Ongoing Research Projects

Research at a Glance: We study adaptive human–AI systems, focusing on mutual adaptation between people and AI over time. Our work examines human-in-the-loop interactions with AI, autonomous systems, and robots across application domains.

NSF EDSE automation intervention project graphic
Sponsor: NSF EDSE PIs: Yunmei Liu and David Kaber

Funded ProjectCognitive-Emotional State Assessment for Adaptive Automation Intervention

This project examines the interplay among cognitive workload, emotion, task demands, and human performance to inform adaptive automation intervention. The research develops cognitive-emotional state assessment methods that support automation responsive to changing task conditions and operator needs.

We integrate physiological sensing, behavioral performance data, and machine learning to assess workload, emotion, and task engagement. The resulting models are intended to guide the timing and form of automation support during safety-critical tasks.

Website
AI-enabled virtual reality manufacturing foundry project graphic
Sponsor: NSF FINDERS FOUNDRY PI: Yunmei Liu

Funded ProjectAI-Enabled Virtual Reality Manufacturing Foundry

This NSF FINDERS FOUNDRY Planning project advances the design of an AI-enabled VR manufacturing assistant that combines intelligent assistance with immersive manufacturing simulation. A central technical focus is the AI assistant itself: its capabilities for supporting manufacturing exploration and decision-making through feedback, reflection, and safety guidance, and its integration into the interactive VR environment.

The planning phase will establish AI assistant requirements, VR scenarios, and human–AI interaction models, informed by student and educator needs. High school manufacturing education provides the application context for the integrated AI–VR system, supporting students’ understanding of manufacturing workflows and exploration of career pathways. These activities will establish a foundation for subsequent platform development.

Aggressive Driving project graphic
Sponsor: UofL Jon Rieger Seed Grant PI: Yunmei Liu

Funded ProjectAggressive Driving in Mixed Traffic

This project examines human driver responses to automated-vehicle driving styles and their implications for human-aware AV control. The research connects driver workload, trust, stress, and behavior with safety and efficiency in mixed human–AV traffic.

We combine CARLA–SUMO human-in-the-loop experiments, empirical driver modeling, and SUMO/TraCI traffic-flow simulation to examine the effects of surrounding AV aggressiveness and human driver aggressiveness on speed choice, braking, lane changes, congestion, and surrogate safety outcomes. The work aims to inform adaptive AV policies that account for human responses.

Presentation 1: HFES 2025

Publications From This Project

  1. Das, U.*, Chen, Y.*, Chowdhury, M. H.*, and Liu, Y.† (2026). A vision-based multimodal framework for quantifying novice driver behavioral responses to aggressive overtaking in continuous simulator traffic. International Conference on Applied Human Factors and Ergonomics. Accepted.
  2. Chen, Y.*, Chowdhury, M. H.*, and Liu, Y.† (2026). From driver intent to road-user perception: a narrative and systematic review of perceived aggressive driving. International Conference on Applied Human Factors and Ergonomics. Accepted.

Video Demo

This demo illustrates a representative aggressive-overtaking scenario in continuous mixed-traffic simulation and demonstrates the behavioral responses detected using our vision-based multimodal framework. It accompanies the study: Das, U.*, Chen, Y.*, Chowdhury, M. H.*, and Liu, Y.† (2026), “A Vision-Based Multimodal Framework for Quantifying Novice Driver Behavioral Responses to Aggressive Overtaking in Continuous Simulator Traffic,” International Conference on Applied Human Factors and Ergonomics (AHFE), accepted.

Teen Driver Safety Education project graphic
Sponsor: NHTSA with KYTC Project No.: M3DA-2027-14 PI: Yunmei Liu

Funded ProjectTeen Driver Safety Education

This project translates human-centered transportation research into simulator-based education that supports teen drivers’ recognition of unsafe behavior and reflection on driving decisions. The curriculum addresses distracted and aggressive driving in Kentucky high schools, aligned with KYTC and KOHS safety messaging.

Teacher lesson plans, student materials, classroom rotations and debriefs, and an interactive driving scenario will integrate feedback on speed choice, following distance, braking, and hazard response. An optional wearable-sensor component will provide a student-friendly driver-state snapshot. A classroom pilot will assess feasibility and learning value before delivery of the educator toolkit.

EMG-based upper-limb prosthetic control evaluation setup
Sponsor: NA

Human-Centered Design for Natural Upper-Limb Prosthetic Control

This project evaluates the effects of prosthetic control strategies on task performance, cognitive workload, and user experience. The research informs interfaces that better align with movement intentions and support natural, intuitive interaction with assistive devices.

Our prior work compared alternative control modes and evaluated VR as a platform for prosthetic manipulation tasks and early-stage design. These studies provide evidence for selecting and refining control interfaces and developing rehabilitation training.

Seizure-EEG project graphic
Sponsor: NA

AI-Driven Scalp EEG Modeling for Seizure Detection and Signal Enhancement

This project develops AI representations of multichannel scalp EEG for seizure detection and signal enhancement. It advances methods for interpreting complex physiological data, with applications in neurological assessment.

Spatial, visual, and geometry-aware representations support robust seizure detection, EEG spatial super-resolution, and biomedical signal modeling.

Communication Barriers project graphic
Sponsor: NA

Communication Barriers in Patient-Provider Interactions

This project characterizes linguistic, cultural, psychological, and mental-model barriers in patient–provider communication and their relationships with healthcare experiences and outcomes. The research informs future adaptive assistance responsive to human needs and clinical context.

Building on a scoping review, we map overlapping barriers and identify opportunities for human and technological support, including interpreter services, cultural and empathy training, plain-language and visual aids, and AI-enabled tools. This evidence helps define the communication needs that future support systems must recognize and address.

Research-Related News

View all news

Showing all 21 news items.

Ph.D. application emails must use this subject format: MINDxAI Ph.D. 2027 — Last Name — Research Fit.