Warehouse MultiCam RF Dataset

This dataset was generated for a warehouse environment using the Gazebo and Sionna simulators. The Gazebo and Sionna scene models are geometrically identical to ensure physically consistent RF ray tracing and multimodal synchronization between wireless communication channels, robot motion, and visual observations. The dataset contains synchronized data from 16 fixed RGB cameras distributed throughout the warehouse environment. Each camera is associated with a transmitter antenna, while an additional transmitter antenna is mounted on a mobile robot navigating through the scene. A base station (BS) receiver equipped with a 16×16 antenna array and 16 OFDM subcarriers is used to capture wireless channel information. For every timestamp, the dataset provides: synchronized multi-camera RGB images, RF channel measurements stored as NumPy tensors, robot position and orientation, linear and angular velocities, camera configuration metadata, timestamp-level synchronization information. The dataset is designed to support research in: RF sensing and localization, multimodal learning, wireless communication, embodied AI, robotics and navigation, sensor fusion, world models, RF-vision alignment, and simulation-to-real transfer research. A key contribution of this dataset is the strict geometric alignment between the Gazebo and Sionna environments, enabling accurate correspondence between visual observations and RF propagation characteristics for multimodal AI and wireless robotics research. Environment Alignment Geometrically aligned Gazebo and Sionna warehouse environments used for synchronized RF and visual data generation.