aSee glasses is an all-in-one solution designed to measure natural viewing behavior in almost any real-world environment and contains a key wearable eye-tracking component with a 60 Hz / 120 Hz binocular sampling rate to capture truly objective and deep insights. Quick set-up procedures and simple calibration are designed to reduce the training time of using the hardware and software, making eye-tracking data collection and analysis more accessible and convenient for developers and researchers working in the fields of psychology and neuroscience, marketing and consumer research, user experience and interaction, education and professional research.
This is a complete project I have led. I have worked together with a six people in-house design team for almost 2 years, I will share mainly the innovative approaches and key results I have made and skip quite a lot of details. aSee Glasses has earned contracts worth millions after it was published.
2017.12 - 2019.07
THE GOAL
Design the eye-tracking mobile accessory
Design the analyzing software
Design the user interfaces for the system
MY ROLE
I am the leader of developing mobile eye-tracking analyzing software and hardware suite from scratch.
During this project, we started from a sketch by collecting users' needs and analyzing competitors' products. Along the way, we have added innovative creations to the product and finally successfully launched the product in Asia markets.
Eye-tracking could show directly the participant's gaze points on the stimulus on the screen. Therefore it has been used as a tool to analyze participants' psychological status, for example, interestedness and distractedness. Eye-tracking algorithms have been improved over the decades by lots of research teams. However, it is still time-consuming to produce eye tracking hardware and software systems all by oneself. Since the last century, eye-tracking product manufacturers have played an important role in popularising eye-tracking study into behavioral studies and market research by providing eye-tracking hardware and analyzing software.
User wearing mobile eye tracker in a shopping scenario
Eye trackers are the devices used to collect participants' eye movement data. Nowadays, the most popular eye-tracking algorithm is called the corneal reflex method which is using the relative positions of the reflections of the fixed IR lights in the eye tracker devices and the pupil to calculate the gaze direction of the participant.
Besides the eye tracker hardware, eye tracking manufacturers always provide analyzing software with basic visualization functions and statistics of the collected gaze data such as heat map, gaze plot, and retrospective think-aloud to support user researchers who don't have programming skills.
The opportunity shows up when several key major players in the eye tracker industry have been purchased by Google, Facebook, and Apple and became their internal research team with no longer sales nor support for the market users. The huge development strength differences of the remaining players gradually result in a monopoly from a dominant player in the market. The mobile eye-tracking analysis product the dominant player is providing can't feed the diverse needs of the clients on the market. Here my team started to do deep market research on the potential markets that we could target and the infiltrating strategy we could use.
The product design and development roles have been controlled by very few leading companies in the eye-tracking field. For the quality of eye-tracking algorithm and the strength of hardware, development has great influences on the product completion, it is necessary to do competitive analysis based mainly on three aspects:
• Functionality feasibility analysis
• Pricing analysis
• Intellectual property analysis
Here shows some results of the above analysis. In consideration of privacy protection, most of the information is not real and some information is altered.
It is acknowledged that in different areas/countries the market can be unique and independent. The key to publishing a successful product is depending on whether the provider/sales can offer a localized service/product to meet the native needs. Even for the eye-tracking system whose functions were supposed to have has been shaped by academic researchers in the last century, the hidden added value must be met to open the market channel. In the light of manpower and time cost, I did researches on criteria for admitting competitors into the market in different countries/areas and ranked the most valuable target market and user group for our product.
By interviewing the targeted clients and analyzing their core needs, the two major scenarios have been summarized for generating system requirements and delivering the system parameters for the hardware and software. For example, because participants might be wearing this mobile eye-tracking system to walk around, the appearance of the eyeglass frame together with the accessory should be normal. The weight and the way of wearing it shouldn't cause discomfort or alienation of the participants in the testing environment. For the researchers, simple and fast data analysis tools should be provided in the analyzing software to make the whole research experience easy and pleasant.
Using scenario when using aSee on PC
Using scenario when using aSee on record assistant
Based on the market research and the developing capacity of my team, aiming to be a strong competitor in the targeted market, I have formulated the requirements for the mobile eye-tracking system and categorized those into three categories which should be developed by professionals from three different fields.
Eye tracker hardware: this part of the design requires optic system engineers to work together with the algorithm programmers.
Recording software: the designer and engineer should pay attention to system latency and communication protocols when dealing with this part.
Analyzing software: user experience designer and interface designer come in to improve the usability of the software.
Due to the range of this project, it is important to get everyone on board with what the team is aiming for and at the same time why the team is designing in this way. Not only creativities should play in this cooperation, but also the limitations on resources and technologies should be brought to the table for all the team members to avoid over-design. Therefore, I held several meetings to deliver a case study to the whole team by using modified scenario description swimlanes in the following section to generate an overall understanding of the work of different teams and the product with user scenarios.
Eye Tracker Hardware
Recording Software
Analyzing Software
For a long time, scenario description swim-lanes have been used as an important tool to find out how to define and design product modules based on their distributed value while being used during the business operational process.
For mobile eye-tracking analysis, the system should be mostly determined by scientific research users’ needs and their purposes. In order to design a highly user-friendly eye-tracking system, I have taken one eye-tracking analysis case study as an explanation to my development team and examined the discoveries of the system demands from this case study under scrutiny regarding the relationship between functions and usabilities in the marketing field where mobile eye-tracking study are very commonly used.
This modified scenario description of swim lanes is organized in a set of streamlined actions of how an eye-tracking test is performed. Its analysis is composed by mainly five parts:
• storyboardI from participant’s view;
• storyboardII from researcher’s view;
• corresponding business value;
• corresponding product module;
• corresponding system responses;
Besides a whole research circle which involves eye-tracking research method and the collaboration of multiple roles, it also contains very valuable information collected from customer interviews with researchers from universities and eye-movement analysis services providers, which was very helpful for my product group to define product packages, separate core demands, organize development tasks.
It very often happens in a big team that each information segmentation is limited to the professionals from different departments. For example, the algorithm developers are not aware of the purposes and the user scenarios of the product. Vice versa, the product managers and the marketing team are not familiar with the logic and reasons behind some dedicated design of some key features. This leads to inefficient communication inside my team.
By providing this case study to the whole team, each member has more knowledge of how this system is working with the customers. Thus they could contribute more ideas when a problem occurs from different perspectives but still talk on the same page.
In the first prototype, my team was struggling with finding a scene camera with a big range of field of vision. The tests showed that even though with the best camera we could find at that moment, the lower part of the recorded scene videos was missing. The idea of stopping to find a better camera but matching the angle of placing the scene camera with the human natural sight direction came to my mind. The following section provided the knowledge and the calculation for this modification.
Horizontal sight: 0°
Normal sight: - 15°
Natural sight: - 30°
Sitting operation sight: - 40°
The placement of the scene camera
For most people, left FOV almost equals to right FOV. So the problem here is how to find the most suitable vertical inclination angle of the scene camera: The FOV ( field of vision ) of scene camera: vertical: 56˚ horizontal: 81˚
According to the vertical eye movement maximum FOV
upper vision/lower vision = 25˚ / 85˚
The distributed upper FOV of scene camera = 56 ˚* 25/(25+85) = 12.7˚
The distributed lower FOV of scene camera = 56˚ * 85/(25+85) = 43.3˚
The inclination angle of the scene camera: 56˚/2 - 12.7˚ = 15.3˚
Because the calculation result does not conflict with the natural eyesight of humans (15˚), the parameter is recommended for product manufacturing.
After designing the wireframe of the analyzing and recording software on desktop and mobile record assistant, my team started to frequently interview my clients, host user tests, and carry out focus groups to define real user needs and validate the design ideas. When iterating the software, I put three principles in mind:
Put users' needs and feedback on usability before developers' opinions.
Put developers' opinions on the system requirements before users' opinions.
When there is no optimal solution for conflicts, try the most suitable direction for the situation.
To simplify the accessibility of carrying out eye-tracking research by using aSee Glasses mobile eye-tracking suit, for researchers with no knowledge in eye tracking, we designed the interfaces, both for recording and analyzing software, in a process of performing an eye-tracking experiment. What the researchers need to do is to follow the "next" button on the interfaces.
the flow of recarding and analysis software on desktop
the flow of recording software on mobile phone
the thousands of tests my team have done and words I have written for making aSee Glasses suit more user-friendly
I am so proud of what my team has done during this big project. When leaving the company, the product has been successfully published on the market and won lots of attention from Chinese and Japanese research teams. Reflecting upon my view on design eye-tracking products, I am so honored to participant in this project.
There is a trend in the industry to include eye-tracking technology into consumer goods, such as mobile phones (iPhone X and above), VR headsets, and AR glasses. It is reasonable since from the business perspective to consumer products could make much more profit and could be much easier to get a market hit, and from the research perspective, the benefits of understanding a user and providing more customized service for this user could be fascinating and considerate. These actions and experiments are bold. ( I call all the design attempts to get eye tracking into consumer goods market "experiments".) However, at the same time, voices from other fields need to be heard, such as philosophical, ethical, environmental on the results of such attempt daily interaction of eye trackers with posthuman, and the usage of collected data. For example, by analyzing the eye movement data, fatigue status could be detected. Thus, an eye tracker on the car could stop a lethal car accident due to distraction or loss of enough sleep for a driver. At the same time, some unconscious information will be also collected by the eye tracker and sent to computing server. What exactly lies behind this unconsciously leaked information of a human being is still under study. Therefore, more details and evidence on commercializing eye tracking into consumer products must be found from multiple channels for eyes as the information they contain are so important for a human being.
A product, or a trend, or an industry published in the market should always be created by two forces, constructors and thinkers. A good product needs people who can ask and reflect on why to be, what to be, how to be, when to be, and where to be. Now, in 2020, I am not a supporter of to-C eye-tracking products but advocate on developing eye-tracking for to-B business. I may change when my opinion grows after I have done more research work related to this topic.
To sum it up, keep skeptical, keep inclusive and keep passionate.