Workshop on Open Science for Intelligent Vehicles

Intelligent vehicle research is benefiting greatly from open datasets, benchmarks and testbeds. Without trustworthy datasets, transparent benchmarks, and credible testbeds, the field can’t measure progress, let alone accelerate it. This half-day workshop brings together open dataset contributors, benchmark creators, testbed operators, and many open science practitioners to bridge that gap head-on. The session highlights emerging open datasets, large-scale simulation and evaluation frameworks, and next-generation physical and virtual testbeds that enable reproducible, comparable, and interpretable research in intelligent vehicles. Through a set of plenary talks and two focused panels, participants will examine what data is still missing despite industry-scale collections, what we should be measuring when we design benchmarks, and where future investments in testbeds can most effectively raise the rigor and trustworthiness of the community’s work. The workshop targets to define concrete, field-wide steps for building an open, reproducible foundation for intelligent vehicle research.

Workshop Program

Note: All speakers and presentation materials might be subject to change.

Format: Each talk within a breakout session is a 5-minute lightning talk, with the remaining time dedicated to open panel discussion.

Plenary Talks

08:30 - 08:45

Opening Remarks and Workshop Overview

Cathy Wu, Associate Professor, Chair of RERITE, MIT

08:45 - 09:00

State of the Union - Open Science in Transportation 2026

Junyi Ji, Research Engineer, Vanderbilt University

Breakout Panels

09:00 - 10:00

Panel 1: Benchmarks

Moderator: Cathy Wu, MIT

One Hour, Decades of Experience: Self-Play RL in PufferDrive
Daphne Cornelisse, PhD Student, New York University
Dynamic Traffic Simulation Package with Multi-Resolution Modelling
Xuesong (Simon) Zhou, Professor, Arizona State University & Cafer Avci, Lecturer, Cornell University
AutoTune: A Unified Benchmark for Highway Traffic Microsimulation Calibration
Cameron Hickert, MIT

When we talk about benchmarks for intelligent vehicles, what exactly are we measuring in the first place?

10:15 - 11:15

Panel 2: Datasets and Testbeds

Moderator: Junyi Ji, Vanderbilt University

Controlled Experiment on the Lane-changing of Transitional Autonomous Vehicles
Danjue Chen, Associate Professor, NC State University
CDA.AI, OPV2V, and V2XReal testbeds
Jiaqi Ma, Professor, UCLA
TeraSim-World: Worldwide Safety-Critical Data Synthesis for End-to-End Autonomous Driving
Jiawei Wang, Lecturer, UMich & Haowei Sun, CEO, SaferDrive AI

If the automated vehicle companies already have all the data, then what exactly is left for us to ask to keep the field moving?

11:30 - 12:00

Ann Arbor Connected Environment and Mcity NextGen

Henry Liu, Director of UMTRI, Mcity, and CCAT (USDOT Region 5 UTC), Bruce D. Greenshields Collegiate Professor of Civil and Environmental Engineering, University of Michigan

12:00 - 12:30

Brainstorming: Open Questions and Fundamental Problems in Transportation

Organizers & All Participants

An open-floor discussion inviting all participants to surface the open questions and fundamental, unsolved problems that should shape the next decade of intelligent vehicle and transportation research. Together we will identify priorities, gaps, and concrete community-wide steps toward an open and reproducible research foundation.

What are the open questions and fundamental problems in transportation that our community should prioritize next?