An Immersion Campus Simulation Protection and Safety Improvement System using Artificial Intelligence and Machine Learning

Eric P. Xing, Laurie Delinois2025 · DOI

Abstract

This project addresses the critical need for immersive crisis-response training to enhance individuals' ability to act effectively under high-stress conditions, such as fires and active shooter events. Traditional training methods often lack the depth and adaptability required for effective skill retention. To bridge this gap, we developed a virtual reality (VR) program that simulates realistic crisis environments using Unity’s physics, collider interactions, and coroutines for dynamic fire propagation and evasive maneuvers. Key components include scenario selection, interactive obstacle navigation, and outcome evaluation, all designed to offer users an engaging training experience. Initial challenges, such as optimizing control responsiveness, were resolved by incorporating an FPS controller, ensuring accurate and responsive movement. Experimental trials across scenarios demonstrated notable improvements in user-response accuracy and engagement, with FPS controllers outperforming standard simulators. This VR training tool offers substantial benefits to schools, workplaces, and communities by providing accessible, impactful training that strengthens emergency preparedness.

Introduction

In high-pressure environments—such as a school fire or an active shooter situation—split-second decision-making is often the difference between safety and tragedy. Traditional training methods, including instructional videos or static drills, frequently fail to replicate the physiological stress and dynamic unpredictability inherent in a real crisis. This lack of "functional realism" often results in poor skill retention when individuals are faced with actual emergencies.

This research aims to bridge this gap by creating a scalable, immersive VR environment where users can practice survival skills in a safe, controlled setting. A primary technical challenge involves balancing simulation fidelity—such as realistic fire spread and adaptive AI behavior—with system performance to ensure a fluid and effective training experience.

System Methodology

The system was developed using the Unity engine, focusing on a structured user experience that transitions from initial setup to active simulation.

The high-level system workflow, illustrating the progression from the start screen to scenario selection and execution.

Technical Implementation

The platform's effectiveness relies on three primary technical pillars:

  1. Dynamic Fire Propagation: Rather than utilizing static "fire zones," the system employs Physics.OverlapSphere to detect flammable objects in real-time. To maintain high performance and avoid simulation lag, the authors implemented C# coroutines to manage fire spread over time. An "ignited" tagging system is used to prevent redundant computational checks on objects already on fire.
Code snippet demonstrating the use of coroutines and physics-based overlap checks for efficient fire propagation.
  1. AI and Evasion Logic: In active shooter scenarios, the system uses distance-check algorithms to govern NPC behavior. An AI character follows a predetermined path but transitions to a "chase" state if the player enters a specific radius while not in a designated "hiding spot."
  2. Control Scheme Optimization: A significant portion of the research focused on the "human-in-the-loop" aspect. The study compared a standard Device Simulator against a dedicated First-Person Shooter (FPS) Controller to determine which provided the necessary responsiveness for crisis training.

Results and Discussion

Experimental trials with 10 participants across fire evacuation and active shooter scenarios yielded significant quantitative benchmarks. The mean completion time for scenarios was 25.4 seconds, with the FPS controller facilitating significantly faster and more accurate movements compared to standard simulators.

Participants achieved an average success rate of 81.5% (median 88.5%). Subjective feedback was also positive, with the simulation receiving a high average realism rating of 4.1 out of 5. Further analysis showed that "Control Precision" (rated 7.1/10) directly correlated with higher user satisfaction and immersion (rated 7.8/10). These results suggest that control responsiveness is a non-negotiable factor in the effectiveness of VR safety training.

Conclusion

While the current system demonstrates the potential of VR for campus safety, several hurdles remain for widespread adoption. High hardware costs for VR setups are a primary limitation. Future work will explore "lightweight" software alternatives for budget-friendly headsets and more robust AI systems capable of adapting to complex user behaviors. Additionally, integrating post-scenario analytics will provide personalized feedback to further enhance the learning outcomes of the tool.

Summary updated

Abstract

This project addresses the critical need for immersive crisisresponse training to enhance individuals' ability to act effectively under high-stress conditions, such as fires and active shooter events [1]. Traditional training methods often lack the depth and adaptability required for effective skill retention. To bridge this gap, we developed a virtual reality (VR) program that simulates realistic crisis environments using Unity’s physics, collider interactions, and coroutines for dynamic fire propagation and evasive maneuvers [2]. Key components include scenario selection, interactive obstacle navigation, and outcome evaluation, all designed to offer users an engaging training experience. Initial challenges, such as optimizing control responsiveness, were resolved by incorporating an FPS controller, ensuring accurate and responsive movement [3]. Experimental trials across scenarios demonstrated notable improvements in userresponse accuracy and engagement, with FPS controllers outperforming standard simulators. This VR training tool offers substantial benefits to schools, workplaces, and communities by providing accessible, impactful training that strengthens emergency preparedness.
BibTeX
@misc{xing2025immersion,
  author = {Eric P. Xing and Laurie Delinois},
  title = {{An Immersion Campus Simulation Protection and Safety Improvement System using Artificial Intelligence and Machine Learning}},
  year = {2025},
  doi = {10.5121/csit.2024.150102},
}
Figure