Visual–motor reaction time (VMRT) plays a crucial role in high-speed operational environments such as driving, aviation, and industrial hazard response. Traditional augmented reality systems rely mainly on reactive cueing, which cannot mitigate inherent neural processing delays. This research proposes a predictive augmented reality framework integrating real-time eye-tracking with gaze-intent forecasting to proactively display visual cues before motor response initiation. The proposed system utilizes a transformer-based gaze prediction model capable of forecasting user intent up to 500 milliseconds ahead. A latency-optimized augmented reality pipeline maintains end-to-end latency below 15 ms to ensure real-time responsiveness. A user study involving 30 participants demonstrated that the predictive AR system reduced visual-motor reaction time by approximately 18.7% and reduced cognitive workload by about 22% compared to baseline AR systems. These results indicate that predictive AR guided by gaze intent can significantly enhance human performance and situational awareness in high-speed environments by reducing reaction latency and improving hazard anticipation.
Sharma et al. (2026) studied this question.