AEKF-SLAM: A New Algorithm for Robotic Underwater Navigation

Area of Science:

• Robotics
• Artificial Intelligence
• Marine Technology

Background:

• Underwater Simultaneous Localization and Mapping (SLAM) is crucial for autonomous navigation.
• Existing SLAM methods face challenges in accuracy, robustness, and computational cost for underwater environments.
• Effective map management, including landmark addition/removal, is vital to prevent error accumulation.

Purpose of the Study:

• To enhance the accuracy and robustness of SLAM-based navigation for underwater robots.
• To develop a low-computational-cost SLAM solution for underwater applications.
• To improve map management and reduce long-term error accumulation in underwater mapping.

Main Methods:

• Proposed AEKF-SLAM (Augmented Extended Kalman Filter-based SLAM) algorithm.
• Integration of robot poses and map landmarks into a single state vector.
• A novel augmentation stage complementing conventional Extended Kalman Filter (EKF) prediction and update steps.

Main Results:

• AEKF-SLAM demonstrated superior performance in map management (landmark addition/removal) compared to FastSLAM 2.0.
• Achieved more precise and efficient self-localization and landmark mapping for underwater robots.
• Significantly lower processing times were recorded, indicating computational efficiency.

Conclusions:

• AEKF-SLAM offers reliable map revisiting and consistent map upgrading, especially on loop closure.
• The method effectively mitigates long-term error and clutter accumulation in underwater maps.
• AEKF-SLAM presents a promising solution for accurate, robust, and computationally efficient underwater robot navigation.