Hi there!
Nokia will soon have devices that has 2-axis acceleration sensor built-in.
How about using it to increase the accuracy of GPS tracking? It should be quite easy to calculate vehicle position using old position and then applying accelerations to it to get new position until next GPS update.
That way accuracy of track maps could be improved. Also, acceleration caused by throttle/brake and steering could be logged. I have not read the specs trough for the G sensor interface but it's update rate is way better than 1 Hz.
In optimum setup you would place GPS BT dongle on your dash to get clear view of the satellites. Then you place the mobile near the center of gravity of the car (between drier and passenger seat usually) to get most accurate results.
Any comments? :)
Comments
I'm not sure how easy it will be to synchronize a Bluetooth GPS position feed to a accelerometer data feed, since the receivers probably have unique "lag" and also Bluetooth will cause some variation on this. If I can solve this problem, then why not!
There was a demo on the phone and I played with it. For me there seemed to be zero delay from actual movement sensed by sensor and shown on phone display. Not sure what are the max G limits and what is the accuracy. I will try to find out if the information is already public.
http://www.st.com/stonline/products/families/sensors/accelerometers.htm
There is a test app called MovingBall, a story link can be found here: http://symbianworld.wordpress.com/2007/11/01/n95´s-next-step-to-get-closer-to-the-iphone-movingball/
A link to download there as well.
More links:
http://wiki.forum.nokia.com/index.php/N95_sensor
http://research.nokia.com/projects/activity_monitor
The accell data will be usefull for performance testing where GPS is week at low speeds. e.g. better definition of when the vehicle starts to move. (on the N95 the internal GPS could be used to time sync with the bluetooth GPS).
The accell data will also be useful to gather data for brake point distributions and lat acc rates for corner turn-in and mid corer max accell/g.
I would not try to use it to join the GPS data points as my BT-1000 definetly suffers from over-shoot on high g corners and I suspect implementation of the GPS/INS filter would be a whole new (long) project to get it working properly.
I have got an N95 so I will try to log some accel data, but I don't know how to yet so it may take a while!
1) to improve the GPS resolution (increase the Hz, and make the traveled route "smooth")
2) to improve the GPS positioning accuracy
3) to improve speed accuracy.
I would think PerformanceBox does 1) and 3). It advertises to have an accelerometer, but it only saves position + speed to flash card, but no g-force.
I would think 3) would be the most obvious thing to implement, to improve the acceleration tests.