Reasoning and Mitigating AI Biases in Connected Vehicle Infrastructure-Pedestrian Systems for Promoting Equitable Pedestrian Safety at Intersections

Pedestrian safety remains a critical challenge in current transportation systems. In the United States(US), fatal pedestrian crashes have increased nearly 50% over the past decade. Data shows that children, older adults, men, people with low income, people experiencing homelessness, and people of color are involved in far greater pedestrian-vehicle crashes compared to the general population. Automated Vehicles (AVs) are expected to effectively detect pedestrians and react to potential crashes. However, due to the constrained mobilities of vulnerable road users, data from vulnerable pedestrians, such as older adults and children, is often limited. This project seeks to uncover, characterize, and mitigate the biases of artificial intelligence (AI) models in connected vehicle-infrastructure-pedestrian systems by leveraging advanced statistical machine learning, representation learning techniques, and real-world video data. The discovery results of the biases of AI models and proposed bias mitigation solutions in connected vehicle-infrastructure-pedestrian systems will help promote the transformation design of future intelligent transportation systems, ensuring the travel safety and justice of vulnerable road users, children, and older adults under such transformations. The project will benefit the design of future management policy and standards by providing a novel and systematic AI bias evaluation framework for general connected vehicle-infrastructure pedestrian systems.