2004 Conference Proceedings

Go to previous article 
Go to next article 
Return to 2004 Table of Contents 


Mel Dashner
Origin Instruments Corporation
854 Greenview Drive
Grand Prairie, TX 75050
Phone: 972-606-8740
Fax: 972-606-8741
Email: mdashner@orin.com

The typical head-tracking client has severe mobility impairment and cannot use a standard desktop keyboard or mouse. Within this group there are two broad classes of users. The first class has become impaired due to an injury that has reduced the functionality of their arms and hands. Depending on the specific injury the level of reduced functionality can vary over a considerable range.

The second class is impaired due to disease. Depending on the disease the client typically has a systemic reduction in his or her ability to accurately control any part of the body.

The distinction between these two groups has implications for the client's ability to accurately control head movements. Although good head control is not required to use a head pointing device, the efficiency and even success in using the device will be in direct proportion to a person's head control. There is a sliding scale of utility for a head-pointing user that is directly proportional to his ability to accurately control his head.

However, if the class of problems a head-pointing client is attempting to access is controlled then a person with limited head control can be successful. For example, if a client with poor head control is attempting to select tiny icons on a complex computer desktop he will have little success. If on the other hand, the computer display has been customized to use large icons he will be more successful. Further, if the client is using a tailored application with very large icons, or buttons like an ACC application, then he can be successful and efficient.

The most successful products used for head tracking take advantage of the following technologies: magnetic, ultrasound, inertial and optical. Up to this point devices using magnetic sensors have not been commercially applied in the field of disabilities and will not be discussed in this paper. Ultrasound has been used in many tracking sensors in several industries including the disability marketplace. Inertial sensors are used for tracking in many industries as well as the disability industry. Probably the most successful head tracking technology applied in the field of disabilities is optical.

All these technologies have strengths and weaknesses and given different design and cost considerations one may be more appropriate for a specific application. The manufacturer must decide the balance between ease of use, device cost, tracking performance and targeted client capabilities. Since device cost is strongly driven by the complexity of a technology and the production volume, a manufacturer will often decide to pick one technology for their entire product line. By specializing on one technology they can reduce their requirements for additional expertise, maximize volume and thereby reduce costs. In addition to tracking performance, there are strong secondary issues that are equally compelling in a head-tracking device, such as the elimination of wiring and weight on the head. For example, a company may have created a wonderfully accurate and responsive head tracking sensor that requires the client to don a head mounted frame with a cumbersome, heavy and wired sensor. However, the client may prefer a device with different performance that does not require a cumbersome head mounted device with attached wiring.

What is important to remember is that head-tracking devices are no different than other manufactured products, there is a tradeoff between the specific tracking technology, delivered device cost, functionality and industrial design. What is also important to remember is that within this tradeoff is where great companies with clever designers distinguish themselves.

The following performance issues are very important when considering a head-tracking device:

Some technologies require mounts or obtrusive devices to be attached to the client's head and can dramatically affect the user's experience with the tracking device. For example, the inertial and ultrasonic based devices require a significantly large device mounted on the head. These devices are not really large in an absolute sense, but with time they can become very uncomfortable, require maintenance and regular re-alignment or adjustment. They also can negatively impact a client's self-esteem.

Some optical head tracking sensors also required a wired sensor to be attached to the head and therefore, suffer the same problems listed above.

The client must decide if a device has additional performance or other advantages that warrant a head mounted apparatus.

Many newer optical tracking devices only require a tiny paper-thin sticker. Ideally, the client would not have to wear anything, but compared to the other technologies this compromise is a minor issue.

There are fundamental differences between the fielded sensors and how they apply the identified tracking technologies. The products using ultrasound and inertial sensors are tracking head angle and the optical sensors are tracking head position. However, to the user it appears that the optical sensors are actually tracking head angle. The reason is that tracking the dot affixed to the forehead is a first order approximation of tracking head angle.

For mouse emulation, tracking resolution of the three technologies is roughly equal given a non-hostile ambient environment and an adequate design. However, the sensor must have sufficient resolution such that the pointer moves in a smooth and precise manner. Low resolution is usually apparent when the sensor gain or mouse gain has been adjusted to the higher settings or one is operating at the extremes of a sensor's range.

A significant, but often overlooked specification for a head-tracking device is known as tracking latency. This term describes the delay in a device reporting head movement. It is manifested in a "sluggish pointer." Latency makes it harder to accurately position the mouse pointer in a given time. Taken to an extreme it can significantly reduce efficiency and induce fatigue. There are small but significant differences in latency between many of the currently fielded head-tracking devices.

Another important characteristic of the currently fielded sensors is known as hysteresis. Specifically this term means that if you move the head an exact amount and then move back by the same amount the mouse pointer should return to the exact same position. Because these sensors use a mouse driver the context of this characteristic is only important in small pointer movements. With large and raid movements that traverse large portions of the screen the mouse driver uses so-called ballistic gain to add hysteresis. In this case it is useful. However, when attempting to make small accurate movements it is not and is observed as a "sticky pointer." For example, you begin a small precise movement and nothing happens until you have moved far enough to overcome the hysteresis and then the pointer moves. It makes precise pointer positioning harder and can reduce the efficiency of a person with good head control.

The typical ambient environment used by a client accessing a computer would be considered non-hostile. Also, many so-called hostile environments are not pleasant for the user and are therefore minimized. However, there are some situations where different technologies and designs behave differently even in what many would consider non-hostile environments. For example, specific types of noise can disrupt ultrasound devices and moderate but changing illumination environments can disrupt poorly designed optical sensors. Quality of the design can have a significant impact on the outcomes in these situations.

Power consumption is becoming a more important specification. As our society becomes more mobile power management in our portable electronic devices is becoming more important. It is therefore equally important that access devices for these portables are power efficient as well. As currently fielded, none of the tracking technologies described in this paper appear to have a fundamental advantage over another. However, implementation details of a specific device can and does have a significant impact on power dissipation. Power specifications should be scrutinized if mobile applications are anticipated.

Integration of the head-tracking device into an efficient system for a client will have a dramatic impact on his or her capabilities to be a productive member of society. If the head-tracking device cannot be adapted, mounted and easily used by a client he will be less effective. Ideally, the head-tracking sensor will have a minimal impact on the other aspects of the device being accessed and be forgotten by the user. The device should have a range of quality mounts to accommodate a client's specific situation. It should be powered by connecting to the device being accessed and have a manageable impact on overall system runtime. It should not require a complicated installation process with special drivers that may cause system side effects and require updates as new operating systems are introduced.

During the presentation, the latest head-tracking sensor from Origin Instruments Corporation will be used to illustrate some of the characteristics described above. It uses optical tracking technology and incorporates high tracking resolution, low latency, low hysteresis, low power consumption and small efficient packaging.

There are several head-tracking sensors to choose from and close study of the specifications and adequate evaluations are required to select the most appropriate device. There are significant differences in the underlying technologies and implementation of currently fielded devices and these differences do have a significant impact.

Go to previous article 
Go to next article 
Return to 2004 Table of Contents 
Return to Table of Proceedings

Reprinted with author(s) permission. Author(s) retain copyright.