Bosch has a 9 DOF IMU including the sensors and an ARM processor in a
*chip* package. Cheap too,.I believe the light version of the sensor fusion library outputs an angular rotation vector.
--sp
I was almost starting to do this for an application, but found enough constraints to make this with an accelerometer.
Your later messages specified light writing, so I'd use 3+3 axis gyro+accelerometer. I did my tests with STM LSM6DS0, which worked fine for me.
Nice!
Wow, I hadn't even considered the act of tossing. I'm fairly sure at least for this job the toy won't be thrown.
"Baton" was probably the wrong word - they would be longer. More like a staff, I guess.
For this application, fortunately I don't think anyone expects it to be thrown and tossed and display coherent images.
"Baton" was the wrong word. Think more like a quarterstaff, something more like a meter and a half long or so, spun hand over hand like Friar Tuck.
Yes, I have, and no, they don't.
If the images don't need to be oriented with the ground then don't use accelerometers at all -- just use gyros.
If you know the center-of-gravity, it just takes a measure of centripetal acceleration (difference axial accelerometers at two ends of the staff) to know the rotation rate. Position and vertical sync might be harder, but a 'maintained' spin rate means you can get that rotation rate by long-time averaging (over a second).
On further thought, you can take the sum of those same two accelerometers to get the vertical, it' reads out the gravity times direction cosine, of a staff at near-constant height.
subtract mean acceleration then watch the zero-crossings. at-least one axis is going to show an amplitude above 0.5g so you can use plenty of hysteresis and get a nice clean square-wave.
Well, the acceleration towards the centre of rotation is omega ^ 2 * r, where omega is angular velocity (in radians per second), and r is the radius of the circle.
A wheel would have a constant 1 g downwards superimposed on that, plus any vibrations.
A 60 centimetre car wheel doing 100km/h imposes an acceleration of over
200 g on its circumference if I've done the math properly, which means you'd have to be careful about the positioning so as not to overload/destroy your accelerometer.Capturing the data from it seems problematic as well.
Sylvia.
Now that we know what you're doing I wouldn't expect 1 second samples to be anywhere near good enough. Your images will suffer massive distortion as h ands keep applying forces to the batons, distorting the scan field. Do it p roperly with way more samples and you can lose all the distortion & inconsi stency. Or your competitor soon will.
NT
Have something to add? Share your thoughts — no account required.
Ask the community — no account required