects and eating meals, to record gaze and multimodal social interaction moment by moment outside the laboratory.
What came out of it
Published in Developmental Psychology: moments of joint attention in which the named object was being held by the child increased the child's attention to that object, revealing how embodied attention coordinates everyday interaction.
doi.org
Who
Joao Amaro, Rafael Ramusga, Ana Bonifacio, Andre Almeida, Joao Frazao, Bruno F. Cruz, Andrew Erskine, Filipe Carvalho, Goncalo Lopes, Ata Chokhachian, Daniele Santucci, Paulo Morgado and Bruno Miranda (Institute of Physiology, Lisbon School of Medicine, and Centre of Geographical Studies, both Unive
What they did
They built and field-tested a wearable backpack combining environmental sensors (microclimate, air pollution, noise) with electroencephalography (EEG) and mobile eye tracking glasses, worn by researchers walking real Lisbon streets, to capture brain activity, gaze and the surrounding urban environment together under natural conditions.
What came out of it
Published in Sensors: it reports successful field deployment at selected Lisbon locations, demonstrating the feasibility of the integrated multisensor platform for urban mobile neuroscience outside the laboratory.
doi.org

Devices used

Talk to us

Back to the use cases