notesum.ai

Published at November 22

DiffusionDrive: Truncated Diffusion Model for End-to-End Autonomous Driving

cs.CV

cs.RO

Released Date: November 22, 2024

Authors: Bencheng Liao¹, Shaoyu Chen², Haoran Yin³, Bo Jiang², Cheng Wang¹, Sixu Yan², Xinbang Zhang³, Xiangyu Li³, Ying Zhang³, Qian Zhang³, Xinggang Wang²

Aff.: ¹Institute of Artificial Intelligence, Huazhong University of Science & Technology; ²School of EIC, Huazhong University of Science & Technology; ³Horizon Robotics

Arxiv: http://arxiv.org/abs/2411.15139v1

Method	Input	Img. Backbone	Anchor	NC $\uparrow$	DAC $\uparrow$	TTC $\uparrow$	Comf. $\uparrow$	EP $\uparrow$	PDMS $\uparrow$
UniAD [13]	Camera	ResNet-34 [11]	0	97.8	91.9	92.9	100	78.8	83.4
PARA-Drive [38]	Camera	ResNet-34 [11]	0	97.9	92.4	93.0	99.8	79.3	84.0
LTF [6]	Camera	ResNet-34 [11]	0	97.4	92.8	92.4	100	79.0	83.8
Transfuser [6]	C & L	ResNet-34 [11]	0	97.7	92.8	92.8	100	79.2	84.0
DRAMA [43]	C & L	ResNet-34 [11]	0	98.0	93.1	94.8	100	80.1	85.5
VADv2- $\mathcal{V}_{8192}$ [3]	C & L	ResNet-34 [11]	8192	97.2	89.1	91.6	100	76.0	80.9
Hydra-MDP- $\mathcal{V}_{8192}$ [22]	C & L	ResNet-34 [11]	8192	97.9	91.7	92.9	100	77.6	83.0
Hydra-MDP- $\mathcal{V}_{8192}$ -W-EP [22]	C & L	ResNet-34 [11]	8192	98.3	96.0	94.6	100	78.7	86.5
DiffusionDrive (Ours)	C & L	ResNet-34 [11]	20	98.2	96.2	94.7	100	82.2	88.1