To investigate the technology involved in magnetic tracking and attempt to build my own 6 degree of freedom magnetic tracking system similar to the Ascension Flock of Birds or the Polhemus FastTrak. At this point in time this project is a future one. I am going to post some very good information I found on the Computer Science page of the Naval Postgraduate School in California. The only reason I am copying the information to here is in case their page is deleted. If someone has any suggestions or requests, please don't hesitate to contact me through the Contact page.
The following information is courtesty of the Naval Postgraduate School in California.
Magnetic trackers are used to capture of [x,y,z] translation coordinates and yaw, pitch, roll [y,p,r] rotation coordinates. Those systems which can track real world objects, i.e. those which have freedom of motion in all six of these directions, are called 6-DOF (6 Degrees of Freedom) tracking devices. Capturing all six DOF is vital for realistic interaction with a virtual environment.
Magnetic tracking is most commonly used as an interface to a virtual world, for instance, by tracking head, hand, or input device motion. However, some magnetic trackers can follow a number of devices simultaneously. Therefore magnetic tracking technology is a viable option for full-motion body capture, for instance by tracking major joints on the human body. This information can be used in real-time, perhaps to drive the motion of virtual character, or can be recorded to give virtual actors realistic motion characteristics.
An excellent example of full-body tracking is the video game industry. Game manufacturers commonly hire martial artists or other actors and record their motions, thereby giving video game characters realistic motion in fight sequences, etc.
A typical set-up contains the following components (see above):
- a transmitter, which is usually permanently installed on a workspace ceiling;
- one or more sensors, often attached to special helmets or gloves;
- an interface device (often called the filter); and
- the computer
Magnetic Field Transmitter
The most complex part of the technology is the field transmission and reception. To better understand the technique of 6-DOF tracking, we begin with a method to determine a single degree of freedom distance.
One Degree of Freedom
Now we need a sensor on the other end, which is a similar coil, but completely passive. The field creates a current in this passive coil, proportional both to the strength of the current in the transmitter and the distance between the transmitter and sensor. Since we know the strength of the transmitter, using these two coils we could figure out how far away the sensor is with some simple math. That's where the filter device comes in.
Two Degrees of Freedom
But what if we want to track the sensor's y distance and its yaw rotation? Our single coils are no longer enough. We already know that the current caused by the magnetic field is weakened in the sensor the further they're separated. But the current in the sensor is also weaker if the two coils are not in a straight line. So if we rotate the sensor, it's going to get a weaker signal. The problem is that we don't know whether the signal is weaker from the sensor's rotation, or from distance. There's no way to tell.
Six Degrees of Freedom
In the full system, the transmitter consists of three coils on orthogonal [x,y,z] axes. A current (either AC or DC, more on that later) is passed through each coil. The sensor consists of a similar set iof three coils. Depending on the system, varying signal strengths or time multiplexing is used so that each of the three magnetic fields can be isolated. This gives enough information to determine the difference between current loss due to rotation vice distance, etc. The filtering device's job is much more important here, as the math is a bit more complicated than just a direct current-to-distance ratio as in the 1-DOF tracker example. The filter can serve a number of other uses, such as filtering jittery readings and the like.
Magnetic trackers are marvelous for motion-capture for a number of reasons:
- No line-of-sight issues. Electromagnetic fields can travel through minor obstructions like human beings with no problem. Therefore, a sensor that is facing away from the transmitter, occluded by a body, can still yield valid data.
- Inexpensive. Magnetic trackers are on the lower end of the spectrum, and can be much cheaper than optical solutions. A key portion is that the processing is not complex compared to other solutions, so the computational equipment demands are slight.
- Accurate. Accuracy is not the highest among the many types of trackers, but it is reasonable, and effective for the price.
- Long ranges possible. Magnetic trackers, in ideal circumstances, can fill large rooms. One product which boosts the field transmission claims a range of 700 sq ft.
- High sample rate. Again, magnetic trackers do not have the highest sample rate, but are quite effective for the price. A typical rate is 120 samples per second.
- Field distortion and interference. Eletronic devices, and conductive or ferrous metals, can distort the projected electromagnetic field. When the field is inaccurate, readings are similarly inaccurate. The 'map and compensate' method is often used, in which the projection space is sampled and inaccurate readings are adjusted. This is a painstaking procedure, and must be repeated whenever the installation is moved, or even when a new computer monitor is added to the room! The recent invention of DC transmitters has reduced these effects. The magnetic field creates "eddies" around the metal objects; the rapidly changing nature of an AC field makes compensation nearly impossible. DC fields can reach a steady state more amenable to compensation.
- Distance diminishes accuracy. Since the field falls off with the square of the distance, readings at the edge of the field are much less accurate
- Latency. All the filter processing introduces latency in the results. Latency can be devastating in interactive VR applications, especially those with a fix visual frame boundary. If the readings are not ready by the time a frame is drawn, the effect is another frame (often 33ms) of latency. A system that reacts slowly to head motion can be disconcerting to the user. For instance, delays greater than 10 ms can cause simulator sickness.
- Jitter. Minor fluctuations in current, the room, etc. can cause a tracker to report motion when none is occurring. Even if this motion is quickly corrected for, the rapid small-range motion gives the jitter effect.