SensorStream, powered by ROS 2 vs EmbodiData

SensorStream turns your iPhone into a sensor platform, streaming LiDAR depth, RGB camera, IMU, and GPS data to your host computer over Wi-Fi or USB. Built for robotics prototypers, researchers, or anyone who wants to use the iPhone's powerful built-in sensor suite. SENSOR STREAMS • LiDAR Depth — 256×192 depth frames at up to 30 FPS from front and back (Wide) cameras (requires a LiDAR-equipped iPhone) • RGB Camera — JPEG-compressed colour frames at up to 30 FPS from the wide-angle camera, or front-facing camera • IMU — accelerometer, gyroscope, and quaternion orientation data at a configurable frequency • GPS — GNSS position TRANSPORT OPTIONS • Wi-Fi (TCP) — connect wirelessly over your local network; scan the server's QR code from the app for instant pairing • USB — low-latency wired connection via PeerTalk / iproxy for demanding real-time applications INTEGRATION • Driver for integration with ROS 2 — a companion driver publishes sensor data directly to ROS 2 topics (/depth_image, /color_image/compressed, /imu/data, /gps/fix) using sensor_msgs message types ON-DEVICE PREVIEW • Real-time depth overlay with adjustable min/max range and colour thresholds • Live camera preview • IMU visualiser for orientation feedback • Per-sensor toggles and settings — configure each sensor independently before streaming HOW IT WORKS 1. Start the companion driver on your computer 2. Open SensorStream, go to Settings → Network, and connect via Wi-Fi (scan the QR code or enter IP/port manually) or USB. 3. Enable the sensors you need on the Dashboard. 4. Tap Start Streaming, that's it. REQUIREMENTS • iPhone with iOS 15.6 or later • LiDAR sensor required for depth streaming (iPhone 12 Pro and later) • Host computer running Python 3.9+ and ROS 2 Humble for the companion server

EmbodiData is an iPhone-based embodied AI data capture tool. It allows users to record first-person video, device motion signals, and structured session metadata during user-started capture sessions. The app is designed for embodied AI research, robotics data collection, and multimodal dataset preparation. Captured data can be reviewed locally and exported by the user for downstream analysis and model development. EmbodiData only records during user-initiated sessions and provides clear controls for starting, stopping, reviewing, and exporting captured data.