Performance Enhancement of Efficient Path Planning Algorithms for Autonomous Underwater Vehicle
Loading...
Date
item.page.authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Self-powered underwater robots known as autonomous underwater vehicles are employed in various maritime applications including bathymetric data collection in an underwater environment, oceanographic surveys and surveillance. Path planning seeks to find the least bad or optimal path between a starting point and a destination. It is challenging due to the underwater environment s complexity and the absence of GPS signals for AUVs. The distinguishing factor between the path planning in this study and other underwater environments is the type of path planning local and global, for example. Local path planning calculates brief sections of the path and modifies it continuously as fresh sensor data is obtainedand#894; in contrast, global environment path planning is completed before the robot moves and is based on a static environment. To circumvent obstacles and address the limitations resulting from quantizers and computational complexity, a local path planning technique based on Grey Wolf Optimization (GWO) and Multi agent System (MAS) as global path planning is suggested. Based on path length, collision penalty and diving cost the effectiveness of advanced GWO (AGWO) as a path planner is confirmed. The proneness of GWO to become trapped in local minima has an impact on the algorithmic performance of local path planning. To solve the issue a global path planning approach is thus suggested in the presence of an obstacle-rich environment with communication limitations. With this approach, path planning is carried out according to a static environment prior to the AUV s motion using Multi agent System (MAS). Because distributed path consensus (DPC) algorithms might be costly to compute, particularly for large-scale or complex scenarios and because they are less flexible in real-time response to unforeseen obstacles, Convolutional Neural Networks (CNN) based systems are utilized in the creation of path planning algorithms based on location estimation. It is capable of learning complex patterns and relationships in the environme