Training Effectiveness Evaluation of an Adaptive Virtual Instructor for Naval Aviation Training

The Navy investigated VR training capabilities within their aviation training and results showcased significant promise for using VR devices, but also highlighted limitations reducing its potential benefit to training (McCoy-Fisher et al., 2019). Two critical issues identified were the need for guidance on the what, when, and how to train with VR and feedback on performance. As a potential solution, both the Air Force and the Navy started work in artificial intelligence (AI) instruction applied to flight training. This study examines an experimental version of an adaptive virtual instructor, the Virtual Instructor Pilot Exercise Referee (VIPER (R)) from Discovery Machine Inc. (DMI), a first step towards an AI instructor Pilot capability. For the Navy to better understand the training impact VR and introduction of an AI-style tutor may have, this study evaluates whether there were performance improvements in student naval aviator (SNA) flight events across different training conditions utilizing the Navy's Immersive Training Devices (ITDs). There were four conditions in this study: 1. Archival: no access to ITDs, 2. Free VR: free-play access to ITDs without guidance or VIPER (R), 3. Assigned VR: required practice scenarios in the ITDs without VIPER (R), and 4. VIPER (R): required practice scenarios in the ITDs with VIPER. Results provide quantitative data on the effectiveness between groups demonstrating VR benefit to training and qualitative feedback on the utility and usability of an adaptive virtual instructor, in this case VIPER (R).