TY - JOUR T1 - AUV Path Planning in Dynamic Cluttered Environment through the Randomized Kinodynamic Sampling-based method TT - JF - IJMT JO - IJMT VL - 15 IS - 0 UR - http://ijmt.ir/article-1-748-en.html Y1 - 2021 SP - 93 EP - 105 KW - Autonomous underwater vehicle (AUV) KW - Kinodynamic constraints KW - Randomized sampling-based path planning KW - Collision avoidance with moving obstacle. N2 - Considering both kinematic and dynamic constraints (kinodynamic constraints) of an autonomous underwater vehicle in a Kinodynamic path planning algorithm in a dynamic large-scale workspace is an NP-Hard problem. Computational and time complexity of the kinodynamic path planning problem increase in the order O (n2) by increasing numbers of moving obstacles, AUV Kinodynamic constraints, degrees of freedoms, and workspace dimensions. This paper proposes a Randomized Kinodynamic Sub-optimal Planning (RKSP) algorithm for a man-portable class AUV. The proposed algorithm solves the path planning problem by applying a randomized sampling-based method to exploring and expanding in the workspace. RKSP re-plans the path to avoid collision with moving obstacles in a cluttered environment through a behavior-based method. RKSP consists of three main components that tightly coupled together. The first component is a Randomized kinodynamic Planning (RKP) module that generates the random offspring waypoints and plans a feasible path by considering the AUV kinodynamic constraints. The second component is a Numerical Path Optimization (NPO) module that prunes the inappropriate edges of the path and reduces the computational complexity. The third component is a Local-Reactive kinodynamic (LRK) module that re-plans the local path through the neighborhood waypoints to avoid collision with moving obstacles in an unknown environment. RKSP path planning method is evaluated through the three different scenarios in a narrow passage, maze-like space and complex space. Results demonstrate the planned path by the proposed method is feasible and the AUV tracks the path appropriately and avoids collision with moving obstacles. Also, the total numbers of waypoints reduce in comparison to the conventional randomized methods and the planned path is near to the optimal. M3 ER -