Single camera_AHRS integrated navigation algorithm for autonomous underwater vehicle

Abstract

As Autonomous Underwater Vehicle (AUV) research in the deep sea is actively underway, further research about the use of human service robots to reduce cost and manpower is needed, especially, for cases where human service robots are used in a pool or off shore for complex, complicated missions. In order to provide continuous and accurate navigation information, integrated navigation systems with several sensors have been used. In particular, vision is a powerful sensing modality that could be used in many tasks. This paper proposes a single-camera/AHRS integrated navigation algorithm for use with a low-performance inertial sensor. The position and attitude estimate from a single camera using direct geo-referencing is compared with that from an Inertial Navigations System (INS) sensor. The measured position and attitude error are inputs of the integrated Kalman filter. The validity of the proposed system is shown through simulation and AUV test. The simulations and experimental results show that a single-camera/AHRS integrated navigation system produces high-grade INS performance.ed in a pool or off shore for complex, complicated missions. In order to provide continuous and accurate navigation information, integrated navigation systems with several sensors have been used. In particular, vision is a powerful sensing modality that could be used in many tasks. This paper proposes a single-camera/AHRS integrated navigation algorithm for use with a low-performance inertial sensor. The position and attitude estimate from a single camera using direct geo-referencing is compared with that from an Inertial Navigations System (INS) sensor. The measured position and attitude error are inputs of the integrated Kalman filter. The validity of the proposed system is shown through simulation and AUV test. The simulations and experimental results show that a single-camera/AHRS integrated navigation system produces high-grade INS performance.