RoA-Planner: Rotatable Area based Path Planner in Dense Spaces

Abstract

Path planning in obstacle-dense environments is a challenging problem, particularly for robots with asymmetric rectangular footprints. To address this problem, we propose a novel collision-checking approach, called a Rotatable Area, which represents a range of heading angles where the robot can rotate without colliding with obstacles. Based on the relationship between two rotatable areas, we define safe local motion and extend this concept to the RoA-Planner, a path planning framework in SE(2) dense space. We validate our planner through extensive simulations and real-world experiments in complex and narrow environments. The results demonstrate that our method achieves fast planning speed while ensuring safety and robustness, making it suitable for practical applications.

Method

Rotatable Area

The Rotatable Area (RoA), denoted by $R$, is defined as the continuous range of yaw heading angles within which the robot can rotate at position $(x, y)$ without colliding with obstacles.

$$ R = (x, y, \psi_s, \psi_e)^T \tag{1} $$

The parameters $\psi_s, \psi_e \in [0, 2\pi]$ denote the start and end of the RoA, and all parameters are referenced to the world frame. If the robot’s heading angle is between $\psi_s$ and $\psi_e$ at $(x, y)$, the robot remains collision-free. Within the range corresponding to RoA, the robot can rotate freely.

Three types of rotatable area. (a) Fully-RoA, $R_F$, (b) Non-RoA, $R_N$, (c) Partially-RoA, $R_{P,i}$

Local Motion between Rotatable Areas

Our approach regards local motion as a relationship between two RoAs, $R_1 \rightarrow R_2$. Local motion consists of 1) performing rotation at $R_1$ from the initial heading angle $\theta_1$ to the target heading angle $\theta_2$, and 2) then translation to the target position of $R_2$. The feasiblility of local motion is determined by area condition and edge condition. If local motion is feasible, a robot can safely traverse between two RoAs without collision.

Area condition

Edge condition

Framework

Based on the concept of RoA, we develop a path planning framework called RoA-Planner for obstacle-dense spaces. This framework utilizes a quadtree structure to efficiently represent planning space as a set of RoAs. By treating each RoA as a node and local motions between adjacent RoAs as edges, we design a modified A* algorithm tailored to RoA-based path planning.

Simulation result

Dense environments

First, we compare RoA-Planner with five planners—Baseline 1, Baseline 2, Wellhausen et al., Art-Planner, and TRG-Planner—to analyze path planning performance in highly obstacle-dense environments.

Narrow corridor

Secondly, we compare RoA-Planner against Baseline 2, Wellhausen et al., Art-Planner, and TRG-Planner in terms of robustness when navigating narrow corridors.

Real-world result

In the real-world experiments, we employed the AiDIN-VIII equipped with an Ouster LiDAR OS0-64 to conduct tests in narrow gaps and various obstacle-dense scenarios.

Narrow gap

Various obstacle-dense scenarios

BibTeX

@ARTICLE{11175463,
  author={Son, Yeongwoo and Lee, Hyunyong and Kang, Hansol and Park, Jiman and Nam, Seongwon and Oh, Jaeyoung and Yi, Bumsu and Song, Junha and Choi, Sooyeon and Kim, Bogeun and Song, Daegeun and Choi, Hyouk Ryeol},
  journal={IEEE Transactions on Automation Science and Engineering}, 
  title={RoA-Planner: Rotatable Area based Path Planner in Dense Spaces}, 
  year={2025},
  volume={},
  number={},
  pages={1-1},
  keywords={Collision avoidance;Robots;Service robots;Radio frequency;Safety;Accuracy;Trajectory;Robot sensing systems;Robot kinematics;Shape;Mobile robot;Rectangular robot;Holonomic robot;Path planning;Collision avoidance;Dense environment},
  doi={10.1109/TASE.2025.3612949}}

}