日本フィジカルAI新聞

世界のフィジカルAIを、日本語で。

週刊ニュースレター購読
arXiv:2509.16812

SMART-3D: Three-Dimensional Self-Morphing Adaptive Replanning Tree

SMART-3D: Three-Dimensional Self-Morphing Adaptive Replanning Tree

シェア:XThreadsFacebookLINEはてブBluesky

著者: Priyanshu Agrawal, Shalabh Gupta, Zongyuan Shen

分類: cs.RO, cs.AI, cs.SY, eess.SY

原文アブストラクト

This paper presents SMART-3D, an extension of the SMART algorithm to 3D environments. SMART-3D is a tree-based adaptive replanning algorithm for dynamic environments with fast moving obstacles. SMART-3D morphs the underlying tree to find a new path in real-time whenever the current path is blocked by obstacles. SMART-3D removed the grid decomposition requirement of the SMART algorithm by replacing the concept of hot-spots with that of hot-nodes, thus making it computationally efficient and scalable to 3D environments. The hot-nodes are nodes which allow for efficient reconnections to morph the existing tree to find a new safe and reliable path. The performance of SMART-3D is evaluated by extensive simulations in 2D and 3D environments populated with randomly moving dynamic obstacles. The results show that SMART-3D achieves high success rates and low replanning times, thus highlighting its suitability for real-time onboard applications.