Undergraduate at South China University of Technology
Research Assistant at HKUST(GZ)AI Agents, Multimodal Intelligence & Embodied AI
I am an undergraduate at South China University of Technology, majoring in Big Data Management and Application with a minor in Computer Science.
My interests lie at the intersection of trustworthy AI agents, multimodal intelligence, and embodied systems. I build end-to-end systems that turn language and multimodal models into controllable, traceable workflows - spanning creative generation, enterprise data analysis, AI safety evaluation, and robot learning in simulation.
Google Scholar
Research impact at a glance
Citation metrics will appear here after a Google Scholar profile ID is added.
Education
-
South China University of TechnologyB.M. in Big Data Management and Application
Minor in Computer ScienceSep. 2023 - Jul. 2027
Honors & Awards
-
Honourable Mention, Mathematical Contest in Modeling2025 & 2026
-
Third Prize (Guangdong), Contemporary Undergraduate Mathematical Contest in Modeling2024
-
University Scholarship & Merit Student2024
Work Experience
-
HKUST(GZ)Embodied Intelligence & Physical Intelligent SystemsResearch Assistant2026 - Present
-
ByFusionAI Short Drama ProductionAI Application Development InternApr. 2026 - Present -
AiCoinCryptocurrency Quantitative TradingQuantitative Development Intern2025 - 2026
Selected Publications
The Computational Handoff: A Pre-Rollout Test of Observation-Buffer Reset at Robot Task Boundaries
Yinuo Chen
IROS 2026 Workshop on Full-Shift Robot Co-Workers: Active Perception and Interaction for Human Environments Accepted Poster
A pre-rollout test checks whether resetting a robot policy's observation buffer reproduces a fresh process across task boundaries. In a controlled RoboFactory case study, explicit buffer resets match fresh-process action outputs, while retained history changes them; the test assesses reset conformance rather than downstream task success.
The Computational Handoff: A Pre-Rollout Test of Observation-Buffer Reset at Robot Task Boundaries
Yinuo Chen
IROS 2026 Workshop on Full-Shift Robot Co-Workers: Active Perception and Interaction for Human Environments Accepted Poster
A pre-rollout test checks whether resetting a robot policy's observation buffer reproduces a fresh process across task boundaries. In a controlled RoboFactory case study, explicit buffer resets match fresh-process action outputs, while retained history changes them; the test assesses reset conformance rather than downstream task success.
Action Response Is Not Factual Direction: A Prospective Audit of Robot Video World Models
Yinuo Chen et al.
IROS 2026 Workshop on Physical World Models for Scaling Embodied AI (PWMS) Accepted Workshop Paper Oral
ActionProbe audits whether action-conditioned robot video world models' branch response reflects factual direction, repeatability, and sample-selection utility, showing that response magnitude alone is not sufficient evidence for these stronger claims.
Action Response Is Not Factual Direction: A Prospective Audit of Robot Video World Models
Yinuo Chen et al.
IROS 2026 Workshop on Physical World Models for Scaling Embodied AI (PWMS) Accepted Workshop Paper Oral
ActionProbe audits whether action-conditioned robot video world models' branch response reflects factual direction, repeatability, and sample-selection utility, showing that response magnitude alone is not sufficient evidence for these stronger claims.