AUV Kalman Filter-Based Trajectory Estimation Using a Combination of Acoustic and Inertial Sensors

In open-water scenarios, acoustic positioning systems are commonly employed to accurately determine the location of autonomous underwater vehicles (AUVs). These systems rely on acoustic signals emitted by transponders, allowing the vehicle to triangulate its position. However, in confined environments, reflections and scattering of acoustic signals off surrounding structures can introduce inaccuracies and signal loss. Consequently, the positioning system becomes unreliable and distorted, compromising the precision required for detailed mapping in confined spaces. To address the limitations of acoustic positioning in confined environments, AUVs must heavily rely on onboard sensors for trajectory estimation. A combination of inertial measurement units (IMUs), depth sensors, acoustic sensors, and vision-based systems is often utilized to enhance the AUV's awareness of its position and orientation. This work proposes the trajectory estimation of an AUV using a Kalman Filter (KF) approach to fuse data from a Doppler Velocity Logger (DVL), an IMU, a magnetometer, and a depth sensor. DVLs provide velocity in the vehicle frame, while the IMU and magnetometer collectively form an Attitude and Heading Reference System (AHRS). Validation for the proposed solution was conducted in real-world confined tunnels and autonomous dam inspection tasks. Experimental results demonstrated that the KF-based estimation yielded more accurate trajectories compared to utilizing only a dead reckoning approach.