Zoned Artificial Repulsion: Path Planning Through Local Minima for Multiple-Robot Dexterous Micromanipulation

Tala Dannawi Aissaoui; Redwan Dahmouche

IROS 20250 citations

Zoned Artificial Repulsion: Path Planning Through Local Minima for Multiple-Robot Dexterous Micromanipulation

Abstract

We introduce in this paper an original planning algorithm, Zoned Artificial Repulsion (ZAR∗), designed for multiple-robot micromanipulation. The algorithm combines a modified Artificial Potential Field (APF) with A∗ to efficiently compute micromanipulation trajectories. The motivation behind ZAR∗ is to integrate the advantages of APF and A∗ while avoiding their limitations. While APF is computationally efficient but prone to local minima, A∗ guarantees completeness but has high complexity. ZAR∗ employs a modified repulsive field to segment the configuration space into distinct zones, each converging to a unique local minimum. These zones are reduced to nodes in a graph, allowing A∗ to compute the inter-zone transitions. The modified APF then handles navigation within the zones. This method ensures the completeness of the algorithm, avoids local minima, and significantly reduces the number of nodes in the graph, leading to a highly efficient algorithm in terms of processing time and path cost. We compare ZAR∗ against A∗, APF, RRT, RRT ∗, and PRM, as well as more recent hybrid algorithms. On average, ZAR∗ reduces the number of nodes by 909 times and speeds up path construction efficiency (time × cost) by 499 times on average, while maintaining a 100% success rate. It also performs more than 4 times better than the best hybrid alternative of standard algorithms, making it suitable for dexterous micromanipulation tasks.

BibTeX

@inproceedings{iros2025_zonedartificialr,
  title = {Zoned Artificial Repulsion: Path Planning Through Local Minima for Multiple-Robot Dexterous Micromanipulation},
  author = {Tala Dannawi Aissaoui and Redwan Dahmouche},
  booktitle = {IROS 2025},
  year = {2025}
}