This question already has an answer here:
I'm looking for ways to accurately navigate from one GPS location to another. I have a GPS sensor and an IMU. I can calculate the distance with the help of GPS I believe and with the help of magnetometer sensor on the IMU(acc + gyro + mag), I can calculate the heading angle and know which way to go. But the GPS data has a lot of outliers and I need to do this navigation accurately. Any suggestions are appreciated.
Thanks.
This question has been asked before and already has an answer. If those answers do not fully address your question, please ask a new question.
GPS itself has about 3.5m of variance. To obtain a better accuracy it is usually fuse the measurements from the IMU with GPS using Kalman filters. The problem of navigation can be decompose into two sections as localization and path planning. Under localization the robot tracks its position after every time step. Then it plans the path for the next move. These two steps need to be iterated until it reaches to the goal. For your concern about outliers, I would suggest to use a Kalman filter and fuse the GPS and IMU measurements. Then you can identify your current position and heading precisely. Then you can use a path planning algorithm.
