An Enhanced Time-Attention-Based Social GAN for Pedestrian Trajectory Prediction and Risk Alert System Utilizing Edge Computing Device

The emergence of smart cities and edge computing has made pedestrian trajectory prediction essential for enhancing traffic safety. This study integrates Nvidia's Jetson AGX Xavier with a social generative adversarial network (GAN) augmented by a temporal attention mechanism. By leveraging edge computing, this approach facilitates real-time processing, reduces latency, decreases dependence on cloud services, and enhances data privacy. The time-attention mechanism effectively captures dynamic pedestrian interactions, improving prediction accuracy. Experimental results indicate that this method outperforms traditional approaches in accuracy and efficiency. These findings underscore the significance of edge computing in enabling real-time decision- making and the critical role of temporal attention in predicting behavior in dynamic environments, ultimately benefitting urban traffic management and enhancing pedestrian safety.