Projects
Human–robot interaction for optimal ageing, conversational AI in autonomous mobility, engineering education, and transport-safety collaborations.
My research follows a path from understanding users of conversational AI in autonomous mobility to measuring and adapting intelligent behaviour. I now extend that work into embodied human–robot interaction for optimal ageing.
Human–Robot Interaction for Optimal Ageing
My current postdoctoral research with Monash Robotics and the ARC ITTC in Optimal Ageing extends my work on adaptive human–AI interaction into embodied human–robot interaction. I am interested in how robots can understand human preferences, adapt their behaviour over time, and interact safely and intuitively with people in everyday environments, particularly in the context of optimal ageing.
Human–Robot Interaction Social & Assistive Robotics Human Preferences Adaptive Robot Behaviour Human-Centred AI Optimal Ageing
From perception to interaction: Psychological drivers and adaptive conversational agents for shared autonomous vehicles (SAVs)
PhD thesis submitted August 2026; currently under examination.
One-line summary: My PhD research developed a closed-loop framework to predict user acceptance, enable scalable real-time interaction, measure user experience from dialogue, and regulate conversational agent responses via closed-loop feedback + bang–bang control—demonstrated in SAV interaction.
What this research developed
- Prediction: data-driven modelling and interpretable visualisation of psychological drivers of acceptance.
- Interaction platform: an LLM-based conversational testbed for open-domain human–agent interaction studies at scale.
- Measurement: interaction-level sentiment/subjectivity analysis to quantify user experience signals from dialogue.
- Regulation: closed-loop feedback control for real-time adjustment of conversational agent response attributes.
Key outputs
- Predictive framework + chord diagram visualisation workflow for psychological drivers.
- Ethics-approved user studies and a curated interaction dataset for quantitative analysis.
- Measurement pipeline for experience signals in open-domain dialogue.
- Control-theoretic framework for adaptive, value-aligned interaction regulation.
Highlights
Closed-loop feedback control LLM conversational agents Experience measurement Human studies (ethics)
- General contribution: controllable, user-aware interaction design (domain-agnostic).
- Case study: SAVs as a real-world interactive autonomous system.
- Why it matters: supports transparency, user controllability, and psychological safety in adaptive interactions.
Methods
Behavioural modelling ML prediction User study design NLP sentiment analysis Control theory Data pipelines
AI in Engineering Education: LLM-based Learning Assistant (CivASK) & Evaluation
One-line summary: I collaborate on designing and evaluating retrieval-augmented LLM systems to provide reliable, equitable student support, with transparent evaluation and classroom-facing deployment.
What this program delivers
- System development: a retrieval-augmented learning assistant grounded in course materials for faithful answers.
- Evaluation framework: protocols to assess accuracy, coverage, and failure modes, supporting responsible use.
- Classroom translation: ongoing pilots analysing real student–assistant interactions and teaching impact.
Key outputs
- RAG pipeline and prompt design to improve answer faithfulness.
- Evaluation design for reliability and coverage across student needs.
- Evidence from teaching-context deployment informing responsible AI use.
Highlights
RAG systems LLM evaluation Real teaching context Equity & accessibility Responsible AI
- Focus: transparent, evidence-based deployment of LLM support tools.
- Strength: connecting system design to real user outcomes (students).
Methods
Retrieval-augmented generation Prompt engineering Learning analytics Human-centred evaluation
Other Collaborations: Behaviour Change and Road Safety
One-line summary: I contribute to interdisciplinary transport-safety research that applies behavioural evidence and intervention design to risky riding behaviour among delivery riders.
What this collaboration adds
- Behaviour-change focus: co-authored a behaviour change wheel-informed intervention study targeting risky riding behaviours among delivery riders.
- User safety lens: extends my interest in human-centred transport systems from automated mobility to road-user behaviour and safety outcomes.
- Interdisciplinary publication: contributed to a Q1 journal article in Transportation Research Part F: Traffic Psychology and Behaviour.
Key output
- Ye, F., Wang, C., Zhang, Z., Zhu, T., Dou, H., & Guo, L. (2026). Behaviour change wheel-informed intervention for delivery-rider safety.
Highlights
Traffic psychology Behaviour change wheel Delivery-rider safety Intervention evaluation
- Contribution type: co-authored interdisciplinary applied road-safety research.
- Connection to my work: human behaviour, safety, and user-centred transport systems.
Methods
Road-safety research Behavioural analysis Intervention evaluation Manuscript development
Collaboration
I welcome collaborations on human-centred AI and HRI, adaptive interaction in autonomous mobility and robotics, and technologies supporting optimal ageing.