notesum.ai

Published at November 17

ModeSeq: Taming Sparse Multimodal Motion Prediction with Sequential Mode Modeling

cs.AI

cs.CV

cs.RO

Released Date: November 17, 2024

Authors: Zikang Zhou¹, Hengjian Zhou², Haibo Hu¹, Zihao Wen¹, Jianping Wang¹, Yung-Hui Li³, Yu-Kai Huang⁴

Aff.: ¹City University of Hong Kong; ²Zhejiang University; ³Hon Hai Research Institute; ⁴Carnegie Mellon University

Arxiv: http://arxiv.org/abs/2411.11911v1

Refer to caption

Method	Ensemble	b-minFDE₆ $\downarrow$	MR₆ $\downarrow$	minADE₆ $\downarrow$	minFDE₆ $\downarrow$
MTR [39]	✓	1.98	0.15	0.73	1.44
MTR++ [40]	✓	1.88	0.14	0.71	1.37
QCNet [54]	$\times$	1.91	0.16	0.65	1.29
ModeSeq (Ours)	$\times$	1.87	0.14	0.63	1.26