From: Thad Beier <firstname.lastname@example.org> Date: 13 Mar 2000 09:55:20 -0800 Organization: Hammerhead Productions References: 1
View raw article or MIME structure
Ed McBride wrote: > I am a consulting engineer in Colorado Springs with a client who wants > to track a few objects using on-board accelerometers and gyroscopes > that communicate with a central processor. The objects are confined to > an area of roughly 25 meters square, and a height of no more than 5 > meters above ground. It is necessary to know the location of each > object within about 0.1 m, over a time period of a half hour or so. > The objects can move at speeds up to 10 m/s, and can experience > accelerations up to 2-3 g (20-30 m/s^2). The objects can also rotate > at angular velocieies up to a few rev/s. (10-15 rad/s). The cost of > the system cannot exceed something like $2,000 per object. Inertial navigation is not the answer to your question. It's too expensive and too imprecise, by a similar couple of orders of magnitude. When some friends of mine were doing the FoxTrak hockey puck tracking system for the NHL, they had remarkably similar requirements to yours. What they used was a system of infrared LEDs on the puck, and a bunch of simple fixed IR cameras around the rink. Cheap computers watched the view through the cameras, and reported back to 2D positions; which were then combined by another computer to get a 3D position. This worked amazingly well, was inexpensive, and robust. My wife never forgave them for adulterating the sacred rubber icon; (they had to cut open pucks to insert the battery and LEDs) but it did the job.