Researchers from the Department of Mechanical and Automation Engineering at The Chinese University of Hong Kong have developed a groundbreaking solution to enhance underwater robot navigation. The team, led by Hao Wei and including Peiji Wang, Qianhao Wang, Tong Qin, Fei Gao, and Yulin Si, has introduced FAR-AVIO, a novel acoustic-visual-inertial odometry framework designed to overcome the unique challenges of underwater environments.
Underwater navigation poses significant hurdles for visual-inertial odometry systems due to factors like light attenuation, marine snow, and turbidity, which degrade sensor performance and lead to frequent tracking failures. Traditional tightly coupled acoustic-visual-inertial fusion methods, while accurate, often require extensive computational resources that are prohibitive for real-time operation on resource-constrained platforms. FAR-AVIO addresses these issues by integrating a Schur complement formulation into an Extended Kalman Filter (EKF). This approach allows for joint pose-landmark optimization, ensuring high accuracy while maintaining constant-time updates by efficiently marginalizing landmark states.
One of the standout features of FAR-AVIO is its Adaptive Weight Adjustment and Reliability Evaluation (AWARE) module. AWARE continuously assesses the reliability of visual, inertial, and Doppler Velocity Log (DVL) measurements, adaptively regulating their sigma weights to ensure robust performance. Additionally, the framework includes an efficient online calibration scheme that jointly estimates DVL-IMU extrinsics without the need for dedicated calibration maneuvers, further streamlining the navigation process.
The researchers conducted numerical simulations and real-world underwater experiments to validate FAR-AVIO’s effectiveness. The results consistently showed that FAR-AVIO outperforms state-of-the-art underwater SLAM baselines in both localization accuracy and computational efficiency. This advancement enables robust operation on low-power embedded platforms, making it a practical solution for underwater robots.
The implementation of FAR-AVIO has been released as open-source software, fostering further innovation and collaboration within the research community. This development is a significant step forward in the field of underwater robotics, offering a reliable and efficient navigation solution that can withstand the challenging conditions of marine environments. Read the original research paper here.