Monday, April 29, 2024, 5pm

As traffic demand continues to increase globally, improving the efficiency and safety of the interconnected network of transportation systems around the world has become an increasingly critical priority. Various AI technologies have been designed to this end, and many have achieved good performance in simulation-based evaluations. However, a disconnect between theory and practice exists: few state-of-the-art AI technologies have been deployed to actually help resolve these challenges in the real world. One important cause of this disconnect is that these AI technologies have made unrealistic simplifying assumptions, which have made them unable to address the pain points of human stakeholders. In this thesis, I propose to answer research questions related to how AI technologies can be better designed for deployment by addressing four common challenges: uncertainty in underlying and observed levels of demand; coordination between individuals and systems; interpretability and controllability for complicated decision-making algorithms; and heterogeneity among end-users and deployment contexts. My completed, in-progress, and proposed work tackles these challenges through the lens of two key problem domains, traffic signal control and gig driving. Ultimately, the goal of this thesis is to design AI systems which are capable of being physically deployed and creating tangible impacts in these domains.

Thesis Commitee
Fei Fang (Co-chair)
Norman Sadeh (Co-chair)
Sean Qian
Matteo Pozzi
Peter Stone (The University of Texas at Austin)

Additional Information

In Person and Zoom Participation.  See announcement.

Event Type: Thesis Proposals
Room Number: In Person and Virtual - ET
Building: Traffic21 Classroom, Gates Hillman 6501 and Zoom
Speaker's Name: REX CHEN
Speaker Website: lythronaxargestes.github.io
Speaker's Professional Title: Ph.D. Student, Ph.D. Program in Societal Computing, Software and Societal Systems Department, Carnegie Mellon University
Talk Title: Rethinking the Design of Coordinated, Human-Centric AI Systems for Deployment in Transportation
Event Poster Title: Poster
Event Poster URL: www.cs.cmu.edu…
For More Information: aroudebu@andrew.cmu.edu
Affiliations: Software and Societal Systems Department (S3D)
Organization(s): School of Computer Science