Eye Tracking in Virtual reality : Pico Neo III Pro Eye Implementation

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216224%3A14210%2F24%3A00139650" target="_blank" >RIV/00216224:14210/24:00139650 - isvavai.cz</a>

Výsledek na webu

—

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Eye Tracking in Virtual reality : Pico Neo III Pro Eye Implementation

Popis výsledku v původním jazyce

With the advent of standalone VR HMDs, eye tracking (ET) implementations quickly followed. Currently, they are mostly created to heighten the gaming experience but can also serve as an advantageous research tool. The most common approach to measuring eye movements in HMDs is video oculography (VOG), and it's employed by the biggest eye tracker providers in the market, including Tobii, Pupil Labs, or Varjo. However, eye tracking data quality and the calibration itself remain one of the biggest challenges in ET in HMDs, as the precision, accuracy, and other metrics are often insufficient or not reported by the manufacturer at all (Adhanom et al., 2023). Researchers are currently trying to get more credible results by implementing their own solutions. The presented solution is a custom implementation of eye tracking calibration and logging functionality for the Pico Neo III Pro Eye standalone VR HMD with the use of the Tobii XR SDK in an immersive virtual environment created in the Unity game engine. Typically, to get raw ET data from the Tobii XR SDK, one would need to purchase a proprietary Tobii Ocumen Studio. However, basic ET data can be obtained from the base SDK as well. With the use of the “Gaze Visualizer'' object from the Tobii XR SDK and a Raycast functionality from Unity Scripting API, we are able to devise a simple script that logs the exact coordinates of the user and his gaze in a virtual environment throughout time while also simultaneously logging the names of particular objects the user is looking at. The frequency of this logging can be modified and can be set as low as approximately 10-15 ms as it is limited by the ET implementation that has a refresh rate of 90Hz. From these data, saccades, fixations, dwell times, gaze heat maps, and gaze paths can be calculated in post-processing. Calibration of ET can be done in a separate Tobii calibration app native to the Pico Neo III Pro Eye. Nevertheless, no raw data can be extracted from this app. For the purpose of objective ET calibration, a separate scene in our app was created where the user is tasked with focusing on appearing targets spread around the environment and then the average distance between the user's gaze and the center of each target in time is calculated; the average distance is then taken as an indicator of ET (in)accuracy. The ET data is logged and saved in the internal memory of the VR HMD in the form of CSV files and thus it is easily accessible and evaluable. This implementation is applied in a custom solution focused on testing the effect of perspective and embodiment on the user's perception and evaluation of cartographic tasks involving altitude, line of sight, and route length, as well as in a solution focused on different cartographic visualization methods for bivariate data in immersive virtual environments.

Název v anglickém jazyce

Eye Tracking in Virtual reality : Pico Neo III Pro Eye Implementation

Popis výsledku anglicky

With the advent of standalone VR HMDs, eye tracking (ET) implementations quickly followed. Currently, they are mostly created to heighten the gaming experience but can also serve as an advantageous research tool. The most common approach to measuring eye movements in HMDs is video oculography (VOG), and it's employed by the biggest eye tracker providers in the market, including Tobii, Pupil Labs, or Varjo. However, eye tracking data quality and the calibration itself remain one of the biggest challenges in ET in HMDs, as the precision, accuracy, and other metrics are often insufficient or not reported by the manufacturer at all (Adhanom et al., 2023). Researchers are currently trying to get more credible results by implementing their own solutions. The presented solution is a custom implementation of eye tracking calibration and logging functionality for the Pico Neo III Pro Eye standalone VR HMD with the use of the Tobii XR SDK in an immersive virtual environment created in the Unity game engine. Typically, to get raw ET data from the Tobii XR SDK, one would need to purchase a proprietary Tobii Ocumen Studio. However, basic ET data can be obtained from the base SDK as well. With the use of the “Gaze Visualizer'' object from the Tobii XR SDK and a Raycast functionality from Unity Scripting API, we are able to devise a simple script that logs the exact coordinates of the user and his gaze in a virtual environment throughout time while also simultaneously logging the names of particular objects the user is looking at. The frequency of this logging can be modified and can be set as low as approximately 10-15 ms as it is limited by the ET implementation that has a refresh rate of 90Hz. From these data, saccades, fixations, dwell times, gaze heat maps, and gaze paths can be calculated in post-processing. Calibration of ET can be done in a separate Tobii calibration app native to the Pico Neo III Pro Eye. Nevertheless, no raw data can be extracted from this app. For the purpose of objective ET calibration, a separate scene in our app was created where the user is tasked with focusing on appearing targets spread around the environment and then the average distance between the user's gaze and the center of each target in time is calculated; the average distance is then taken as an indicator of ET (in)accuracy. The ET data is logged and saved in the internal memory of the VR HMD in the form of CSV files and thus it is easily accessible and evaluable. This implementation is applied in a custom solution focused on testing the effect of perspective and embodiment on the user's perception and evaluation of cartographic tasks involving altitude, line of sight, and route length, as well as in a solution focused on different cartographic visualization methods for bivariate data in immersive virtual environments.

Druh

W - Uspořádání workshopu

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

<a href="/cs/project/GA23-06187S" target="_blank" >GA23-06187S: Identifikace bariér v procesu komunikace prostorových sociálně-demografických informací</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2024

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Místo konání akce

CAL, Olomouc

Stát konání akce

CZ - Česká republika

Datum zahájení akce

—

Datum ukončení akce

—

Celkový počet účastníků

15

Počet zahraničních účastníků

4

Typ akce podle státní přísl. účastníků

EUR - Evropská akce