Does an accelerometer sensor measure the distance accurately when it is used inside a car? and what is the misjudge percentage when it is compared with its odometer?
\$\begingroup\$
\$\endgroup\$
4
-
1\$\begingroup\$ Have a look at this stackoverflow.com/questions/17572769/… \$\endgroup\$– Steve RobillardAug 7, 2016 at 21:29
-
1\$\begingroup\$ @Steve: you just saved me from writing an answer. I went straight to the double-integration issues in my mind. But there's no need. OP should just go read and assimilate the info there, to start. \$\endgroup\$– jonkAug 7, 2016 at 21:38
-
1\$\begingroup\$ @Abdulkader - In addition to the Stack Overflow topic kindly linked above, your question is answered by these previous questions here on EE.SE : Limits of dead reckoning using MEMS sensors | Can I “integrate” the data from an accelerometer to record a motion trajectory? | Accelerometer double integration error \$\endgroup\$– SamGibson ♦Aug 7, 2016 at 22:48
-
1\$\begingroup\$ Might be suitable for measuring during a drag race (a few seconds total, zero'd at the beginning, high acceleration, and the wheels are slipping so the odometer is inaccurate) but the errors rapidly accumulate with dead reckoning using even a very accurate and expensive strap-down inertial measurement system. \$\endgroup\$– Spehro PefhanyAug 7, 2016 at 22:49
Add a comment
|
1 Answer
\$\begingroup\$
\$\endgroup\$
An accelerometer requires integration over time to get velocity and integration again for position. But without any motion is subject to thermal bias and/or dc offsets so errors can accumulate in velocity when stopped and even more with position after the 2nd integration. Thus Accelerometers often are not specified to have a DC response but it may be very low such as 0.01Hz. In other words this low frequency period is an indication during 1 cycle where position errors can become significant.
An odometer might have a speed error with under or oversized tires, but there is never a position error penalty for time as in the case with accelerometers.
One such MEMS system used for short distances was adequate for a robotic application for short durations. It had a 2g range using Kalman Filters to reduce random noise and achieved these results during accelerated motion measured in milli-g's. [mg]
- Velocity error = 0.589 m/s per mg per min
- Position error = 17.66 m per mg per min
ref: http://link.springer.com/article/10.1023/A:1008113324758
