AFI-RRT$$^*$$: Adaptive Free Interval Guided RRT$$^*$$ for Path Planning in Cluttered Environments

Path planning for mobile robots and unmanned aerial vehicles in complex environments remains challenging, particularly when efficient exploration, short path generation, and low computational cost are required. RRT \(^{*}\) and its variants can provide high-quality solutions, but they often suffer from excessive exploration and large search trees, especially in geometrically constrained environments. To address these limitations, this study proposes AFI-RRT \(^{*}\) , an Adaptive Free Interval guided RRT \(^{*}\) framework that combines coarse A \(^{*}\) -based global guidance with structured local free-space exploration. Its main advantages include focused sampling through adaptive free-interval layers, reduced unnecessary tree expansion, preservation of global exploration, and direct extraction of the final solution from the RRT \(^{*}\) tree without additional geometric post-processing. The proposed method was evaluated in nine environments over 30 independent runs and compared with Informed RRT \(^{*}\) , A \(^{*}\) , AdaRRT, RRT \(^{*}\) , RRT, RRT-Connect, and QS-RRT \(^{*}\) . AFI-RRT \(^{*}\) achieved the lowest overall mean path length while reducing average execution time by 13.8% and average search-tree size by 52.5% relative to RRT \(^{*}\) . Compared with Informed RRT \(^{*}\) , the corresponding reductions were 35.7% and 53.0%, respectively. The method also maintained a 100% success rate. Wilcoxon signed-rank tests showed statistically significant path-length differences against RRT \(^{*}\) in all nine environments and against Informed RRT \(^{*}\) in seven environments. Overall, AFI-RRT \(^{*}\) provides a favorable balance between path quality and computational efficiency by focusing the search on geometrically relevant free-space regions.