Looxid Labs (Sunnyvale, CA and Seoul, South Korea) was recently in the news because the company’s VR headset was awarded a CES 2018 Best of Innovation Award in Virtual Reality. The company has also had investment from HTC.
The unique aspect of the Looxid smartphone-based VR headset is that it includes two embedded eye-tracking cameras to measure eye movement and pupil dilation and six EEG brainwave sensors to monitor brain activity. The eye and EEG measurements are captured synchronously with the VR content. Using this data, the company claims that the LooxidVR system can provide information not only about where the user has looked but also the users’ emotional state while interacting with the VR content.
A photograph of the LooxidVR headset appears in the figure below. A video illustrating and discussing the LooxidVR headset can be found at the end of this article.
The LooxidVR headset.
The company has reportedly been conducting extensive experimental research using machine learning algorithms to identify user emotions based on the eye tracking and EEG sensor data.
As a first step, Looxid Labs explains that electrical signals from brainwaves are sensitive to noise deriving from sources that include eye-blinking, muscle movement and interference from other electrical devices. To address these issues, the company developed a so-called Independent Component Analysis based algorithm that identifies the important features in EEG signals by automatically extracting and reducing the noise contained in the original signal thus improving the quality of the signal. The pre-processing needed to accomplish this includes four steps.
- Detrending
- Filtering
- Spike artifact removal
- Blink artifact removal
Once the signal has been “cleaned up” the analysis proceeds through several steps that include feature extraction, feature selection and classifier learning.
The insights derived from the EEG analysis are reported as three types of measurements that interpret the user’s current emotional state. The measurements fall into three categories: happiness/sadness, dominance/submissiveness and excitement/depression.
“Changes in pupil size are also monitored because they are closely related to the user’s cognitive status, memory, behavioral decisions and emotional changes. The Looxid AR infers the user’s exact pupil size by applying information tracking, filtering and 3D reconstruction methods.”
At this time, there is little information publically available regarding the conventional hardware aspects of the Looxid AR headset. Some of the hardware specifications relating to the headset’s EEG and eye tracking features are known and include the following.
- Brainwave sensor
- 6 channels
- Sampling rate 250, 500 and 1000Hz
- Resolution 24 bits
- Connection USB 2.0/3.0
- Eye tracking camera
- 2 binocular
- Sampling rate 60 Hz
- Resolution 640 x 480
- Accuracy 0.3 degrees
- Trackable FOV 90 degrees
- Connection USB 2.0/3.0
- Data output: pupil position, pupil size and gaze
It might be noted that currently available apps do not use the kind of data produced by the LooxidVR headset. It is, however, not hard to imagine the LooxidVR capabilities being applied to various VR applications that require a better understanding of users’ emotional state or, perhaps, in the development of more immersive VR systems.
Several such applications come to mind. As an example, “a video game that modifies its own difficulty based on the user’s stress reactions or a VR video with narrative paths that branch in response to the user’s brain’s whims.”
Although such applications are possible, the Looxid team is currently more focused on turning LooxidVR into a research tool exploring implications for business and health.
In medicine, for example, “it could be used in pain management or physical therapy. Detecting traits like confusion could be important to educational and training apps, preventing important concepts from escaping users.”
Pre-orders for the LooxidVR headset can be placed on the company web site. Deliveries are planned to start on February 1st. The price is not indicated. -Arthur Berman
