Available Data Types
RoboX collects sensor data through smartphone sensors and (in the future) wearable devices.
Visual Data
RGB Camera
Standard color video from the device's rear camera.
Resolution
Up to 4K (varies by device and campaign)
Frame Rate
30fps standard, 60fps available
Format
H.264 or H.265 encoded
FOV
Device-dependent, typically 70-120°
Capture Considerations
Automatic exposure and white balance
Stabilization applied when available
Anonymization (face/plate blurring) applied on-device
Metadata includes device model and lens characteristics
Use Cases
Visual perception and object recognition
Scene understanding and classification
Visual odometry and SLAM
Imitation learning from demonstrations
Depth / LiDAR
3D depth information from LiDAR-equipped devices (iPhone 12 Pro and later) or Time-of-Flight sensors.
Range
0.5m to 5m typical
Resolution
Device-dependent
Frame Rate
Up to 60Hz
Format
Point cloud or depth map
Capture Considerations
Works in all lighting conditions (active sensing)
May have interference in direct sunlight
Limited range compared to industrial LiDAR
Paired with RGB for color point clouds
Use Cases
3D scene reconstruction
Obstacle detection and avoidance
Spatial mapping and room scanning
AR applications and occlusion
Motion Data
IMU (Inertial Measurement Unit)
Combined accelerometer and gyroscope data capturing device motion.
Accelerometer Range
±16g
Gyroscope Range
±2000°/s
Sample Rate
100-200Hz
Format
Timestamped 3-axis readings
Capture Considerations
High frequency captures fine motion details
Requires calibration for drift correction
Gravity component included in accelerometer readings
Device orientation affects axis mapping
Use Cases
Motion pattern recognition (walking, running, driving)
Dead reckoning / inertial navigation
Activity classification
Gesture recognition
Device placement detection
Magnetometer
Compass/heading information from the device's magnetic sensor.
Range
±4800µT
Sample Rate
50-100Hz
Format
3-axis magnetic field readings
Capture Considerations
Affected by nearby metal and electronics
Requires calibration for accurate heading
Indoor readings less reliable than outdoor
Combined with accelerometer for orientation
Use Cases
Heading/compass direction
Indoor positioning augmentation
Magnetic field mapping
Device orientation determination
Position Data
GPS / GNSS
Satellite-based positioning data.
Accuracy
3-10m typical (device-dependent)
Update Rate
1Hz standard
Format
Latitude, longitude, altitude
Privacy Processing
Raw coordinates are never uploaded. Position data undergoes:
Grid quantization (~100m cells)
One-way hashing
Temporal offset (±15 min)
Only anonymized grid hashes are transmitted.
Use Cases
Outdoor localization
Route and trajectory analysis
Geographic coverage mapping
Movement pattern analysis
Wi-Fi RSSI (Received Signal Strength Indicator)
Signal strength from nearby Wi-Fi access points.
Range
-100 to 0 dBm
Scan Rate
Campaign-dependent (limited by OS)
Format
BSSID (hashed), RSSI, timestamp
Capture Considerations
Access point identifiers are hashed, not raw MAC addresses
Scan frequency limited by iOS power management
Signal strength varies with device model
Indoor environments provide richer data
Use Cases
Indoor positioning and localization
Fingerprint-based navigation
Building/floor detection
Coverage mapping
Bluetooth / BLE RSSI
Signal strength from nearby Bluetooth devices and beacons.
Range
-100 to 0 dBm
Scan Rate
Campaign-dependent
Format
Device ID (hashed), RSSI, timestamp
Capture Considerations
Beacon detection provides precise indoor landmarks
Consumer device detection is less consistent
UWB (where available) provides higher accuracy
Combined with Wi-Fi for robust indoor positioning
Use Cases
Beacon-based indoor navigation
Proximity detection
Retail and venue analytics
Indoor localization augmentation
Environmental Data
Barometer
Atmospheric pressure measurement.
Range
300-1100 hPa
Resolution
0.01 hPa
Sample Rate
1-10Hz
Format
Pressure in hPa, timestamp
Capture Considerations
Highly accurate for relative altitude changes
Affected by weather (need calibration for absolute altitude)
Excellent for floor/level detection
Can detect elevation changes of ~1 meter
Use Cases
Floor-level detection in buildings
Altitude tracking for drones
Stair/elevator detection
Terrain variation mapping
Microphone / Audio
Environmental sound capture for acoustic mapping.
Sample Rate
44.1kHz or 48kHz
Format
AAC or WAV
Duration
Campaign-defined (typically 30-60 second clips)
Privacy Processing
Audio undergoes on-device processing:
Speech detection using ML model
Speech segments replaced with ambient noise
Amplitude normalization
Only processed audio uploaded
No intelligible human speech leaves the device.
Use Cases
Environmental acoustic mapping
Noise level assessment
Audio-based localization
Drone routing (noise avoidance)
Sound source detection
Ambient Light
Light sensor readings for environmental brightness.
Range
0 to 100,000+ lux
Sample Rate
1-10Hz
Format
Lux value, timestamp
Capture Considerations
Sensor location on device affects readings
Affected by device orientation
Useful for context detection (indoor/outdoor, time of day)
Use Cases
Indoor/outdoor classification
Lighting condition assessment
Context-aware behavior adaptation
Camera exposure optimization
Data Synchronization
All sensor streams are timestamped to enable synchronization:
Hardware timestamps where available
Monotonic clock for relative timing
NTP synchronization for absolute time
Interpolation support for mismatched sample rates
Data packages include synchronization metadata allowing accurate alignment of multi-modal data.
Last updated