Learning Unions of Convex Sets via Invertible Latent Decomposition for Path Planning
Learns an invertible map plus convex polytopes in its latent space, so convex-set planners work in high-DoF spaces and paths decode back feasible. Tested from 2D to 14-DoF; no false positives after refinement.联合学习可逆映射及其潜空间中的凸多面体并集，使基于凸集的规划可用于高自由度空间，解码回原空间的路径仍可行。在 2D 至 14 自由度场景测试，细化后未观察到误报。
BibTeX
@misc{yoon2026learning,
title = {{Learning Unions of Convex Sets via Invertible Latent Decomposition for Path Planning}},
author = {Taerim Yoon and Dongho Kang and Kisang Park and Junha Cha and Stelian Coros and Sungjoon Choi},
year = {2026},
eprint = {2606.12027},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.12027},
}