Feature Article: DOI. No. 10.1109/MAES.2023.3266180 An Integrated INS/GNSS System With an Attention-Based Deep Network for Drones in GNSS Denied Environments Sina Taghizadeh and Reza Safabakhsh , Amirkabir University of Technology, Tehran 15875-4413, Iran INTRODUCTION The use of artificial intelligence is revolutionizing navigation techniques [1]. Recent years have seen substantial effort put into applying convolutional neural networks (CNN) to navigation systems [2], [3], [4]. Inertial navigation systems (INS) and global navigation satellite systems (GNSS) have traditionally discussed methods for acquiring position and velocity data using filtering algorithms. INS is a self-contained navigation system that uses accelerometers and gyroscopes to estimate a vehicle's relative position, orientation, and velocity. INS can carry high-frequency measurements, but a consistent slight error in the rapid measurement of acceleration or angular velocity is leading to drift due to accumulated errors in the integration calculations; hence an external correction system is required [5]. The GNSS can effectively play this external error resetting role. GNSS errors are time-invarianthttps://orcid.org/0009-0008-2871-3682 https://orcid.org/0000-0002-4937-8026