Publications

A collection of my research work in robotics, 3D computer vision, and embodied intelligence.

PolySplat: Workload-Regime-Aware Rasterization for 3D Gaussian Splatting

Longzan Luo, Bin Cui, Xupeng Miao

Under Review at the 40th Conference on Neural Information Processing Systems (NeurIPS 2026) 2026

PolySplat is a workload-regime-aware 3D Gaussian Splatting rasterizer combining adaptive tile-key emission, asynchronous shared-memory staging, and persistent scheduling. It delivers geometric-mean speedups of 1.26x-6.48x over state-of-the-art rasterizers at lossless visual quality across extreme resolutions and primitive counts.

DARTS: Distribution-Aware Active Rollout Trajectory Shaping for Accelerating LLM Reinforcement Learning

Yujie Wang, Siwei Chen, Longzan Luo, Xinyi Liu, Xupeng Miao, Fangcheng Fu, Bin Cui

Proceedings of the 43rd International Conference on Machine Learning (ICML) 2026

DARTS introduces active distribution shaping for LLM reinforcement learning rollouts via distribution-aware trajectory sampling and adaptive redundancy allocation, delivering up to 1.77x end-to-end acceleration over state-of-the-art RL systems without compromising model quality.

Lift3D Foundation Policy: Lifting 2D Large-Scale Pretrained Models for Robust 3D Robotic Manipulation

Yueru Jia, Jiaming Liu, Sixiang Chen, Chenyang Gu, Zhilue Wang, Longzan Luo, Lily Lee, Pengwei Wang, Renrui Zhang, Zhongyuan Wang, Shanghang Zhang

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2025

We propose the Lift3D framework, which progressively enhances 2D foundation models with implicit and explicit 3D robotic representations to construct a robust 3D manipulation policy.

PaperProjectCode

GarmentLab: A Unified Simulation and Benchmark for Garment Manipulation

Haoran Lu, Ruihai Wu, Yitong Li, Sijie Li, Ziyu Zhu, Chuanruo Ning, Yan Shen, Longzan Luo, Yuanpei Chen, Hao Dong

Advances in Neural Information Processing Systems (NeurIPS) 2024

GarmentLab provides realistic simulation for diverse garments with different physical properties, benchmarking novel garment manipulation tasks in both simulation and the real world.

PaperProjectCode