Skip to main content

ClientSDK Changelog

This page tracks the version history and changes made to the Aria Gen2 ClientSDK.

Version 2.4.0​

Release Date: [TBD — fill at release]

New Features​

Multi-Device Time-Aligned Recording & Streaming​

Orchestrate multiple Aria Gen 2 devices from a single CLI invocation. Devices receive coordinated start/stop, and one device acts as a time-domain-mapping broadcaster so receivers can align timestamps to a shared clock.

Key additions:

  • recording start --all / streaming start --all — discover and orchestrate every connected device
  • --broadcaster <serial> — override the default broadcaster (smallest-serial wins by default)
  • --disable-time-domain-mapping — opt out of TDM when only coordinated start/stop is needed
  • recording stop / streaming stop — auto-stop all connected devices when no --serial is given
  • recording sessions list / info and streaming sessions list / info — inspect multi-device sessions
  • recording download --session <id> — download every recording in a session into per-device subdirectories
  • Sessions persisted as JSON at ~/.aria/multi-device-session/<id>.json

See Multi-Device Recording & Streaming for the full guide.

Country-aware Time-Domain-Mapping Channel Validation​

The CLI now validates the time-domain-mapping channel against the connected glasses' regulatory region before forwarding it to the device, mirroring what the Aria companion app enforces.

Key additions:

  • ETSI countries permit channels 0–2; FCC countries permit channels 10–40
  • Channel auto-picked per region when --time-domain-mapping-channel is omitted (ETSI → 0, FCC → 10)
  • Multi-device sessions infer an empty country from any sibling that does report one (logged as a WARNING); cross-region mismatches are surfaced as a hard error
  • Country code source: each device's wifi_country_code from HmdStatusResponse, surfaced via the new aria::sdk::DeviceStatus.countryCode field
  • Behavioral break: existing scripts that relied on the silent default of channel 0 in FCC regions will now error out with an actionable message pointing to the right channel

Per-Device Identity in Callbacks​

Optional device_id parameter on every Python streaming callback. Auto-detected via inspect.signature() — declare device_id in your callback to receive the device serial; omit it to keep existing single-device behavior unchanged.

Key additions:

  • device_id: str | None keyword argument on RGB, SLAM, ET, audio, IMU/IMU batch, magnetometer, GPS, phone-location, PPG, barometer, Bluetooth/WiFi beacon, eye-gaze, hand-pose, time-domain-mapping callbacks
  • handler.device_serial property on StreamDataInterface
  • See Multi-Device Streaming Example for the pattern

Multi-Device Streaming Example​

New Python tool multi_device_streaming_example.py — a headless HTTPS server that demonstrates how to receive RGB frames and time-domain-mapping offsets from multiple Aria Gen 2 devices in a single process and disambiguate them per device.

Key additions:

  • Concrete reference for the new opt-in device_id callback parameter
  • Per-device aggregation pattern (thread-safe dicts keyed by serial; threading.Lock for cross-callback state)
  • Time-domain-mapping offset summary printed on shutdown for sanity-checking sync quality
  • Use as a starting point for downstream tools (notebooks, custom dashboards, perception pipelines) that need per-device frames in one process
  • See Multi-Device Streaming Example for the walkthrough

Multi-Device Streaming Viewer​

New Python tool multi_device_streaming_viewer.py — a live Rerun visualization that receives RGB streams from N Aria Gen 2 devices, applies TDM-corrected timestamps for synchronized cross-device replay, and rebuilds its panel layout as devices connect and reconnect.

Key additions:

  • mDNS-resolved HTTPS server (oatmeal_server.local:6768) shared across all devices
  • TDM activation validation with per-device timing windows; fails fast on a broken receiver before the user spends minutes wondering why the timeline is desynced
  • --broadcaster <serial> to override the auto-detected broadcaster
  • --expected-device-count <N> strict mode for scripted/CI runs
  • --skip-tdm-validation escape hatch for partial-sync investigations
  • Inherits the persistent shared-cert workflow from streaming install-certs
  • See Multi-Device Recording & Streaming for usage

Version 2.3.0​

Release Date: Apr 2026

New Features​

Multi-Device Support​

Refactored device connectivity to support connecting multiple Aria Gen 2 devices simultaneously.

Key additions:

  • DeviceTarget struct bundling IP and serial for streamlined device identification
  • usb_network_devices() returns List[DeviceTarget] for direct use with connect()
  • CLI supports --serial flag to specify a target device
  • Auto-prompts when multiple devices are detected

Real-Time Latency Measurement​

Added real-time latency measurement to the aria_streaming_viewer.

Key additions:

  • Re-enabled TicSync time synchronization
  • TimeSyncRef Python binding with compute_latency_ms(), is_connected(), and is_valid() methods
  • CLI flags: --show-latency and --latency-offset-ms

Multi-Device Streaming Cert Install​

The CLI streaming install-certs command now supports batch certificate installation across multiple devices.

Key additions:

  • --device option for specifying target devices
  • Certificates generated once and installed on all specified devices

Recording List API​

listRecordings() now returns List[RecordingInfo] with uuid, size, and metadata, enabling Python scripts to directly retrieve and process recording lists.

Recording State Query API​

Added APIs for querying the current recording state of a device.

Key additions:

  • is_recording() and get_recording_state() APIs
  • RecordingState data class with is_recording, recording_name, start_time_ms, and profile_name fields
  • Enables Aria Studio and Companion App recording state synchronization

Thumbnail Download API​

Added APIs for downloading recording thumbnails.

Key additions:

  • download_thumbnail(uuid, output_path) for downloading a single thumbnail
  • download_all_thumbnails(uuid, output_dir) for batch downloading
  • Thumbnail naming matches VRS files (e.g., example_recording_20230101_123456_thumbnail_0.jpeg)

Device Profiles API for Python​

Added device_profiles() API to the Python SDK for retrieving device recording profiles.

Online Calibration Streaming​

VIO data stream now includes real-time Online Calibration state.

Key additions:

  • FrontendOutput.online_calib containing:
    • Camera projection model parameters (real-time updated intrinsics)
    • IMU biases (accelerometer/gyroscope offsets)
  • Full Python SDK support for OnlineCalibState, ProjectionModelParameters, and ImuMeasurementModelParametersFloat

Custom Streaming Profile Support​

The device_streaming.py example now supports custom streaming profiles.

Key additions:

  • --profile-name argument for selecting a named profile
  • --custom-profile-path argument for using a custom JSON profile

Version 2.2.0​

Release Date: Dec 2025

New Features​

macOS Support​

ClientSDK now supports macOS, enabling developers on Apple hardware to work with Aria Gen2 devices natively.

Key additions:

  • Full Python SDK support on macOS (Apple Silicon and Intel)
  • Hardware-accelerated image decoding for improved performance on macOS
  • Native USB network connection support via Companion App pairing
  • All CLI tools (aria_gen2, aria_doctor) available on macOS

Supported platforms:

  • macOS 13 (Ventura) and later
  • Python versions: 3.10 - 3.12

Wireless Streaming​

Added support for wireless data streaming from Aria Gen2 devices over WiFi, providing greater flexibility and mobility during development and testing.

Key capabilities:

  • WiFi Router Streaming: Stream data from Aria devices connected to the same WiFi network as your development machine
  • On-Device Hotspot Streaming: Connect directly to Aria's built-in WiFi hotspot for standalone wireless streaming without requiring external network infrastructure

Benefits:

  • Untethered development workflows - no USB cable required for streaming
  • Improved mobility for testing in various environments
  • Simplified setup for multi-device scenarios

Resources:

Version 2.1.0​

Release Date: Nov 2025

⚠️ Breaking Changes​

Breaking Change

Device OS V14 (version: 51021960009400500) and V15 (version: 51041610004100500) Compatibility Update

ClientSDK 2.1.0 introduces a breaking change for recording operations with devices running OS V14 only:

  • Recording commands now require the --disable-tls flag when interacting with Aria devices running OS V14
  • Existing scripts and workflows that perform recording operations must be updated to include this flag
  • This change only affects recording functionality; streaming and other operations remain unchanged

OS V15 Compatibility: Devices running OS V15 are fully compatible with ClientSDK 2.1.0 and do not require the --disable-tls flag. Recording commands work normally without any modifications.

Example (OS V14 only):

# Old command (will fail with OS V14 devices)
aria_gen2 recording list

# New command (required for OS V14 devices)
aria_gen2 recording list --disable-tls

# OS V15 devices (no flag needed)
aria_gen2 recording list

Migration:

  • OS V14 devices: Update all recording-related CLI commands and Python scripts to include the --disable-tls flag
  • OS V15 devices: No changes required - all commands work as before

New Features​

ROS2 Integration Examples​

Added comprehensive ROS2 integration support for advanced users, enabling seamless integration of Aria device data with ROS2 robotics workflows.

Key additions:

  • Custom ROS2 message type (AriaRaw) for transporting raw Aria sensor data
  • Publisher example for streaming Aria data to ROS2 topics
  • Subscriber example for receiving and processing Aria data in ROS2 nodes
  • Support for all sensor types (SLAM cameras, RGB camera, IMU, VIO, eye gaze, hand tracking, etc.)
  • Device calibration sharing across ROS2 nodes

Resources:

Enhanced Data Converter API​

Expanded the OssDataConverter class with additional functionality for converting raw FlatBuffer messages to typed Python data structures.

New capabilities:

  • Message Type Utilities: Added MessageType.to_string() and MessageType.from_string() methods for converting between message IDs and their string representations

    # Convert message ID to string
    name = sdk_gen2.MessageType.to_string(0x8010)  # "SLAM_CAMERA_FRAME"

    # Convert string to message ID
    msg_id = sdk_gen2.MessageType.from_string("IMU_EVENT")  # 0x8020

  • SharedMessage Constructor: Enhanced SharedMessage class with constructor for creating messages programmatically

    msg = sdk_gen2.SharedMessage(
        id=sdk_gen2.MessageType.IMU_EVENT,
        payload=b'\x00\x01\x02\x03',
        timestampNs=1234567890
    )

  • Additional Sensor Conversions: Expanded support for converting various sensor message types:

    • Location sensors: GNSS, phone location
    • Environmental sensors: barometer, magnetometer
    • Wireless beacons: bluetooth (BLE) and WiFi
    • Machine perception: eye gaze, hand tracking, VIO (Visual-Inertial Odometry)
    • Other sensors: audio, PPG (photoplethysmogram)

Resources:

Improvements​

Data Converter Performance Optimization​

Significantly improved data processing efficiency by leveraging native FlatBuffer serialization from Aria Gen2 devices:

  • Reduced serialization overhead: The data converter now directly processes FlatBuffer-serialized data from the device, eliminating redundant serialization/deserialization steps

This optimization maintains full API compatibility while delivering better performance across all sensor data types.

Documentation & Developer Experience​

  • Enhanced documentation with detailed code walkthroughs for ROS2 examples
  • Added troubleshooting guides for common ROS2 integration issues
  • Improved message type handling with string conversion utilities

Version 2.0.0​

Release Date: Oct 2025

Initial Release​

ClientSDK is a comprehensive toolbox designed to facilitate control and interaction with Aria Gen2 devices. This initial release provides a complete set of libraries, utilities, and example code to streamline device management, authentication, recording, and streaming.

Key Features​

Device Management

  • Core classes and utilities for connecting to, authenticating, and controlling Aria Gen2 devices
  • USB network connection support via Companion App pairing
  • Device discovery and connection management

Streaming Support

  • Components for starting and managing data streams from Aria Gen2 devices
  • On-device machine perception data streaming (VIO, hand pose, eye gaze)
  • Raw sensor data streaming from all device sensors (cameras, IMU, audio, etc.)
  • HTTP-based streaming architecture with certificate support

Recording Support

  • Components for starting and managing on-device recording
  • Recording configuration with predefined profiles
  • Recording management (list, download, delete)
  • VRS format output

Security

  • Built-in support for streaming certificates and security options
  • Secure device authentication via mobile Companion App
  • SSL/TLS support for streaming connections

Python & CLI Tools

  • Python SDK with intuitive APIs for device control and data streaming
  • Command-line utilities for scripting and automation (aria_gen2, aria_doctor)
  • Comprehensive device control commands (authentication, recording, streaming, device info)

Example Code

  • Ready-to-use Python examples demonstrating common workflows
  • Sample code for device connection, authentication, recording, streaming, and text-to-speech
  • Comprehensive documentation with setup guides and tutorials

Prerequisites​

  • Hardware: Aria Gen2 device
  • Mobile App: Aria Companion App (installed and paired with device)
  • Connection: USB cable for device-to-PC connection

Supported Platforms​

  • Linux (x64): Fedora 40/41, Ubuntu 22.04 LTS, Ubuntu 24.04 LTS
  • Python versions: 3.10 - 3.12

Support​

For issues, questions, or feature requests, please visit: