Smooth-RRT*: An Improved Motion Planner for Underwater Robot

In underwater search and rescue, it is very important for underwater robot to reach the rescue position quickly. Planning path in advance is very important to save rescue time and energy consumption. Therefore, it is meaningful to find a better and shorter path as soon as possible. As a common method of path planning, RRT* has the disadvantages of high cost and slow convergence. To solve these flaws, an improved motion planner for underwater robots is proposed in this paper. In this study, the simulation experiments were divided into twodimensional conditions and three-dimensional conditions, where used point cloud of real underwater scene to find a better initial solution. Based on RRT*, this paper finds the ancestor node farthest from the sampling point and without collision in the random tree as parent node, adds intermediate nodes in the path according to the step size, and uses trigonometric inequality many times throughout the process, so as to obtain an optimized path. Through a large number of simulation experiments, the results show that the cost of path is less and the convergence speed is faster than RRT* and Q-RRT*.