From: emfuchs@irts.mit.edu (Eric M. Fuchs) Subject: Ultrasonic position detection Date: Fri, 4 Oct 91 14:08:48 -0400 Greg Alt writes: >In article <1991Oct1.045624.25978@milton.u.washington.edu>, "Mark Anthony Brown;E200" writes: >> Does anybody have any information as to the accuracy/operation of ultrasonic >> trackers used in VR hardware? I've heard that the Mattel power glove uses >> ultrasonics for tracking. The only other information I've got is on Sutherland's >> original HMD, which used continuous wave ultrasonics or (shudder) a mechanical >> arm attached to the user's head. Comparative performance information for >> ultrasonics versus electromagnetic trackers would also be much appreciated. >Well, I can give you some information about the powerglove ultrasonics... >You can get position info acurate to 1/4 inch, but the ultrasonics must be >aimed at the receivers (I think you have up to 90 degrees to play with). >Also, with the powerglove, you can get informartion about 30 times a second. >I think the electromagnetic trackers have a longer lag time, but they are not >"directional". Also, to get full 6D information from ultrasonics, you need >3 emitters. >I am thinking about building my own ultrasonics to be used for head tracking, >but it does seem to be a large project. If you're willing to spend $10k for the complete system (or about $4-5k OEM version), another option to consider is the GP8-3D Sonic Digitizer from Science Accessories Corp. 200 Watson Blvd. Stratford, CT 06497 203-386-9978. It can send 150 3-D points per second to your host computer. Since three points are needed to get the 6-DOF description of the head, this lets you remeasure the head position and orientation 50 times per second. (They provide software to convert the three points into a 6-DOF description in your host computer, which should take less than a millisecond.) Unlike the electromagnetic trackers, the actual latency of the measurement should be less than the 20 milliseconds implied by the update rate, since the only cause for latency is the time it takes the 60kHz tone burst to propogate from your head to the sensors on the ceiling at about 1.1 ft/ms. Since the GP8-3D comes in models that can multiplex either 8 or 16 sound emitters, you could use one unit to simultaneously track up to 2 or 5 body parts, with the refresh rate for each body part being 50/sec divided by the number of body parts. The resolution and accuracy for each point measurement are .01 cm and .05 cm , respectively, contingent on uniform atmospheric conditions in the room. The maximum tracking volume is a 9 foot cube. If you can wait, a much better tracking technology is supposed to be released this winter from Gaertner Research Division 140 Water Street Norwalk, Connecticut 06854 203-866-3200 fax: 203-838-5026. It uses multiple infrared diodes mounted on the subject, and offers " 1 mRad accuracy, up to 240 Hz update rate, 4 ms phase lag/latency, 6 degrees of freedom, no mapping required, magnetic & sonic noise immunity ..." I asked Robert McGill at SIGRAPH what it will cost and he said it's expensive now but should eventually be down around $2000. If that's true, it seems to me they will put all the other tracker manufacturers out of business. Any information to the contrary will be appreciated. - Eric Fuchs - grad. student, Electrical Eng., M.I.T. - email: emfuchs@irts.mit.edu