Playing With Accelerometers

[Defiant]

For Accelerometers, I just recently found this small text:
http://bildr.org/2011/04/sensing-orient ... 5-arduino/

Its good for the general understanding.

[Defiant]

Do you have a link to the paper?

Update: never mind, found it.

[thkais]

Hi,

I think it is not possible to get an *accurate* position out of the data of an accelerometer/gyro/compass combination.
The most compass modules I know have a poor resolution (AFAIK 0.1°), not to mention the accuracy (about 3°). So a compass module is not really useful for calculating a position. By the way, every big amount of iron will disturb the magnetic field - not to mention loudspeakers or something else.
Gyros are useful to get informations about the actual rotation speed - like in a quadrocopter. But, because of the (real) big drift, they don't work for a longer period - in quadrocopters the correction is done by the accelerometers. I use quadrocopters as an example because exactly the combination accelerometer/gyro/compass is used in them - but this sensors are not able to calculate a position in 3D. For positioning they use GPS and sometimes additionally a pressure sensor for the height.
However, I may not know everything. So if there exists an example of a working device using this sensors, I appreciate to get some information about.

[thkais]

hi,

counting steps may also cause errors because the wheels may slip on the ground...
In the video (really nice...) you can see what I mean: After 15 seconds the position is lost. With an IMU you may detect relative positions, but not absolute positions - which you need to navigate a robot...

edit:
The IR/US-combination I posted some time ago ("jugend forscht" project) sounds promising, it's a "in-door GPS" with a resolution better than 1 cm.

[vleeuwen]

The MS-Kinect sensors overs a lot of new possibilities for detecting positions.

This sensor +/- 140 E does a lot of on board preprocessing and is very powerfull.
Image processing and sound processing.
Depth information, skeleton tracking,
Microsoft is offering a good SDK, API, Examples and technical documentation.
Is also integrated into MS- Robotics Developer Studio

[Defiant]

oh vleeuwen spamming again...did you know that in Chine, a Sack of Rice fell over?

vleeuwen, one do not want to detect your position by visual information.

[vleeuwen]

@Defiant,
Ik kan werkelijk geniet van uw obsessie jegens mijn persoon. Maar misschien dat u zich voor uw persoonlijke problemen beter tot een daarvoor opgeleide functionaris zou kunnen richten.
Wees dan ook een echte kerel en verschuil u niet achter een alias.
Met vriendelijke groet,
Carel van Leeuwen

[vleeuwen]

Cher Rei Vilo,
It was not my intention to let you switch to a new PC but more the supply an idea for the position problem as discussed by Thomas.

Maybe it will be cheeper to develop your own optical sensor and sound system.
For this looks like a very nice challenge.
In fact it is simple, take an infrared camera, color camera and a distance sensor, develop the calculations and construct the virtual images.
My perspective is the end functionality, having a image with good distance and position information.
This sensor is offering a good reference.
That's all.

From another source I receive a idea for a new challenge which rely heavely on your sensors.
The one leg hopper robot.

http://www.youtube.com/watch?v=moENDzu_rS0

[Defiant]

The problem is the more plug and play it will be, the more intransparent it gets. Quite the opposite to a compressor you had to build by ft parts a few years ago.
It also limits your abilities to do changes to the system.

vleeuwen: what?