Choosing Your Motion Data Output: Raw, Quaternion, BVH and CSV

Choosing the Right QSense Motion Data for Your Application

QSense Motion sensors are designed to give you full control over what data you receive, how it is processed, and how it is exported. Depending on your application, you may want raw inertial signals, real-time orientation, derived kinematics, or motion capture–ready formats.

To support this flexibility, QSense offers:

• Multiple real-time streaming modes
• Multiple data representations from the same recording
• Standardized export formats for analysis, visualization, and integration

This page explains how these options work and how to choose the right configuration.

Overview: From Raw Signals to Motion Representation

QSense IMU sensors measure motion at the lowest physical level, but do not force you into a single abstraction layer.

From one measurement session, you can work with:

• Raw IMU signals (accelerometer, gyroscope, magnetometer)
• Processed orientation (6-DoF or 9-DoF quaternions)
• Derived motion quantities (e.g. linear acceleration, gravity vector)
• Kinematic motion formats (e.g. BVH skeleton motion)

All outputs share a common, time-synchronized clock, enabling direct comparison and validation across representations.

Real-Time IMU Data Streaming Modes

QSense supports four streaming modes. Each mode defines what data is transmitted per packet and how bandwidth is allocated.

Mixed Mode — IMU + Orientation Combined

Best for: general motion tracking, biomechanics, robotics, exploratory research

Mixed Mode provides both low-level and high-level motion information:

• Raw accelerometer, gyroscope, magnetometer samples
• Quaternion orientation (6- or 9-DoF)
• Free (gravity-compensated) acceleration

This mode is useful when you want orientation for interpretation but still retain access to raw signals for verification, filtering, or later processing.

Raw Mode — Maximum IMU Data Access

Best for: algorithm development, machine learning, validation studies

Raw Mode streams only unprocessed sensor data:

• Accelerometer, gyroscope, magnetometer
• Maximum available resolution and sampling rate
• No onboard sensor fusion applied

Use this mode when you:

• Develop your own fusion or estimation algorithms
• Train ML models on inertial data
• Need full transparency over signal processing

Quaternion Mode — Orientation-Only Streaming

Best for: VR/AR, robotics, real-time control, visualization

Quaternion Mode streams orientation only:

• High-frequency quaternion samples
• No raw IMU signals transmitted

This minimizes bandwidth and simplifies integration when orientation is the primary output, such as:

• Avatar control
• Head or limb tracking
• Closed-loop feedback systems

Optimized Mode — High-Rate Orientation + 6-DoF IMU

Best for: sports science, motion capture, real-time analytics

Optimized Mode balances detail and efficiency:

• High-frequency quaternions
• Raw accelerometer + gyroscope
• No magnetometer data

By excluding magnetometer data, this mode:

• Reduces sensitivity to magnetic disturbances
• Increases effective bandwidth
• Is well-suited for dynamic, indoor, or metal-rich environments

Streaming Mode Summary

Mode	Best for	Includes
Mixed	General motion analysis	IMU + quaternions + free acceleration
Raw	Custom processing & ML	Full raw IMU
Quaternion	Orientation-focused apps	High-rate quaternions
Optimized	Balanced real-time motion	Quaternions + accel/gyro

Data Logging & Multiple Output Representations

QSense recordings are not limited to a single output type.

A single measurement session can contain multiple synchronized representations, allowing you to:

• Inspect raw signals
• Use processed outputs for analysis
• Validate algorithms against the same dataset

Available Data Representations

Raw IMU Data

Time-series data from:

• Accelerometers
• Gyroscopes
• Magnetometers (if enabled)

Use cases:

Signal processing, filtering, validation, ML training, custom estimation

Orientation Data (6-DoF / 9-DoF Quaternions)

Orientation estimates derived via onboard or post-processing sensor fusion.

Use cases:

3D visualization, VR/AR, robotics, posture and segment orientation analysis

Derived Motion Quantities

Pre-calculated values such as:

• Linear (gravity-free) acceleration
• Gravity vector
• Angular velocity

Use cases:

Biomechanics, real-time feedback, reduced-complexity pipelines

Motion & Kinematic Formats (BVH)

QSense supports export to BVH (Biovision Hierarchy) format.

BVH represents motion as:

• A skeletal hierarchy
• Time-varying joint rotations

Typical use cases:

• Motion capture analysis
• Biomechanics and ergonomics
• Animation and avatar pipelines
• Interoperability with third-party MoCap tools

QSense BVH files:

• Are human-readable
• Import directly into standard BVH viewers
• Preserve consistent timing across joints and segments

BVH Motion Output

BVH output allows QSense data to integrate into existing motion capture ecosystems without proprietary tooling.

Typical applications include:

• Comparing IMU-based motion to optical MoCap
• Rapid visualization of full-body motion
• Feeding motion into animation or simulation environments

Example Dataset

The example dataset provided demonstrates:

• A minimal sensor configuration
• Synchronized raw IMU and processed outputs
• Identical timestamps across representations

It is intended for:

• Evaluating data structure
• Testing import pipelines
• Comparing raw and processed motion data

Choosing the Right Output for Your Application

• Developing algorithms or ML models? → Raw Mode + raw exports
• Doing biomechanics or sports analysis? → Optimized or Mixed Mode
• Building VR or control systems? → Quaternion Mode
• Need MoCap-style output? → BVH export

QSense is designed so you don’t have to choose too early—you can record once and decide later.

Conclusion

QSense Motion gives you control over data, not just access to data. By separating streaming modes from data representations, the platform supports everything from low-level research to high-level motion capture and real-time applications.

Whether you need raw physics, clean orientation, or full kinematic motion, QSense lets you work at the level that fits your problem—without locking you into a single format or workflow.