Enhanced SIRRT*: A Structure-Aware RRT* for 2D Path Planning with Hybrid Smoothing and Bidirectional Rewiring

TL;DR

Enhanced SIRRT* (E-SIRRT*) addresses slow convergence and high variance in 2D path planning by combining skeleton-based initialization with two refinements: hybrid path smoothing and bidirectional rewiring. The method deterministically initializes from the environment's skeleton to produce a high-quality initial path, then refines the accompanying tree to improve cost propagation before performing informed optimization within an ellipsoidal sampling region. Experimental results show that E-SIRRT* consistently outperforms IRRT* and SIRRT* in initial path quality, convergence rate, and robustness across 100 trials, with deterministic behavior reducing variance. The work demonstrates that integrating structure-aware initialization with targeted geometric and connectivity refinements yields reliable, high-quality motion plans suitable for practical navigation tasks.

Abstract

Sampling-based motion planners such as Rapidly-exploring Random Tree* (RRT*) and its informed variant IRRT* are widely used for optimal path planning in complex environments. However, these methods often suffer from slow convergence and high variance due to their reliance on random sampling, particularly when initial solution discovery is delayed. This paper presents Enhanced SIRRT* (E-SIRRT*), a structure-aware planner that improves upon the original SIRRT* framework by introducing two key enhancements: hybrid path smoothing and bidirectional rewiring. Hybrid path smoothing refines the initial path through spline fitting and collision-aware correction, while bidirectional rewiring locally optimizes tree connectivity around the smoothed path to improve cost propagation. Experimental results demonstrate that E-SIRRT* consistently outperforms IRRT* and SIRRT* in terms of initial path quality, convergence rate, and robustness across 100 trials. Unlike IRRT*, which exhibits high variability due to stochastic initialization, E-SIRRT* achieves repeatable and efficient performance through deterministic skeleton-based initialization and structural refinement.

Figures (6)

• Figure 1: Overview of the SIRRT* algorithm: (a) Skeletonization of the binary occupancy grid map via morphological thinning; Harris corner nodes are marked in yellow. (b) Initial tree (blue nodes and lines) and initial path (cyan line) constructed using an MST over the extracted corner nodes, start (yellow dot), and goal (red dot) positions. (c) Optimization phase via informed sampling within an ellipsoidal region (green ellipse); the final optimized path is shown in red.
• Figure 2: Illustration of bidirectional rewiring around a smoothed path node $p$. (a) In forward rewiring, the parent of a neighbor $q$ is updated to $p$ if it yields a lower cost and the edge is collision-free. (b) In reverse rewiring, $p$ adopts $q$ as its new parent under similar conditions. Solid arrows indicate parent-to-child direction, red arrows show newly rewired edges, and dashed circles represent the rewiring radius.
• Figure 3: Initial tree refinement and optimization result in E-SIRRT*. (a) Initial tree (blue nodes and edges) constructed via grid map skeletonization, with the smoothed initial path (cyan) merged into the tree. The magenta line shows the splined initial path before collision-aware correction, and corrected segments are highlighted with grey dotted rectangles. (b) Tree after forward rewiring around the smoothed path; rewired regions are highlighted with grey dotted rectangles. (c) Tree after bidirectional rewiring, further enhancing local connectivity. (d) Refined initial tree and path (green). (e) Final optimized path (red) obtained through informed sampling within the ellipsoidal sampling region (green ellipse). Blue lines show the expanded tree generated during the optimization phase.
• Figure 4: E-SIRRT* result from Experiment #2. (a) Initial tree structure (blue nodes and edges) with the extracted MST path (cyan line) and its hybrid-smoothed version (magenta line). (b) Rewired tree structure and the refined initial path (green line) after bidirectional rewiring. (c) Final optimized path (red line) after informed sampling, along with the sampling ellipse (green) used during optimization.
• Figure 5: Convergence of path cost over post-initial iterations for IRRT*, SIRRT*, and Enhanced SIRRT* (E-SIRRT*) in Experiment #1. Each plot shows the path cost after the initial solution is obtained. (a) Best-case performance. (b) Median-case performance. (c) Worst-case performance.