Attitude and Heading Estimation


Despite recent advancments in Micro-Electro Mechanical Systems (MEMS) inertial and magnetic sensors, percices and accurate attitude estimation is a challenging task, especillay in the existance of magnetic distubances or high dynamic motions. This problem cannot be significantly tackled by conventional methods and clasical estimators. In this paper, an end-to-end deep learning framework is develped to estimate the attitude and heading using inertial and magentic sensors obtained from a low-cost IMU. The proposed model consists of two-layer stacked bidirectional Long-Short Term Mermory (LSTM) and Feed Forward Neural Network layers. The model is trained using a large dataset of IMU measurements collected from publicly availabe datasets inertial orientaion and inertial odometry datasets.


Problem Defenition

Literature Review



Deep Learning Model

Loss Function



Arman Asgharpoor Golroudbari
Arman Asgharpoor Golroudbari
Space-AI Researcher

My research interests revolve around planetary rovers and spacecraft vision-based navigation.