Research | ICEL | NC State ISE
Research
Last Updated: 02/26/2025 | All information is accurate and up-to-date
ICEL’s research brings together ideas from cognitive psychology, ergonomics and artificial intelligence. This combination helps create systems that boost your abilities and ensure safety, usability and satisfaction. Key research areas include human-AI teamwork, social intelligence in AI, human behavior modeling and Human-in-the-Loop (HITL) computing. These projects support fields like automated driving, healthcare and other critical areas.
Research Projects
Human Cognition Models to Inspire AVs in Interaction Scenes
Autonomous vehicles (AVs) face challenges predicting pedestrian behavior due to uncertainty, sudden changes, and longer prediction time frames. AV-pedestrian negotiations demand higher precision than safety features. We suggest generative models to create multiple trajectory options or heatmaps and recommend studying human driver-pedestrian interactions to develop better strategies, making AVs safer and more reliable in complex scenarios.
Explainable AI Algorithm for Driving Decision-Making
An explainable AI algorithm helps improve driving decisions by making them easier to understand. First, the implicit visual-semantic module captures action-inducing components in specific regions, turning them into explanations people can follow. Next, the explicit reasoning module combines human-annotated explanations with intention predictions. By using this multi-task approach, the AI creates decisions that are both accurate and clear.
Pedestrian Behavior Prediction Models
Automated vehicles (AVs) work well on highways but struggle in urban areas. One big challenge is dealing with the “unpredictable” and rapidly changing actions of pedestrians and other vulnerable road users. AVs rely on cautious strategies that use short-term kinematics to avoid crashes. To improve this, we are developing deep-learning algorithms. These tools will better predict pedestrian behavior and help AVs make smarter decisions in complex urban environments.
Driver State Monitoring System
The Driver State Monitoring (DSM) system improves road safety by tracking head movements and eye glances with cameras. It detects distractions, issues warnings, and prevents accidents. Researchers tested it with over 1,000 participants in various conditions, considering demographics, facial features, and lighting scenarios like night glare and daylight shadows. The data provided valuable insights to enhance its performance.
Behavior Modeling of AV Users
Behavior modeling of autonomous vehicle (AV) users focuses on understanding how people interact with AVs. For example, researchers study the behavior of drivers, passengers, pedestrians and other road users who might encounter AVs. By examining these patterns, they aim to improve human-centric AI systems. This research helps guide the design, development and testing of intelligent transportation systems, ensuring they work safely and effectively for everyone involved.