Research
Research Overview
Our research interests lie in the area of Human-Technology Interaction and Performance Modeling in Cyber Transportation Systems (CTS). With the continued advances in CTS, the communication between vehicles and the increased automation capabilities aim to support a new generation of safety applications and provide information to drivers in a reliable and timely manner. Our current research focuses on the interaction between humans and CTS by proposing a theoretical framework that combines automation and communication technologies, in-vehicle information systems (cooperative collision warning systems), and human drivers. We adopt experimental and modeling approaches to investigate and quantify the effects of system characteristics on human behavior and performance in CTS. This framework aims to (1) provide insights into human cognition mechanisms regarding how humans perceive, process, learn and respond to information; and (2) provide a way to optimize and evaluate the design of CTS by addressing human needs.
Ongoing Research Projects
1. Driver Risk Perception, Trust, and Behavior in Human-Automated Vehicle Interaction in Mixed Traffic
Funding source: NSF: CRII: CHS: Addressing Individual Differences in Human-Autonomous Vehicle Interaction in Mixed Transportation Systems, Press Release
With the increasing deployment of automated vehicle (AV) technologies, we will experience a long transition period of mixed-autonomy traffic in which human-driven vehicles (HVs) will share the road with AVs, at least for the next few decades. In mixed traffic, there are two forms of human-AV interaction: drivers directly interact with AVs within the vehicle (Driver-AV interaction) and drivers of human-driven vehicles (HV) interact with AVs on the road (HV-AV interaction). This project has the following objectives:
Driver-AV interaction: Understand and quantify how the design of automated driving styles and individual differences among drivers impact their level of trust and acceptance of AV technology.
HV-AV interaction: Elucidate and model how drivers perceive risks and adapt their behavior in mixed traffic environments with varying AV ratios/penetration rates.
Human-Machine Interface Design:
Mitigate aggressive driving behaviors and enhance driver trust through thoughtful designs of HMIs and automated driving styles.
Utilizing AVs for Behavior Alteration: Explore the potential of AVs in shaping the behavior of human-driven vehicles sharing the road with them.
2. Driver Situation Awareness and Risk-Taking Behavior in Connected and Automated Vehicle Systems
With the assistance of connected and automated vehicles (CAVs), drivers might depend on these technologies for hazard detection and some driving tasks, potentially leading to reduced situational awareness (SA). Ensuring the safe use of CAVs requires designing interfaces that help drivers quickly and effectively regain SA. This research focuses on quantifying the relationship between driver SA and performance, exploring the impact of different warning interface designs on driver SA in connected vehicles and conditionally automated vehicles, and utilizing HMI designs to enhance driver SA in CAVs.
Furthermore, our research adopts a dual-modeling approach, incorporating both a cognitive modeling framework (Queuing Network-Model Human Processor) and machine learning techniques. This allows us to forecast driver situation awareness through the analysis of eye movement data while also quantifying the influence of HMI interface design on driver SA, decision-making, and driving/takeover performance.
Funding source: PSU/Technion Marcus Funds
3. Modeling of Human Cognition and Performance in Intelligent Systems
To understand the psychological processes involved in driver information processing and assess driver performance, our third research avenue focuses on developing cognitive driver performance models in CAVs. This research employs the Queueing Network-Model Human Processor (QN-MHP) cognitive framework and mathematical modeling techniques to establish a computational framework for driver-vehicle interaction. It aims to uncover the cognitive processes involved in how drivers process warnings and information from CAVs and predict the impact of HMI designs on driver performance. This research contributes to the computational cognitive modeling field by advancing our comprehension of driver cognitive processes in driver-vehicle interactions and facilitating the design of user-centered interfaces for CAV systems.
Mathematical modeling of human performance in CTS
4. Enhancing Situation Awareness of Adversary ML in Human-AI Collaboration for Safe Implementation of Automated Driving Systems
AI systems integrated into automated driving systems (ADS) are susceptible to adversarial attacks, thereby compromising the safety and security of ADS and posing significant risks on the road, potentially leading to accidents and fatalities. Drivers often overestimate the AI's capability to detect objects, resulting in a lack of awareness that can be hazardous, particularly in SAE Level 3 ADS, where drivers are expected to intervene in automation controls when necessary.
This collaborative project with Dr. Aiping Xiong from IST is to develop a comprehensive framework for enhancing collaboration between drivers and AI systems in response to adversarial attacks, with a specific focus on improving driver situation awareness. We will assess the influence of system capabilities and reliability on various facets, including driver situation awareness, trust in AI systems, acceptance of AI technology, and driver performance when faced with diverse adversarial attack scenarios.
Funding source: Center for Socially Responsible AI-Big Ideas Grants, Press Release
5. Impact of CAV Technologies on Driver Behavior
Funding source: PSU: Multidisciplinary Research Seed Grant, Press Release
This project focuses on the predictive understanding of the impacts of connected and autonomous vehicle (CAV) technology on changes in driver behaviors during commute travel. We aim to develop driving simulators to enable live interactions with traffic microsimulations of community-level traffic flows. This project seeks to combine the best-practice tools in transportation engineering, human factors, vehicle dynamic simulation, and urban land-use choices to produce a comprehensive understanding of human-vehicle interaction in CAVs.