I know I should be using the concept of Dead Reckoning. But how do I use prediction algorithms to determine/estimate the 2D coordinates and yaw angle, if I can get the gyroscope data, wheel encoder data, and global system time or timestamp of each input?
Dead Reckoning is basically an open-loop estimation method (open-loop integration), which is not robust for long-term autonomy. For instance, given some persisting noise that corrupts the gyro or encoder sensor measurements, the integrated outputs (which are coordinates or heading angles) will drift in an unbounded fashion in long run. You need some extra sensors (e.g., GPS) to periodically correct the coordinates/yaw-angle estimation from the dead reckoning.
That being said, if your sensors are navigational grade (very expensive), perhaps dead reckoning can give you decent results for a medium-length period of time.
Hi Xingyu Zhou , thank you for your time. I agree it is not the best method, but that is all I was given, and I am not sure how to start solving with the given parameters.
Aditya Sivaraj see http://aplicaciones.cimat.mx/Personal/sites/default/files/jbhayet/files/clase9.pdf
- Badges
- Science topic