POSTDOCTORAL RESEARCHER · PEKING UNIVERSITY 博士后研究员 · 北京大学

RUIMA

A latent manifold maps observations into multiple predicted future states

I build trustworthy intelligent systems at the intersection of embodied AI, multimodal learning, and content security. 我的研究聚焦于具身智能、多模态学习与内容安全的交叉领域，致力于构建可信智能系统。

ruima AT pku.edu.cn GitHub ↗ Google Scholar ↗ 谷歌学术 ↗

Explore publications 查看论文

18 Papers 论文

04 Focus Areas 研究方向

07 Patents 专利

05 Honors 荣誉

361 Google Scholar 谷歌学术

01 / ABOUT01 / 个人简介

About 个人简介

I am a postdoctoral researcher at Peking University, co-advised by Researcher Shanghang Zhang. I study safe embodied intelligence, trustworthy multimodal models, and verifiable content. 我现为北京大学博士后研究员，合作导师为仉尚航研究员，研究方向为安全具身智能、可信多模态模型与可验证内容。

02 / EXPERIENCE02 / 工作与教育

Experience & Education 工作与教育经历

2026 — PRESENT至今

WORK工作
Peking University北京大学

Postdoctoral Researcher 博士后研究员

Co-advisor: Researcher Shanghang Zhang 合作导师：仉尚航研究员

Embodied AI具身智能 Multimodal Trust多模态可信 Content Security内容安全
2025.07 — 2025.12

WORK工作
Ningxia University宁夏大学

Assistant Professor & M.S. Supervisor 助理教授、硕士生导师

Computer vision research, teaching, graduate supervision, and grant development 开展计算机视觉研究、教学、研究生指导与科研项目申报
2024.09 — 2025.05

WORK工作
Yunsheng Intelligent Technology Co., Ltd.云圣智能科技有限公司

AI Algorithm Researcher & Research Intern 人工智能算法研究员、研究实习生

Full-time and internship roles · End-to-end autonomous flight for quadrotor drones under the VLA framework 历任人工智能算法研究员与研究实习生 · 基于视觉—语言—动作（VLA）框架开展四旋翼无人机端到端自主飞行研究
2020 — 2025

EDUCATION教育
Peking University北京大学

Ph.D. · Computer Science · Computer Vision & Information Security 博士 · 计算机科学 · 计算机视觉与信息安全方向

Advisors: Dr. Xiaodong Xie and Academician Wen Gao 导师：解晓东研究员、高文院士
2017 — 2020

EDUCATION教育
University of Chinese Academy of Sciences中国科学院大学

M.S. · Physics · Optical Imaging & Information Hiding 硕士 · 物理学 · 光学成像与信息隐藏方向

Advisor: Prof. Yishi Shi 导师：史祎诗教授
2013 — 2017

EDUCATION教育
North China Electric Power University华北电力大学

B.E. · Automation · Measurement & Control Technology and Instrumentation 学士 · 自动化 · 测控技术与仪器方向

03 / RESEARCH03 / 研究方向

Research Interests 研究方向

Editorial illustration of safe embodied intelligence

R/01

Embodied Intelligence Safety 具身智能安全

Safe decision-making, continual adaptation, and robust perception for agents acting in physical environments. 面向物理环境智能体的安全决策、持续适应与鲁棒感知。

Editorial illustration of trustworthy multimodal learning

R/02

Trustworthy Multimodal Models 可信多模态模型

Reliability, privacy, and generalization across visual, language, and action spaces. 研究视觉、语言与动作空间中的可靠性、隐私保护与泛化能力。

Editorial illustration of copyright authentication and watermarking

R/03

Copyright Authentication & Watermarking 版权认证与水印

Robust blind watermarking and content authentication under real-world distortions and black-box noise. 面向真实失真与黑盒噪声的鲁棒盲水印和内容认证。

Editorial illustration of generated-content traceability

R/04

Generated-Content Traceability 生成内容溯源

Attribution, provenance, and forensic signals for synthetic images, video, and generative systems. 面向合成图像、视频与生成系统的归因、来源追踪与取证信号。

04 / PUBLICATIONS04 / 论文成果

Publications 论文成果

A selection of representative published and formally accepted papers; the full publication count follows Google Scholar. 以下仅展示部分代表性论文；首页论文总数以谷歌学术为准。

SafeDojo interactive world model and constrained reinforcement learning pipeline

2026CoRLCONFERENCE会议

SafeDojo: Safe Reinforcement Learning for VLA via Interactive World Model

K. Tang, P. Jia, Z. Chu, J. Wu, R. Ma, J. Cao, F. Zhao, S. Chen, Y. Guo, X. Chi, C.-K. Fan, K. Zhang, J. Xu, F. Yang, W. Mi, X. Ju, J. Tang, S. Zhang

Accepted to CoRL 2026 · arXiv:2606.20698被 CoRL 2026 接收 · arXiv:2606.20698

Method overview figure from Group-of-Latents

2026ACM MULTIMEDIACONFERENCE会议

Group-of-Latents: Perceptual Video Compression at Extreme Bitrates via Masked Latent Generative Modeling

S. Wang, J. Xu, P. Jia, Z. Hao, S. Qian, F. Zhao, R. Ma, X. Ju, J. Tang, X. Xie, S. Zhang, H. Jia

Accepted to ACM Multimedia 2026 · arXiv:2607.19437被 ACM Multimedia 2026 接收 · arXiv:2607.19437

Core method figure from See to Act, Adapt to Act

2026ACM MULTIMEDIACONFERENCE会议

See to Act, Adapt to Act: Action-Stable Continual Test-Time Adaptation for Multimodal Embodied Intelligence

P. Jia, X. Chi, R. Ma, J. Xu, C.-K. Fan, S. Zhang

Core method figure from Scene-Adaptive Unsupervised Crowd Counting

2025IEEE TCSVTJOURNAL期刊

Scene-Adaptive Unsupervised Crowd Counting for Video Surveillance

R. Ma, Y. Hou, C. Li, H. Jia, X. Xie

vol. 35, no. 7, pp. 6910–6925第35卷第7期，6910–6925页

Core method figure from Enhanced Blind Watermarking Against Black-Box Noise

2024IEEE ICMEORAL口头报告

Enhanced Blind Watermarking Against Black-Box Noise: Leveraging CIN Framework

R. Ma, M. Guo, P. Jia, C. Li, Y. Hou, Y. Li, X. Xie, S. Zhang

2023IEEE TCSVTJOURNAL期刊

Frame-Recurrent Video Crowd Counting

Y. Hou, S. Zhang, R. Ma, H. Jia, X. Xie

vol. 33, no. 9, pp. 5186–5199第33卷第9期，5186–5199页

Core method figure from Towards Blind Watermarking

2022ACM MULTIMEDIACONFERENCE会议

Towards Blind Watermarking: Combining Invertible and Non-invertible Mechanisms

R. Ma, M. Guo, Y. Hou, F. Yang, Y. Li, H. Jia, X. Xie

Project & code 项目与代码

2022JOURNAL OF UCASJOURNAL期刊

Chosen-Plaintext Attack to Linear Unitary Optical Encryption Systems

D. Miao, R. Ma, X. Sun, Y. Shi

vol. 39, no. 2, p. 165第39卷第2期，第165页

Core method figure from Optical Information Hiding with Non-Mechanical Ptychography

2021OPTICS & LASERS IN ENGINEERINGJOURNAL期刊

Optical Information Hiding with Phase-Only and Non-Mechanical Ptychography

R. Ma, Y. Li, H. Jia, Y. Shi, X. Xie, T. Huang

vol. 141, article 106569第141卷，文章编号106569

Core method figure from Multi-Image and Color Image Encryption

2021OPTICS COMMUNICATIONSJOURNAL期刊

Multi-Image and Color Image Encryption via Multi-Slice Ptychographic Encoding

J. Zhang, D. Yang, R. Ma, Y. Shi

vol. 485, article 126762第485卷，文章编号126762

Core method figure from Scanning-Position Error-Correction Algorithm

2020CHINESE PHYSICS LETTERSJOURNAL期刊

Scanning-Position Error-Correction Algorithm in Dual-Wavelength Ptychographic Microscopy

R. Ma, S. Zhang, T. Ruan, Y. Tao, Y. Shi

vol. 37, no. 4第37卷第4期

Core method figure from Sharpness-Statistics-Based Auto-Focusing

2020OPTICS & LASERS IN ENGINEERINGJOURNAL期刊

Sharpness-Statistics-Based Auto-Focusing Algorithm for Optical Ptychography

R. Ma, D. Yang, T. Yu, T. Li, Y. Shi

vol. 128, article 106053第128卷，文章编号106053

Core method figure from Natural Speckle-Based Watermarking

2020OPTICS EXPRESSJOURNAL期刊

Natural Speckle-Based Watermarking with Random-Like Illuminated Decoding

X. Sun, S. Zhang, R. Ma, Y. Tao, Y. Zhu, D. Yang, Y. Shi

vol. 28, no. 21第28卷第21期

2020ACTA PHYSICA SINICAJOURNAL期刊

Enhanced-Visual-Cryptography-Based Optical Information Hiding System

Y. Tao, D. Yang, R. Ma, Y. Zhu, Y. Shi

vol. 69, no. 14第69卷第14期

05 / AUTHORIZED PATENTS05 / 已授权专利

Authorized Patents 已授权发明专利

Seven granted Chinese invention patents in optical imaging and information security. 共 7 项中国发明专利获得授权，主要涉及光学成像与信息安全。

2023-03-21CN111563938B