VDEgo-C2 — Professional Egocentric Multimodal Data Collection System

The VDEgo-C2 is VIRDYN’s purpose-built egocentric data acquisition device, engineered for researchers, AI engineers, and institutions who need high-quality first-person-perspective multimodal datasets for training vision-language-action (VLA) models, world models, and embodied intelligence systems.

Key Features

Binocular Global Shutter Fisheye Camera
Equipped with dual global shutter image sensors (pixel size 3.0μm × 3.0μm) and ultra-wide fisheye lenses with diagonal FOV ≥160°, the VDEgo-C2 captures full stereo egocentric video at up to 3840×1200 @ 60fps — covering a broad field of view ideal for first-person perception tasks.

Precision Hardware Frame Synchronization
Hardware-level binocular sync ensures a stereo frame time offset of ≤1μs, while binocular-to-IMU synchronization is maintained within <50μs. This tight temporal alignment is critical for accurate depth estimation, visual odometry, and sensor fusion pipelines.

High-Rate 6-Axis IMU
The integrated 6-axis inertial measurement unit (accelerometer + gyroscope) runs at 300Hz to 600Hz sampling rate, delivering precise motion and orientation data timestamped and synchronized with each video frame.

Offline Standalone Capture with Encryption
Operates fully independently without a connected PC — simply press the onboard control button to start and stop recording, with audio prompts for confirmation. All captured data is encrypted locally, with an offline decryption and trajectory computation tool provided.

Web-Based Real-Time Control
When connected to the same WiFi network, any PC or mobile browser can access the built-in web interface for live preview, capture control, data upload/download, and device configuration — no software installation required.

WiFi 6 + Bluetooth 5.4 Connectivity
Equipped with WiFi 6 (with external antenna) and Bluetooth 5.4 for high-speed wireless data transfer and seamless integration with multi-device workflows.

High-Performance Computing Core
Powered by a high-performance hybrid 8-core processor with 4GB RAM and 128GB onboard storage, capable of handling real-time compression, encryption, and multi-stream data logging simultaneously.

Hot-Swappable Battery System
The 5000mAh rechargeable lithium battery supports rapid replacement for uninterrupted long-duration capture sessions. Built-in IC protection covers over-current (OCP), over-voltage (OVP), short-circuit (SCP), and over-temperature (OTP) — four-layer safety mechanism. Compatible with standard 5V/2A or 5V/3A chargers, and supports charge-while-use via USB-C.

Technical Specifications

Output Data Package

Every capture session produces a complete, structured data package:

• video.avi — Binocular stereo video (left+right stitched, MJPEG high-quality compression)
• video_frames_timestamp.csv — Per-frame timestamps for precise temporal alignment
• imu.csv — IMU timestamps, accelerometer and gyroscope readings
• trajectory.csv — Egocentric pose data (position coordinates + quaternion)
• calibration.json — Full camera and IMU calibration parameters (resolution, intrinsics, extrinsics, vignetting)

Use Cases

• EGO First-Person Dataset Collection: Capture rich egocentric visual and motion data for human activity recognition, scene understanding, and embodied AI research.
• VLA Model Training: Collect synchronized multimodal demonstrations for Vision-Language-Action model fine-tuning and behavioral cloning.
• World Model Research: Gather high-fidelity first-person video and trajectory data to train predictive world models.
• Academic & University Vision Research: Ideal for labs working on visual SLAM, optical flow, depth estimation, and sensor fusion.

What’s in the Box