Computer ScienceNeurIPS 2024 (Conference on Neural Information Processing Systems)
DiffuserLite: Towards Real-time Diffusion Planning
Z. Dong, J. Hao, et al.
DiffuserLite introduces a super fast, lightweight diffusion planning framework that uses a planning refinement process (PRP) to generate coarse-to-fine trajectories, dramatically reducing redundant modeling and boosting decision-making frequency to 122.2Hz (112.7x faster than prior approaches). It attains state-of-the-art results on D4RL, Robomimic, and FinRL benchmarks and can plug into other diffusion planners. This research was conducted by Zibin Dong, Jianye Hao, Yifu Yuan, Fei Ni, Yitian Wang, Pengyi Li, and Yan Zheng.
Related Publications
Explore these studies to deepen your understanding
Adjacent work that informs or extends this paper's methodology and findings.
Engineering and Technology
Real-time observation of optical rogue waves in spatiotemporally mode-locked fiber lasers
U. Teğin, P. Wang, et al.
Physics
Real-time observation of a correlation-driven sub 3 fs charge migration in ionised adenine
E. P. Månsson, S. Latini, et al.
Earth Sciences
A machine learning estimator trained on synthetic data for real-time earthquake ground-shaking predictions in Southern California
M. Monterrubio-velasco, S. Callaghan, et al.
Engineering and Technology
Thread-based multiplexed sensor patch for real-time sweat monitoring
T. Terse-thakoor, M. Punjiya, et al.
