When working with an Xsens DOT, it is important to remember that besides measuring raw data from the internal sensors (accelerometer and gyroscope), orientation and free acceleration data is also available. These last ones are obtained by combining information from different sensing elements through a process called sensor fusion. To do so, Xsens has developed its own algorithms, resulting in extremely accurate and stable data, within one degree of accuracy [2,3]. Then for a user, one of the most interesting questions is: what is the difference among all different type of data that Xsens DOT provides?
Before addressing that, it is important to clarify how to interpret the data that you see in the output sheet or app. Acceleration and angular velocity are expressed in a local-reference frame, meaning that X, Y and Z axes will be well defined with respect to the sensor case. On the other side, orientation and free acceleration are expressed relative to the Earth-reference frame, whose axes are defined in the picture below and aligned with fixed points on Earth, such as the magnetic north and the gravity vector pointing downwards.
- X: along the long edge
- Y: along the short edge
- Z: pointing up, out the case
- X: pointing East
- Y: pointing North
- Z: pointing up
X: pointing East, Y: pointing North, Z: pointing up. Also called ENU = East – North – Up
Acceleration (also known as Specific Force) and gyroscope data are initially obtained by direct measurement from accelerometer and gyroscope sensing elements at 800Hz (note that this type of raw data is not available as output at any point). These are then fitted into the Xsens Kalman Filter Core (XKFCore)  and down-sampled at 60Hz through a protocol called strap-down integration, which preserves the interesting part of the original signal (essentially peak values within the waveform) while flattening out on less “interesting” parts of the signal [1,2]. This is essential to lessen the amount of data being transmitted while maintaining accuracy in acceleration and angular velocity. Since these types of data come directly from the sensing elements within the case, it is natural to express them in the sensor-reference frame.
Strapped-down data is then further processed by the XKFCore Sensor Fusion algorithms, together with magnetometer data in order to obtain orientation and free acceleration data. These two types of data are then expressed in the Earth-reference frame in order to give a universal reference, regardless of how the sensor is placed on the body.
Difference between acceleration and free acceleration
Acceleration is a simple reading of the accelerometer sensor, which includes the gravity component in it. For instance, holding the sensor flat on the back side will sense a reading of about positive 9.81m/s^2 on the Z axis (figure 1A). However, when holding it flat on a side, acceleration reading will be approximately +/- 9.81 on the Y axis (or X depending on the side).
Figure 1: Differences in acceleration data when the sensor is kept flat (A and B) or on its side (C and D)
Free acceleration is instead the acceleration sensed by the IMU, without the gravity component, expressed in the Earth-reference frame. To better understand this concept, you can compare the two data types. By keeping the sensor flat on any of its axes, will always give a reading of about 0 m/s^2 as expressed in figure 1A and 1B (be mindful of noise and bias that slightly skew this value. More info on noise and bias can be found in this article). In addition to this, any movement of the sensor in any direction will be detected as an acceleration on the corresponding axis (on the Earth frame) regardless of how the sensor is oriented in space.
Figure 2: Free acceleration is not influenced by how the sensor is oriented in space. For example, it remains the same whether the sensor is kept flat (2A) or on its side (2B)
Both types of data can be useful, depending on the goal of the project or application. Since acceleration is the rawest form of acceleration data obtainable from Xsens DOT, it is perfectly suitable fur users who want to develop their own orientation algorithms, without solely relying on Xsens XKF core. On the other hand, free acceleration is the end result of Xsens Sensor Fusion algorithms, which makes it suitable for users that are not interested in the computational part but, rather, are only interested in the final ready-to-use data and implement it into their own application.
 Xsens DOT User Manual, 2020, Document XD0502P https://www.xsens.com/hubfs/Downloads/Manuals/Xsens%20DOT%20User%20Manual.pdf
 M. Paulich, H. M. Schepers, and G. Bellusci, “Xsens MTw White Paper: Miniature Wireless Inertial Motion Tracker for Highly Accurate 3D Kinematic Applications.” https://www.xsens.com/hubfs/3446270/Downloads/Manuals/MTwAwinda_WhitePaper.pdf
 A. Vydhyanathan and G. Bellusci, “Xsens MTi White Paper: The Next Generation Xsens Motion Trackers for Industrial Applications.” https://www.xsens.com/hubfs/Downloads/Whitepapers/MTi_whitepaper.pdf