Density-Driven Progressive Shape Formation for Swarm Robots in Dynamic Environments
Yalun Xiang, Jintao Song, Zhicheng Zheng, Xiaokang Lei, Xingguang Peng
Abstract
This letter presents a novel density-based control framework for shape formation in swarm robotic systems, operating exclusively on local perception without requiring communication or identity recognition among robots and targets. Inspired by Smoothed Particle Hydrodynamics (SPH), each robot estimates a local spatial density using kernel functions and regulates its motion based on the gradient of a constructed target density field. Unlike existing methods that depend on global coordination or explicit target assignments, the proposed approach enables fully distributed shape formation through local density regulation alone. By adjusting the reference density, the system supports autonomous scaling in both the number of robots and the geometric size of the target shape, seamlessly transitioning among shape-surrounding, double-layer contour, and boundary-conformal formations. A composite control law integrates density-driven forces, repulsive interactions for collision avoidance, and damping for dynamic stability. The effectiveness and robustness of the method are validated through extensive simulations and physical experiments with 50 robots under a motion capture laboratory setup. Results demonstrate the scalability, resilience, and practicality of the approach for decentralized swarm shape formation.
BibTeX
@inproceedings{ral2026_densitydrivenpro,
title = {Density-Driven Progressive Shape Formation for Swarm Robots in Dynamic Environments},
author = {Yalun Xiang and Jintao Song and Zhicheng Zheng and Xiaokang Lei and Xingguang Peng},
booktitle = {RA-L 2026},
year = {2026}
}