"Comparative Analysis of Machine Learning and Deep Learning Based Error Correction in GPS\/IMU Localization for Autonomous Vehicl"

"Machine learning and deep learning techniques offer promising solutions to the challenge of integration drift in autonomous vehicles localization. This study and ongoing research program evaluates the performance of various machine learning models for predicting longitude and latitude in autonomous vehicles. The models considered include Decision Tree (DT), Random Forest (RF), Linear Regression (LR), Long Short-Term Memory (LSTM), and Gated Recurrent Unit (GRU). The performance of each model was assessed using the Mean Squared Error (MSE) metric, with the following results: the Random Forest model achieved the lowest MSE of 8.5322e- 06, indicating superior accuracy in predicting geographic coordinates. Linear Regression closely followed with a MSE of 8.3323e-06, demonstrating strong performance as well. The Decision Tree model showed a MSE of 1.6934e-05, slightly less accurate than RF and LR. The LSTM model, which produced a MSE of 0.0005968, and the GRU model with a MSE of 9.3090e- 05, performed significantly worse, indicating that they are less suitable for this specific longitude and latitude prediction task. These results determined by quantitative analysis highlights that ensemble-based methods like Random Forest provide the most accurate and reliable predictions for autonomous vehicle localization, making them the preferred model for this application "