When VR Shooters Meet Discrete Events: Training Security Policies Without Endless Human Trials
Training a security policy sounds simple until the training data involves people role-playing traumatic emergencies inside a virtual school. That is the uncomfortable starting point of this paper. Virtual reality can help researchers study rare and dangerous events under controlled conditions, but it does not solve the scaling problem. Every new intervention, policy variation, or robot behavior still needs another human-subject experiment. That is slow, expensive, ethically constrained, and not exactly a cheerful afternoon in the lab. ...