MTx FAQ
- General Topics
- Communication and Programming Topics
- Sensor Hardware Topics
General Topics
Please contact us and we will send you a copy of the Software Development Kit Documentation, containing details on the MotionTracker object interfaces, the MT Communication C++ class (sources included), example source code (C/C++, Matlab, LabVIEW, Excel) including tested code for Linux, etc.
The orientation software uses the earth magnetic field as a global reference for North/heading (just like a compass). If the STATIC earth magnetic field is distorted, this will influence the orientation calculation. Any ferromagnetic material, such as iron or nickel, will distort the earth magnetic field, however, most metals do NOT influence the earth magnetic field (e.g. most stainless steels, aluminum). The MTx / MTi is also not sensitive to oscillating EM-fields.
The degree of distortion in the earth magnetic field is dependent on the amount (mass) of ferromagnetic material present and the distance between the sensor and the disturbing mass. Small amounts of iron, e.g. present in an office chair, will cause the MTx / MTi to sense a disturbed magnetic field typically within a distance of 20-30 cm, larger masses such as e.g. a car will have a stronger influence over longer distances.
Furthermore, the algorithms developed by Xsens, have different options to strongly reduce any negative influence of such distortions using advanced filtering of the earth magnetic field measurements.
For example, the influence of a STATIC distortion can be fully calibrated for using a one time procedure in the Magnetic Field Mapping Tool included in the MT SDK. This is also often referred to as a soft and hard iron calibration. The calibration can both be performed in a full 3D mode (for hand-held devices, cameras etc.) as well as a 2D mode (cars, marine vessels, etc.).
It is possible that strong DC electrical currents can influence the accuracy of the orientation measurement of the MotionTracker as a DC current generates a (static) magnetic field. In practice currents below 1 A does not influence the MotionTracker greatly if the distance between the wires and the MT is more than a few centimeters. If this is a problem in your application, consider using twisted-pair wires that will partically eliminate any radiated field.
Yes.
Large linear acceleration, of long duration, will decrease the accuracy of the orientation calculation. The linear accelerations cannot be distinguished from the acceleration due to gravity and therefore the reference for 'down', which is gravity, will be less well defined. However, the gyros are mainly used for fast movements so any accelerations of short duration (fast) will be "filtered" out to minimize this effect.
 It's possible to double integrate accelerometer data, after proper co-ordinate transformations and subtraction of the acceleration due to gravity, to obtain 3D position data. To implement this in a practical implementation some issues will be encountered: 1) You will need a 'starting point', a reference 3D position, from which you can start to integrate the 3D acceleration data. 2) Noise on the acceleration data (about 1 mg RMS) and small offset errors and/or incorrectly subtracted acceleration due to gravity, will be integrated and over time will cause huge (drift) errors in the position and velocity estimate. The conclusion is that it depends very much on the (type) of motion you want to register if this approach is feasible. Typically, short duration movements, preferably cyclical, with frequent known reference positions will work well.
The software algorithms that Xsens has developed for its MotionTrackers are implemented to use earth gravity field as a global Earth-referenced for the vertical and the earth magnetic field as a global reference for North/heading. This is also known as an Attitude and Heading Reference System (AHRS) inertial measurement unit. This method will eliminate any drift that typically would occur from integrating the angular rate of turn data from the rate gyros (especially MEMS gyros).
This is caused by an (mathematical) artifact in the description of Euler-angles (also known as Cardan angles). In the Euler-angles description of orientation there is a singularity (also called "gimbal-lock") when the pitch is +90 deg or -90 deg. This means that, as the pitch approaches ±90 deg the output angles in the yaw and roll will start to fluctuate greatly and that you will not reach pitch = ±90deg.
Communication and Programming Topics
Yes. The MTx/MTi can send orientation data and/or calibrated sensor data (accelerations, gyro data etc). When using calibrated sensor data, user can also select to receive all data or just for example accelerations data.
The serial output of the MTx/MTi is dependant on the configuration. The output can be configured to output any combination of 3D orientation data (quaternion, matrix, Euler) in combination with 3D acceleration, rate gyro, magnetometer, temperature.
Please check the MTx/MTi Users Manual for information on different settings and the corresponding binary data. Contact us to obtain a copy.
Bits/second: 115200
Data bits: 8
Parity: None
Stop bits: 2
Flow control: None
All binary data communication is done in big-endian format. Example:
16 bits accelerometer output = 1275 (decimal) = 0x04FB (hexadecimal)
big-endian: 04 FB The bit-order in a byte is always: [MSB...LSB] » [bit 7...bit 0]
About 50 ppm (±0.5 us) within temperature operating range.
The duration of the 16 bit AD-conversion (sampling) of the 9 channels of data inside the MotionTracker takes ±200us. The transmission over the serial RS232 link (@115k2 bps) takes 2.17 ms. The total time is 2.37 ms. At higher baud rates (the supplied USB-RS232 cable that comes with the MTi and MTx Development Kit can handle up to 921k6 bps) the latency can be reduced further.
No, not "officially". However, the DLL COM-object can certainly be used inside the Delphi development environment through the IDispatch interface.
Yes, the DLL supplied with the MT Software Development Kit can be used in most programming environments on the Windows platform such as popular tools like C, C++, Matlab, LabVIEW, Excel, Visual Basic, Delphi etc. Tested sample source code is included for C, C++, Matlab, LabVIEW and Excel (VBA) and Linux.
Sensor Hardware Topics
Rate of turn sensors (gyros): MEMS, solid state, "vibrating structure" concept.
Linear accelerometers: MEMS, solid state, capacitative readout.
Magnetic field sensors: Thin-film magneto-resistive.
Yes. But keep in mind that for optimal accuracy the electronics of the MTi/MTx should be "warmed up" for about 10-15 minutes.
To avoid aliasing the MT uses low-pass (LP) filters on all inertial data channels (not on temperature and magnetometers). The filters used are effective second order and are available in different cut-off frequencies depending on your requirements. The standard MT is delivered with 40 Hz LP on the rate-of-turn sensors and 30 Hz LP on the accelerometers.
The MTi and MTx are delivered with an Xsens USB Converter. The communication from sensor to USB Converter can be either RS-232, RS-422 or RS-485.
For the MTx the connector type is "Binder 719". For the MTi the connector type is "ODU L-series 7". Pin-definitions are listed in the MTi and MTx User Manual. Please contact us and we will send you a copy.
The pin-definitions are listed in the MTi and MTx User Manual. Please contact us and we will send you a copy.
When we say 100 Hz we mean that each sensor channel is sampled at 100 Hz, so actually for the MT 10 channels of data (3D accelerometer, 3D rate of turn, 3D magnetometer and temperature) are transmitted at 100 Hz, some would call it 1000 Hz throughput.
Yes, most PocketPC's support standard RS232 input (mostly with supplied RS232 synch cable) so you can develop software for the PocketPC using binary data from the MTi or MTx or log the inertial data on the PocketPC (please contact us if you would like to receive a simple binary datalogger for the PocketPC).
Using the Xbus Master you can also easily connect to a PocketPC using Bluetooth.
|
