Robust Attitude Estimates in Cases of Sensor Failure with Minimal Sensor Suite

• Published in: 2023 DGON Inertial Sensors and Systems (ISS) pp. 1 - 24
• Main Authors: Panten, P., Bestmann, U., Hecker, P.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 24.10.2023
• ISSN: 2377-3480
• DOI: 10.1109/ISS58390.2023.10361907

The ongoing technological advances in the development of small scale sensors and computing units promote the miniaturization of autonomous vehicles in the air, on land and underwater. While small scale sensors facilitate the implementation of a redundant sensor setup on board of these vehicles, redundancy is mostly avoided in favour of smaller size, lower complexity and cost. However, depending on the field of application and mission scenario, sensor faults and failures can have severe consequences regarding the mission success, environmental impacts and safety of third parties. Hence, crucial sensors, e.g. such needed for navigation and control purposes, should be monitored closely and counteracting measures taken in case of fault or failure. The latter is subject of this paper which aims to identify suitable methods to maintain a viable attitude estimate in case of failures within the navigation sensor suit. More specifically, the paper depicts how attitude sensor failures on earthbound vehicles may be compensated through the remaining sensor data. The methods described in this paper were developed for the TRIPLE nanoAUV, an Autonomous Underwater Vehicle (AUV) for the exploration of sub-glacial lakes. Due to the small size of the vehicle and its operating environment, attitude estimates can only rely on a tactical grade Inertial Measurement Unit (IMU) and Three-Axis Magnetometer (TAM). This implies that analytical redundancy must be used in place of hardware redundancy. Even though this very limited sensor equipment may not be representative for other vehicle types, the developed methods may be adapted for larger sensor suits on land, water or aerial vehicles. The use of analytical redundancy provides a fall-back layer that reduces the inherent risk given by a non-redundant sensor setup. Thus, the potential for mission success can be increased at the slender expense of a larger computational cost.