MIT Transit Lab Wins $2.1 Million to Build an Open-Source AI Assistant for Transit Control Rooms

When a train breaks down or a bus route is blocked, staff in a transit control centre have to make quick decisions while watching a wall of separate screens and data feeds. A new project at the Massachusetts Institute of Technology wants to make that job easier. MIT's Transit Lab will receive $2.1 million from Google.org, Google's philanthropic arm, to develop an open-source platform called the Public Transit Intelligence Hub, or PTIQ.

The project is one of 15 selected worldwide in the Google.org Impact Challenge: AI for Government Innovation, announced on 15 September. The funding runs for three years, and Google.org will also provide pro bono support from its own engineers and AI product experts.

PTIQ aims to bring together the real-time monitoring, operations control and passenger communication systems that transit agencies currently run separately, in one AI-orchestrated platform. Its decision-support interface will combine predictive models, optimisation engines and large language model-based contextual reasoning. The expected benefits include faster responses to disruptions, less crowding on platforms and at bus stops, and quicker, more accurate information for riders, supporting everyone from dispatchers to vehicle operators and communications teams.

The team is emphatic that the aim is not to automate transit decisions: the final call stays with transit staff. The project is led by Awad Abdelhalim, associate director of the Transit Lab and PTIQ's technical lead, and Jinhua Zhao, head of MIT's Department of Urban Studies and Planning. Program manager Jim Aloisi, a former Massachusetts secretary of transportation, directs the Transit Research Consortium, which also includes researchers from Northeastern University. The lab has previously worked with transit agencies in Washington, Chicago, London, Boston, Tokyo and Hong Kong.

The researchers are also candid about the challenges. Zhao argues that adoption depends less on the technology than on whether an organisation can work with AI and whether staff trust it, and Abdelhalim notes that AI is usually tested on problems with a single right answer, while real transit operations rarely have one. The benefits remain goals for now, but an open-source tool could let agencies of every size share what works.