0

I am doing a project where I am using accelerometer to perform location tracking task.

How do I differentiate between a state if object is under constant acceleration or is at rest?

In both cases it will return a g, and as I understand, I will have to monitor very small variations in this case as when the object will be at rest, which is proving to be tough for me. If any of these cases is misinterpretted, of course the location information will be wrong.

I am just using an accelerometer, nothing else.

Ali
  • 56,466
  • 29
  • 168
  • 265
user1466072
  • 1

2 Answers2

0

I think you must read about Inertial navigation systems.

Laser
  • 6,652
  • 8
  • 54
  • 85
  • hi. at rest only the gravity component is shown, which is along z axis (using a tri axis accekerometer). what i am assuming is that when the object will be moving in let's say y direction (forward), at constant acceleration motion, it will then also return zero; at which moment i will be experiencing only gravity at z axis. – user1466072 Jun 19 '12 at 10:23
  • To determine the slowdown you need to track changes in acceleration in opposite directions. So if you dont track this changes and accelerometers return 0, that mean you reach const acceleration. – Laser Jun 19 '12 at 10:29
  • well i was also thinking in the same way, i am thinking that when the object will be slowing down, there will come a point when the acceleration will go negative(or opposite direction), and through that i can say that the object is slowing down or stopping, same way it can be determined for constant acceleration. however i am not very sure of this,, that's why i need second opinion. – user1466072 Jun 19 '12 at 10:36
  • I think it's a right way. I dont find another way to solve this problem. And look at Inertial navigation systems. – Laser Jun 19 '12 at 10:49
0

You see it right, you cannot distinguish between these cases. FYI: you won't be able to distinguish between the static case and constant rotation around the vertical axis either.

To go a step further, you will have to substract gravity from the measured acceleration and for that you will need gyroscopes anyway.

And even if you do that, calculating the position won't work, the current sensors are not accurate enough to track position.

Community
  • 1
  • 1
Ali
  • 56,466
  • 29
  • 168
  • 265
  • Thanks. I think i get it where you are coming from. However, I am not concerned on rotation around vertical axis. Now, there is one more thing, I am using a Kalman filter also. But since I donot have a good grasp at this topic, I would like to seek some feedback on, if that would make any difference or not, even if I manage to distinguish between const acceleration and rest position. In my view, the feed to Klaman will be pos and vel information from accelerometer, and if the feeded values are not correct or are off margin, how much reliable would the Kalman prediction be ? – user1466072 Jun 19 '12 at 11:47
  • I can only repeat myself: **it won't work**, these sensors are not accurate enough. Please read the linked answer. – Ali Jun 19 '12 at 11:50