# Training SONAR spatial interpretation using virtual reality

## TL;DR

This study shows that a VR-based conformal display improves SONAR interpretation speed and accuracy but may lead to over-reliance in some users.

## Contribution

A novel VR display format for SONAR data interpretation is introduced, showing improved performance but variable over-reliance effects.

## Key findings

- The conformal VR display improved speed and accuracy in localizing enemy vessels from SONAR data.
- Over-reliance on the VR aid varied between participants, with some showing no cost in unsupported performance.
- Gaze distribution during training predicted whether participants would suffer performance costs without the aid.

## Abstract

Submariners must interpret SONAR data and make rapid tactical decisions under challenging conditions. The broadband time-bearing plot is a common visualisation method which requires mental spatial transformations to generate an actionable representation. Participants (N = 81) were trained on a VR scenario that required them to spatially localize an enemy vessel indicated on a time-bearing plot. In the supported condition, participants were given access to a novel egocentric, conformal display, which wrapped the time-bearing plot around the user's field of view. The unsupported group received no assistance. Compared to a baseline unsupported condition, we found that the VR overlay markedly improved both speed and accuracy. Test performance revealed over-reliance in the supported group, but this phenomenon varied notably between participants. Those who jointly used the initial time-bearing plot and the conformal aid during training showed good performance with the aid, and no cost of prior exposure at test. Comprehension of SONAR data can be aided by novel display formats, but care must be taken to avoid over-reliance.

Modern technical roles often require operators to make rapid interpretations of complex, high-level mathematical abstractions or plots of incoming sensor data. This study examines how novice adults learned to use time-bearing “waterfall” representations of broadband SONAR data—a common display format for submariners and other naval officers—in order to localize the depicted contact. We found that a novel, virtual reality-enabled display format markedly improved performance when present, but hindered subsequent unsupported performance. We found that over-reliance varied notably between people, with some people showing no cost of prior exposure to the support. These people were identifiable by their gaze distribution when the aid was present, which may inform policies regarding how visual supports are monitored and assessed during training.