Hi Sajid Bhat,

Thanks for your answer. Does it mean that we can compare the accelerometer and gyroscope output only when there is considerable angular momentum that can affect the accelerometer output too? What about the physical vector whose properties the 2 sensors are trying to measure? In case of accelerometer won't this physical vector be both the force causing an acceleration + gravity, but in case of gyroscope, it is only the force causing the acceleration? Isn't this another issue for which the accelerometer and gyroscope do not complement each other?

An accelerometer measures the combined effect of both _linear_ acceleration and gravity. For very fundamental reasons, there is no way, locally, to distinguish these two components..

A gyroscope measures _angular_ speed and acceleration. It is not affected by gravity because there is no such thing as angular gravity. Also, angular acceleration is not caused by a force, it's caused by a torque.

Hope this helps.

As I understand, accelerometers work for translational motion and gyroscopes for rotation. Hence, they are not interchangable in most cases.

Since they measure different things, they can be integrated together to better describe the motion of a device. For example rotations, translations and tilt caused by undesired shaking when taking a picture might need the data from both accelerometer and gyro.

Hi All, Thanks for your answers. I believe now that the accelerometer gives the combined effect of the linear motion and gravity . My sensor showed me the gravity acting on it even though I gave no linear acceleration to it and kept it still.

Also I understood that Gyroscope ,only and very sensitively gives the angular velocity though it suffers from drift.

Yes they both can be used to together describe the motion of a device, but they can also be used to correct each other.

For example, I read and experimented that the drift in the gyroscope can be corrected by using the accelerometer even though they don't essentially measure the same thing. The trick to this is to assume that the accelerometer is not affected much by any linear acceleration when the linear acceleration is very very small, like may be that which is created when we move our hand very slowly.

In that case accelerometer measures effectively only the gravity and its orientation w.r.t the axes of the sensors. And the gyro gives the angular velocity of the rotation of the axes of the sensor integrating which over time we get the angle of rotation and hence a representation of orientation. This angular velocity also causes the gravity to change its orientation w.r.t the sensor axes and thus affects the accelerometer output. Hence we have 2 sources of orientation, one from accelerometer and another from gyro. The orientation information from accelerometer is used to correct the drift in gyro and get an even accurate orientation information.

Thanks all for your answers. Helped me a lot!

Varun, you are right, given the implicit assumption (the center of the circle is not accelerated in tangential direction, for example due to a tilt against gravity).

@Yashaswini Prathivadi ("apples to oranges comparison?"), yes. And the outcome depend on what you are going to measure. As gyros are unaffected by linear acceleration (ideally), lets talk about angles. Gyros show angular velocity. Integrating over time yields change of orientation. If you know the initial orientation, you can estimate the final orientation. The problem is the unsteady offset of the angular velocity. Now for the accelerometer. Hold it quietly and it shows you the direction of gravity. No need to integrate. Problem: This is only part of the orientation. You don't get the azimuth angle.