A Novel Image and Audio-based Artificial Intelligence Service for Security Applications in Autonomous Vehicles
Dimitrios Tsiktsiris, Anastasios Vafeiadis, Antonios X. Lalas, Minas Dasygenis, Konstantinos Votis, Dimitrios K. Tzovaras · Transportation research procedia · 2022
Autonomous Vehicles (AVs) can potentially reduce the accident risk while a human is driving. They can also improve the public transportation by connecting city centers with main mass transit systems. Creating a system that can provide a sense of security to the passenger, when the driver is missing, remains a challenging task. In this work, an image and audio-based approach, supported by novel Artificial Intelligence (AI) algorithms, is proposed as a service to increase the security and trust inside an autonomous shuttle. The two modalities, running in real-time, can detect petty crimes scenarios such as screaming, bag snatching, people fighting and vandalism and enable notifications to authorized personnel for proper actions. The proposed solution is deployed on a Jetson AGX Xavier to favor power efficiency and seamless integration and achieves up to 96% accuracy. Thus, the envisioned system exhibits high potential for transforming security and safety in emerging autonomous public transportation infrastructure.