Mingsheng Li

Qwen-VLA: Unifying Vision-Language-Action Modeling across Tasks, Environments, and Robot Embodiments

Co-First Author

project paper github

• An all-in-one unified vision-language-action generalist model that handles complex manipulation tasks, long-horizon navigation, and trajectory prediction within a single framework, generalizing across diverse tasks, environments, and robot embodiments.

Qwen3-VL Technical Report

Core Author

project paper github

• A next-generation vision-language model featuring advanced visual reasoning, native dynamic-resolution processing, long-context video understanding, and multilingual support with both dense and MoE variants.

Qwen3.5-Omni Technical Report

Author

project paper

• An upgraded end-to-end omni model with enhanced multimodal reasoning, real-time streaming interaction, and stronger cross-modal understanding and generation across text, image, audio, and video modalities.

Qwen3-Omni Technical Report

Author

project paper github

• A natively end-to-end multilingual omni model supporting real-time streaming interaction across text, image, audio, and video with unified perception, understanding, and generation.

StarVLA: A Lego-like Codebase for Vision-Language-Action Model Developing

Author

project paper github

• A modular, Lego-like open-source codebase for agile VLA development, supporting seamless integration of diverse VLMs, action heads, and world models with unified training, evaluation, and deployment across robotics benchmarks.

Unify Robot Actions in Camera Frame

Sicheng Xie, Lingchen Meng, Zijie Diao, Haidong Cao, Zhiying Du, Shuyuan Tu, Jiaqi Leng, Qiuyue Wang, Mingsheng Li, Shuai Bai, Zuxuan Wu, Yu-Gang Jiang

paper github

• A training-free, robot-independent pipeline that estimates camera extrinsics for offline datasets and unifies robot actions in camera frame, enabling scalable cross-embodiment learning without any manual calibration.

FineVLA: Fine-Grained Instruction Alignment for Steerable Vision-Language-Action Policies

Xintong Hu, Xuhong Huang, Jinyu Zhang, Yutong Yao, Yuchong Sun, Qiuyue Wang, Mingsheng Li, Sicheng Xie, Yitao Liu, Junhao Chen, Yixuan Chen, Yingming Zheng, Shuai Bai, Tao Yu

paper project github benchmark

• Fine-grained VLA supervision that improves steerable robot control.

StructChart: Perception, Structuring, Reasoning for Visual Chart Understanding

Renqiu Xia, Haoyang Peng, Hancheng Ye, Mingsheng Li, Xiangchao Yan, Peng Ye, Botian Shi, Yu Qiao, Junchi Yan, Bo Zhang

IEEE Transactions on Pattern Analysis and Machine Intelligence (T-PAMI), 2026.

project paper github

• A unified approach for visual chart perception and reasoning using Structured Triplet Representations.

VLM4VLA: Revisiting Vision-Language-Models in Vision-Language-Action Models

Jianke Zhang, Xiaoyu Chen, Qiuyue Wang, Mingsheng Li, Yanjiang Guo, Yucheng Hu, Jiajun Zhang, Shuai Bai, Junyang Lin, Jianyu Chen

International Conference on Learning Representations (ICLR), 2026.

paper github

• A unified framework for studying how VLMs affect VLA performance.

Chimera: Improving Generalist Model with Domain-Specific Experts

Tianshuo Peng, Mingsheng Li (co-first), Jiakang Yuan, Hongbin Zhou, Renqiu Xia, Renrui Zhang, Lei Bai, Song Mao, Bin Wang, Aojun Zhou, Botian Shi, Tao Chen, Bo Zhang, Xiangyu Yue

IEEE/CVF International Conference on Computer Vision (ICCV), 2025.

paper github

• Cost-effective general-specialist collaboration to improve LMMs.

GeoX: Geometric Problem Solving Through Unified Formalized Vision-Language Pre-training

Renqiu Xia, Mingsheng Li (co-first), Hancheng Ye, Wenjie Wu, Hongbin Zhou, Jiakang Yuan, Tianshuo Peng, Xinyu Cai, Xiangchao Yan, Bin Wang, Conghui He, Botian Shi, Tao Chen, Junchi Yan, Bo Zhang

International Conference on Learning Representations (ICLR), 2025.

paper github

• Formalized pre-training for geometry problem solving.

WI3D: Weakly Incremental 3D Detection via Vision Foundation Models

Mingsheng Li, Sijin Chen, Shengji Tang, Hongyuan Zhu, Yanyan Fang, Xin Chen, Zhuoyuan Li, Fukun Yin, Tao Chen

IEEE Transactions on Multimedia (T-MM), 2025.

project paper

• Introducing new categories to 3D detectors via 2D foundation models.

3DET-Mamba: State Space Model for End-to-End 3D Object Detection

Mingsheng Li, Jiakang Yuan, Sijin Chen, Lin Zhang, Anyu Zhu, Xin Chen, Tao Chen

Conference on Neural Information Processing Systems (NeurIPS), 2024.

project paper

• End-to-end 3D detection with Mamba-based representation learning.

M3DBench: Let's Instruct Large Models with Multi-modal 3D Prompts

Mingsheng Li, Xin Chen, Chi Zhang, Sijin Chen, Hongyuan Zhu, Fukun Yin, Gang Yu, Tao Chen

European Conference on Computer Vision (ECCV), 2024.

project paper github

• Large-scale 3D-language dataset with interleaved multimodal prompts.

Vote2Cap-DETR++: Decoupling Localization and Describing for End-to-End 3D Dense Captioning

Sijin Chen, Hongyuan Zhu, Mingsheng Li, Xin Chen, Peng Guo, Yinjie Lei, Gang Yu, Taihao Li, Tao Chen

IEEE Transactions on Pattern Analysis and Machine Intelligence (T-PAMI), 2024.

paper github

• Decoupled queries for 3D localization and dense captioning.

LL3DA: Visual Interactive Instruction Tuning for Omni-3D Understanding, Reasoning, and Planning

Sijin Chen, Xin Chen, Chi Zhang, Mingsheng Li, Gang Yu, Hao Fei, Hongyuan Zhu, Jiayuan Fan, Tao Chen

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024.

project paper github

• 3D-LLMs for visual and textual interactions in complex 3D scenes.

Lightweight Model Pre-training via Language Guided Knowledge Distillation

Mingsheng Li, Lin Zhang, Mingzhen Zhu, Zilong Huang, Gang Yu, Jiayuan Fan, Tao Chen

IEEE Transactions on Multimedia (T-MM), 2024.

paper github

• Language-guided knowledge distillation for lightweight model pre-training.