Building a human-in-the-loop simulation of market-based traffic flow management

Abstract

Human-in-the-loop (HITL) simulations can be invaluable in the study of traffic flow management concepts. However, the design and execution of HITL experiments is often very challenging and risk-prone. In our market-based Traffic Flow Management (TFM) research, we develop an approach to developing HITL simulations that leads to effective experiments that yield great results, while reducing programmatic risk. The software used for the HITL experiment is developed using the Probabilistic NAS Platform (PNP). The architecture of PNP allows software clients, representing various National Airspace System (NAS) elements, to interact with real-world NAS data provided by the PNP server. The architecture of PNP allows clients to be distributed across a network, such as the Internet. This flexibility allows HITL study participants to be located anywhere in the world, provided Internet access is available. This was a requirement for our simulation, in which participants would be participating from various locations. In addition to software risk, we also tackle user acceptance challenges. HITL experiment results can be detrimentally skewed by participants who are not fully engaged in the simulation. In this study, we incorporate a feedback solicitation process into our project planning mechanism. The development of HITL simulations, while challenging and risk-prone, can provide insights that cannot be derived from automated, fast-time simulations. The approach discussed here can lead to the rapid development of robust software that engages users and collection of useful results.