Virtual Reality Simulator for Police Training with AI-Supported Cover Detection.

Training of different scenarios is essential to prepare emergency services for real-world situations. In this paper, we introduce a virtual reality (VR) simulator specifically designed to enhance the training of police officers in the context of apartment search and area securing, including the evaluation of potential threats or hazards. At this point, multiple police officers train simultaneously by employing virtual reality headsets and full-body suits, enabling them to see and interact with each other’s avatars. We additionally developed a configuration tool to enable operational trainers to easily create new buildings with custom interiors. Moreover, we developed an algorithm that automatically recognizes possible dangerous areas, especially when police officers have no cover. In this case, a non-player character is spawned, representing either neutral individuals or potential assailants, strategically placed to confront police officers and accentuate tactical mistakes. Field tests have substantiated the efficacy of our approach, demonstrating a high F 1 score of 0.93 in identifying tactical mistakes made by police officers. Furthermore, the training outcomes have revealed that virtual training is comparable to traditional training in terms of maximum recovered area. These findings demonstrate the immense potential of VR-based simulators for police training, as they enable the creation of various scenarios and thus increase the number of training sessions. The simulator is particularly attractive also due to alternate training scenarios with individual difficulty levels making it suitable for students and special units.