multi-agent

Distributed Reinforcement Learning for Optimal Speed Limit Control Over Network, sponsor: Safer-Sim

The goal is to optimize variable speed limit control (VSLC) strategies over network to improve both traffic safety and mobility. The proposed research will advance the current knowledge and practice of VSLC in three aspects. First, this research will enlarge the scope of VSLC from link­based to network­ based control to bring a new understanding about its system­level safety implications. Second, it will optimize the impact of VSLC using multi­objective learning approaches considering both safety and mobility. Third, distributed artificial intelligence approaches proposed in this research introduce new opportunities for network control effectiveness and scalability compared with traditional model/rule based approaches.

Optimizing Information Value in Heterogeneous Multi-agent Transportation Systems (OPTIMA), sponsor: NSF, Safer-Sim(seed)

This project addresses the use of advanced sensing, communications, and computing technologies in studying value of information in transportation systems made up of heterogeneous traffic (cars, autonomous and connected vehicles, buses, bicycles, etc.) The wealth of data available on these systems enables new approaches to information provisioning that have the potential to improve transportation system efficiency, reliability, and resilience. The project develops a unified modeling framework for information provision considering heterogeneous non-cooperative stakeholders and addresses three critical questions: (1) How do we model adaptive behavior of different traffic components in response to evolving information updates? (2) What are the limits of positive and negative information on systems efficiency, reliability, and resilience? (3) Given limited resources, what, when, and where should the information be communicated with which groups of stakeholders to optimize system performance?