Algorithm 1 INS Solution Improvement Using ML
Input	IMU’s sensor measurements of three gyroscopes and three accelerometers $({\omega }_x,{\omega }_y,{\omega }_z,a_x,a_x,a_x)$ for the MEMS-IMU and the reference IMU, initial PVA states $\left({Lat}_0,{Long}_0,{Att}_0,V_0^N,V_0^E,V_0^D,p_0,r_0,y_0\right)$, and the navigation solution of the reference IMU ($pos\_ref,vel\_ref,att\_ref$).
Step 1	Prepare and tune the ML-ANFIS options (input data, output data, type of clustering, MF type, number of Ms, F and epochs/iterations).
Step 2	Apply the ML-ANFIS on $50\%$ of the input data (training phase).
Step 3	Generate the ML-ANFIS.
Step 4	Evaluate and apply the ML-ANFIS on the remaining data (testing phase).
Step 5	Evaluate the ML-ANFIS’s output (improved IMU sensor measurements $({\omega }_x,{\omega }_y,{\omega }_z,a_x,a_y,a_z)$.
Step 6	Compare the MEMS IMU’s sensor measurements and the ML-ANFIS IMU’s sensor measurements to the reference IMU’s sensor measurements to compute the percentage of improvement caused by the ML-ANFIS (RMSE). $RMSE=\sqrt{\frac{1}{n}{\sum }_n{\left(X_{n,Ref}-X_{n,ML}\right)}^2}$ where $X_{n,Ref}$ and $X_{n,ML}$ are the reference IMU and trained IMU measurements, respectively.
Step 7	Compute the ML-ANFIS’s navigation solution (PVA) by using the output of the ML-ANFIS as the input to the INS.
Step 8	Compare the MEMS IMU (PVA) and the ML-ANFIS (PVA) to the reference IMU (PVA) to compute the percentage of improvement of the ML-ANFIS (PVA) using the RMSE metric.
Output	The INS solution (PVA) of the MEMS-IMU and the ML model compared to the output using the reference IMU.