3-Dimensional Path Planning for Autonomous Underwater Vehicle

This paper proposes a new algorithm for Autonomous Underwater Vehicle (AUV) path planning in 3D space to visit multiple targets using Dubins curves. For a given target-sequence, the 3D path planning is usually solved by two steps: Step 1 projects 3D targets to the X-Y plane and designs a 2D path on this plane; Step 2 maps the 2D path into 3D via interpolation. The proposed new algorithm defines a local coordinate system (LCS) for each pair of targets and designs the 2D Dubins curve in the LCS, then uses the Euler's rotation transformation to convert the 2D dubins curve into the 3D global coordinate system (GCS). Applying the proposed rotation method to a given target sequence and given incoming-outgoing angles yields 3D Dubins path with guaranteed G2 continuity at the joints of two Dubins curves. The proposed method is compared with the interpolation method and Bezier curve method. Computer simulations demonstrate that the proposed algorithm provides better G ² continuity in 3D space or shorter path lengths than the existing linear or spline interpolation methods and Bezier curves.