LRRT*: A robotic arm path planning algorithm based on an improved Levy flight strategy with effective region sampling RRT*

Aiming at the problems of blind sampling points and slow planning speed of path planning Rapidly-exploring Random Trees algorithm, an effective region sampling Levy Rapidly-exploring Random Trees algorithm (LRRT*) is proposed based on the improved Levy flight strategy. Divide the entire path planning process into two stages: quickly finding the initial path and optimizing the path. Goal oriented strategy is used to explore the path when finding the initial path quickly. The Levy flight strategy is used to regenerate nodes after obstacles are encountered to improve the quality of the expansion points. They can quickly plan a collision-free path. In the phase of optimizing the initial path using the effective region sampling method, each sampling is only sampled around the initial path. Meanwhile, node rejection strategy is introduced to reduce the number of collision detection and accelerate the convergence speed. In 2D and 3D environments, the LRRT* algorithm reduces the initial path planning time by 17.6% and 91.9% respectively compared to the RRT* algorithm, and shortens the average planning time by 12.3% and 65.5%, and the path smoothness is 3.4% and 79.4% shorter respectively. Applying the LRRT algorithm to a robotic arm allows for the planning of collision-free paths.