研究经费 Research Grants

  • 面向不稳定通信链路的安全人工智能沙箱控制架构设计研究，广东省面上项目, 在研
    资助单位：广东省基础与应用基础研究基金委员会
    主持(PI), Grant No:2026A1515010222，2026.01-2028.12, 100k RMB

  • 形式化方法驱动的具身智能安全，青年科技人才培育计划, 在研
    资助单位：广东省科学技术协会
    主持(PI), Grant No:SKXRC2025468，2025.01-2026.12, 200k RMB

  • 面向信息物理系统的大语言模型可控策略生成研究，广州市基础与应用基础研究项目, 在研
    资助单位：广州市科技局
    主持(PI), Grant No:2025A03J4493，2025.01-2026.12, 250k RMB

  • AI推理结果的自动化质量评价机制研究，校企联合实验室子项目, 在研
    资助单位：香港科技大学（广州）- 蜂助手AI与IOT智核联合实验室
    主持(PI),2024.12-2025.10, 450k RMB

  • 融合学科下数学学科教学实践探索——以《人工智能中的数学》课程教学为例，港科广实践项目（萌芽项目）, 在研
    资助单位：香港科技大学（广州）
    主持(PI), Grant No: HKUST(GZ)-ROP2025044, 2025.04-2026.04, 20K RMB

  • 面向低空安全的无人机智能航迹规划与控制研究，广东省普通高校重点领域专项人工智能（智能机器人）重点领域项目, 在研
    资助单位：广东省教育厅
    参与（Participating）, Grant No: 2025ZDZX3052, 2026.01-2028.12, 200K RMB

  • 广州市人工智能前沿交叉科学重点实验室, 基础研究计划-市重点实验室建设项目, 结题
    资助单位：广州市科技局
    参与（Participating）,Grant No:2023A03J0008; 2023.04-2025.03, 3M RMB

部分项目简介 Selected Research Projects

Safe-sec-visor Architecture for Cyber-Physical Systems

Make AI trustworthy in CPS through sandboxing


In this project, we propose a Safe-sec-visor Architecture for sandboxing arbitary types of unverified controllers (e.g., AI-based controllers) to provide a system-level safety and security guarantees for Cyber-Physical Systems.

Secure-by-construction Controller Synthesis

Constructing safe and secure Cyber-Physical Systems


In this project, we focus on provably-correct synthesis scheme for building controllers that ensure safety and security properties over Cyber-Physical Systems simultaneously.

Model-based Controller Synthesis

Towards correct-by-construction embedded controllers


In this project, we investigate both abstraction-based and abstraction-free approaches for synthesizing correct-by-construction controllers against high-level logical properties, such as \(\omega\)-regular properties, over systems with continuous state and input sets.

Data-driven Controller Synthesis

Achieve formal safety guarantees from data


In this project, we investigate how to synthesize correct-by-construction controllers directly based on data collected from the black-box systems with continuous state and input sets.