Turning Heads

[Newz2000]

Can anyone suggest a cheap sensor that you could put on your head to detect head motion?<p>You may know that I'm into human-computer interaction and user interfaces. Well, along those lines, I'm interested in figuring out a way to tell a computer which way you're looking.<p>It would be *ok* to detect left motion, right motion, up motion and down motion, but it would be better to detect what angle you're looking (on two axis left/right and up/down).<p>The recent servo magazine had an article about using a laser pointer to detect distance. It would be possible to mount a laser pointer on your head and watch which direction it moves using the same technique as that for the distance finder. However, you'd probably also have to put the web cam on your head to prevent it from losing the dot because of its narrow field of view. I figure there has to be a better way. I will say I'm not opposed to having a webcam on my head for this experiment, but it still seems there's a better way.<p>By the way, has anyone ever seen an IR laser pointer or know a way to focus an IR led so that it has a narrow beam? I suspect their would be little market value for an IR laser pointer since few would be able to see the dot. :p But I think you can grasp how it would be useful for this experiment.

[philba]

<blockquote><font size="1" face="Verdana, Helvetica, sans-serif">quote:</font><hr>Originally posted by Matt Nuzum:
Can anyone suggest a cheap sensor that you could put on your head to detect head motion?<p>You may know that I'm into human-computer interaction and user interfaces. Well, along those lines, I'm interested in figuring out a way to tell a computer which way you're looking.<p>It would be *ok* to detect left motion, right motion, up motion and down motion, but it would be better to detect what angle you're looking (on two axis left/right and up/down).
...
<hr></blockquote><p>At the risk of being accused of having one solution for every problem, you could use an accelerometer to determine head tilt angle (by measuring the angular g field). You could use a digital compass to determine head direction. These aren't cheap solutions but not hundreds of $$ either. <p>I think you will need 2 dimensions for head tilt and 2 for head turn.

[josmith]

How about a qyro?

[Chris Smith]

The only way to focus the IR led or laser is with a viewer, camera, Ir scope or Starlight scope.<p> There are two ways to get a point from a laser or led, but only one way to get a point that is in focus all the time. A collimating lens will give you a beam that stays a dot at any distance while any common lens can make a pin point, but only at one point, at a time.<p>
Some times putting back to back any pair of compound lens assemblies will form a collimating lens. <p>Laser collimating lenses however are small, light weight, and made for the frequency specific.<p> I have used two eye pieces off a telescope back to back and formed a colimating lens, however it was one inch in diameter and three inches long compared to my laser collimating lenses that are less than 1/4 inch in diameter, and 3/4 inches long. <p>You need a camera or Ir scope to focus the lens no matter which type you use. <p>Head motion detection can be from gyro or proximity detectors. As you turn your head the magnetic detection gets stronger in one and weaker in the others. Gyros are great to tell the motion, but calibrating that motion into data is a little harder. <p>So gyro and proximity sensors would be the best in combination. One detects the direction, the other calibrates its degree.

[Michael J]

A few years back I recall they had invented a set
of VR (Virtual Reality) goggles or glasses for a
game, they actually had inbuilt sensors that tracked the movement of the users eye itself to provide feedback for the game.<p>Probably overkill for what you need, and over
budget as well.

[Newz2000]

<blockquote><font size="1" face="Verdana, Helvetica, sans-serif">quote:</font><hr>Originally posted by josmith:
How about a qyro?<hr></blockquote><p>One of those greek sandwhiches with cucumber dip? Sure, I'd love one.<p>Chris, it doesn't need to be a laser point. As a matter of fact, I can see some benefit to having it gradually get larger as the distance increases. Because the res of my webcams is pretty low, a larger dot at greater distances would be nice.<p>So what you're saying then, is that just using a lense to keep the light focused is enough, right?<p>That makese sense. Actually, I've got the cheap lens that originally came with the webcam that might work. Will have to experiment.<p>I'll try the dot first. I'm not going to make any money on this project, so if I can do it cheap, I will. This is actually a question for my list of winter projects. I'm getting ready now because I'm sure I won't be going to the beach this Christmas.<p>How can you use a gyro to detect tilt? I've seen them used to help prevent tilt, but never to detect it.<p>[ October 14, 2005: Message edited by: Matt Nuzum ]</p>

[Chris Smith]

The cheap lens is like any lens, it will focus a point but before and after the focal point,... gets out of focus real quick. <p>Think of two cones point to point, and a pin point on the end where they meet.<p> Move forward or back from the point and the cone diameter gets big, real quick.

A collimating lens will keep a point, from zero to infinity for the most part with a 2 degree or less divergence overall.<p>Gyros for tilt and movment are attached to a Pot or a Rheostat. When the gyro stays in place, the pot [body of the gyro] does not and reads the angle or increase by resistance readings. This is good for tilt and angle, but motion through space is a little harder. <p>There you can use a Piezo gyro, and it measures G forces so that force divided by time equals distance, etc.<p>Here the complexity comes from the variable mass due to velocity. <p>The faster you move, the more the mass increases. <p>The more the mass increases, the greater the resistance to move that mass. <p>The more the resistance to move the mass, the greater the force against the Piezo element.<p>The greater this force, the greater the signal strength [piezo electric effect] and all this has to be translated into a electrical equation representing “all of this variable movement”, plus the time it has been applied over.<p>Any change in mass [the variable or a summed equation] has to be summarized into a equation equal to the movement of distance, over time.<p>A processor is definitely needed even though the calculations are small, they are varying and need summing through the variable time, force, and velocities involved.<p>[ October 14, 2005: Message edited by: Chris Smith ]</p>

[L. Daniel Rosa]

An optical mouse tracks a moving surface. If you could modify one to have a depth of field of about a meter instead of 10 millimeters it could track the user's face.

[Enzo]

Google "laser gyroscope"<p>These things could be mounted - one per axis: pitch, roll, yaw - on headgear to sense when the head moved. These would be relative motion readings. To keep track of the real world (absolute motion) you would have to start with an index point. Something like "look this direction and press START." Then it would sense motions every which way from there. If all you care about is turn left or look up and you don't need to know whether you are looking east or west, you can skip the indexing.

[Chris Smith]

Laser gyros are great, however the timing is in the NS and PS range. <p>Not for the beginner and not a small project.

[bcurrey]

Hows this for simple ?
http://img.photobucket.com/albums/v344/ ... sensor.jpg

[Chris Smith]

bcurrey<p>Excellent idea, and perhaps a speedometer inner cable [flexible and stiff] would allow a little more flex, and while still transmitting the movement correctly.

[philba]

Matt, what kind of accuracy are you looking for? Also, does this need to be self contained on the head? <p>For me the ideal would be a self-contained unit that sits in a hat/helmet/headband and requires no external sensors/reflectors/... That's why I recommended the compass and accelerometer.<p>Today, I was looking at the Oct issue of CCI and they have an article on the HM55B compass. It gives you 64 points of direction (5.6 deg or 4 brad8 resolution) and looks bone head easy to interface to a micro. Parallax has a unit on a breakout board for $30 but it's nothing more than the chip on a little PCB - looks pretty simple to build. It looks like it can be pretty low power so a battery would work quite well.

[Newz2000]

bcurrey:
yeah, it doesn't get much easier than that, but I'll tell you what... In highschool I made some alpha-wave inducing goggles out of a scuba mask (to trick your brain into being more creative) and got laughed at to no end. Imagine whering a device like this! But actually, you're right, it would work very well. My brother is the ME in the family so he'd probably be better at making it then me, but I think constructing it to work like the front wheel axel on a car would be a good idea.<p>philba:
I'm sitting here at my desk turning my head left and right trying to figure out how far 5.6 degrees is. I hope nobody walks in. I think 5.6 is fine enough resolution. I'd just have to add some type of tilt sensor. But the compass idea is the way to go because then it can work regardless of which way your body is facing. Instead of relative only detection you can use it for either relative or absolute. But of course, I'm most interested in relative motion at the moment. Putting two sensors in a hat with a wire running down to a computer in my shirt pocket would be awesome.<p>I will say that I'm not 100% sure this project is going to ever get done. It sounds like it might be growing too complex for what we were thinking originally. It's fun to think about though.

[rshayes]

There was an early digitizing tablet that used a pen with a small spark gap. When the tablet was touched, a short spark across the gap generated an acoustic pulse. This was picked up by two separated microphones and the pen position computed from the time of arrival of the pulse at each microphone.<p>Pulse generators (not necessarily spark gaps) at three points on a headband could be pulsed in sequence. These pulses could be received with an array of microphones, possibly only three or four if suitably located, and the position of each generator located from time of arrival measurements. This should be enough information to calculate the position and orientation of the headband.