This project can predict the imu error by using ANFIS. And we provide a LSTM method to compare with our method. Our codes have been tested by a KITTI dataset. By using our code, the KITTI dataset should be downloaded.
If you want to see the result visually, run the Anfis.m to see the train and test result. The data of the demo used is the Y-direction of sequence-09300018.
The quick start is based on KITTI dataset.
We use Microsoft excel to delete the repeated data. But if it's too much trouble for you, you can write your own code to delete.
Using ErrorCalculation.m to calculate after put your data path in the code, and obtain the position error and other input and output values.
Run Anfis.m to predict the position error of IMU after put your data in the code, and to compare the results between the predicting error and truth error.
Run Lstm.m to get the results based on the lstem method after put your data in the code.