preprint

Action Emergence from Streaming Intent

Streaming Intent — a chain-of-thought-driven, intent-steerable VLA that delivers intent-faithful controllability on Waymo E2E.

Pengfei Jing, Victor Shea-Jay Huang, Hengtong Lu, Jifeng Dai, Xie Yan, Benjin ZHU

Driving Intents Amplify Planning-Oriented Reinforcement Learning

DIAL — intent-CFG sampling + multi-intent GRPO lift WOD-E2E preference RL past prior best (RAP 8.5) and human demos for the first time.

Hengtong Lu, Victor Shea-Jay Huang, Chengmin Yang, Pengfei Jing, Jifeng Dai, Yan Xie, Benjin ZHU

MindVLA-U1: VLA Beats VA with Unified Streaming Architecture for Autonomous Driving

A unified streaming VLA architecture that surpasses experienced human drivers on WOD-E2E for the first time, at VA-level latency.

Yuzhou Huang, Benjin ZHU, Hengtong Lu, Victor Shea-Jay Huang, Haiming Zhang, Wei Chen, Jifeng Dai, Yan Xie, Hongsheng Li

EqCo: Equivalent Rules for Self-supervised Contrastive Learning

EqCo (Equivalent Rules for Contrastive Learning) to make self-supervised learning irrelevant to the number of negative samples in the contrastive learning framework.

Benjin ZHU, Junqiang Huang, Zeming Li, Xiangyu Zhang, Jian Sun

EqCo: Equivalent Rules for Self-supervised Contrastive Learning

AutoAssign: Differentiable Label Assignment for Dense Object Detection

End-to-end dense object detection without hyperpamameters.

Benjin ZHU, Jianfeng Wang, Zhengkai Jiang, Fuhang Zong, Songtao Liu, Zeming Li, Jian Sun

AutoAssign: Differentiable Label Assignment for Dense Object Detection

Class-balanced Grouping and Sampling for Point Cloud 3D Object Detection

This report presents our method which wins the nuScenes 3D Detection Challenge held in Workshop on Autonomous Driving(WAD, CVPR 2019).

Benjin ZHU, Zhengkai Jiang, Xiangxin Zhou, Zeming Li, Gang Yu

Class-balanced Grouping and Sampling for Point Cloud 3D Object Detection