RoboMNIST: A Multimodal Dataset for Multi-Robot Activity Recognition Using WiFi Sensing, Video, and Audio

This dataset is designed for multi-robot activity recognition (MRAR) and integrates data from two Franka Emika robotic arms using three modalities: WiFi channel state information (CSI), video, and audio. The dataset includes synchronized data from three cameras, three WiFi sniffers, and three microphones, offering a rich multimodal resource for enhancing robotic perception, decision-making, and autonomous operations in dynamic environments.