Dynamics, Infrastructure Networks, and Mobility (DINaMo)

The Dynamics, Infrastructure Networks, and Mobility (DINaMo) group at MIT‘s Department of Aeronautics and Astronautics conducts research on topics related to the modeling, analysis, control, and optimization of modern infrastructure systems, including air traffic networks, airports,  advanced aerial mobility, aircraft emissions modeling and mitigation, control of networked systems, and congestion management in transportation systems.

Research includes:

(*) Advanced Air Mobility -- Advanced air mobility (AAM) operations (for example, autonomous air taxis and drone deliveries) are expected to significantly increase the demand for limited airspace resources. In our research, we tackle challenges — some unique to AAM, others not — posed by these emerging types of aircraft operations: previously unprecedented numbers of flights, on-demand services, and highly-competitive environments.

(*) AI-Assisted Optimization of Schedules -- Air Force flight and training scheduling is a labor-intensive and largely manual process. Complex training requirements and dependencies, operational constraints, numerous qualifications, and unforeseen missions confound the schedule development process. In this project, we develop deterministic and robust optimization formulations for the Air Force training scheduling problem. We account for the propagation of disruptions, incorporate user preferences, and identify metrics and objective functions to compare candidate schedules.

(*) Modeling and Control of Queuing Networks -- Long queues of aircraft taxiing on the surface contribute significantly to the fuel burn and emissions at airports. Motivated by the problem of airport surface congestion, we develop data-driven modeling approaches for queuing networks, and use them to build congestion management algorithms.

(*) Multi-Agent Navigation in Dynamic Environments -- We study the problem of coordinating teams of vehicles with limited sensing and communication to navigate in environments with dynamic (adversarial) and static obstacles using graph neural networks and multi-agent reinforcement learning.

(*) Network Models of Air Transportation -- Air transportation is vulnerable to a variety of disruptions (for example, weather), and its networked nature results in their impacts being felt across the system. In order to achieve a resilient and robust air transportation system, we develop tools that give us a better understanding of system behavior, and the algorithms needed for recovery post-disruption.

(*) Trajectory-Based Operations -- The concept of Trajectory-Based Operations (TBO) is a paradigm shift in air traffic management from traditional clearance-based control to trajectory-based control. It is intended to make flight operations more efficient and predictable, while maintaining operational flexibility. TBO relies on four dimensional trajectories that are managed by specifying a sequence of metering points, each of which is associated with a controlled time of arrival (CTA) that must be met by the aircraft within a specified time tolerance. In this research, we address the design aspects of TBO, such as the optimal number and location of metering points, and their impact on various system performance metrics (e.g., throughput and fuel burn).