Dr. John Handley and Dr. Lei Lin, of the Rochester Data Science Consortium, will be traveling to Buffalo, NY to speak at the 2019 TransInfo Symposium.
The Symposium – taking place at the The State University of New York at Buffalo on Thursday, August 8th, 2019 – will gather transportation and big data professionals from academia, industry and government together to discuss the latest in the field of Transportation Informatics. The focus of this year’s symposium is on Research at the Intersection of Big Data and Connected & Automated Transportation Systems.
The conference will feature informative lectures, networking, and vehicle demos from the University of Buffalo, plus a keynote address from Alison Pascale, Senior Policy Strategist at Audi of America.
Dr. John Handley will be presenting at the 2019 TransInfo Symposium. More information about his talk is below:
A case study in the impact of public transit network redesign on accessibility: Introduction and application of a novel method of assessment
Abstract: In November 2014, Regional Transit Service, the public transit authority of Rochester, New York, introduced a new hub-and-spoke network design. We analyzed the impact of this redesign using a novel method that takes into consideration the actual operational impact on accessibility. One way to measure the service value of a transportation system is through the area that can be accessed in a given amount of time. The greater the reachable area per time interval, the more connected the system is and the greater value that is sold to a passenger. Historically, accessibility is based on published schedules. But there is a difference between what is planned and what actually happens. Owing to on-time performance issues, the actual accessibility is often less than what is planned. In fact, that difference can be greater during peak travel times and can vary spatially. In our study of Rochester’s network change, we introduced a new spatio-temporal methodology that measured the spatial temporal changes in accessibility and we were able to identify times and areas where service improved and where it was degraded.
Dr. Lei Lin will also be presenting at the 2019 TransInfo Symposium. More information about his talk is below:
Vehicle Trajectory Prediction Using LSTM with Hierarchical Attention Mechanism
Abstract: The ability to accurately predict vehicle trajectories can benefit many Intelligent Transportation System (ITS) applications such as traffic simulation, advanced driver assistance system, and the enhancement of navigation and safety of various types of road users. This ability is pronounced with the emergence of autonomous vehicles, as they require a thorough understanding of the environment in order to be safe and efficient. Recent studies have adopted deep learning for vehicle trajectory prediction. Although many advancements have been achieved, one prominent issue is that these models often lack explainability. We propose a long short-term memory model with hierarchical attention mechanism (HA-LSTM) to alleviate this issue. The results using the NGSIM datasets show that our approach not only outperforms other state-of-the-art models in terms of trajectory prediction accuracy, but also identifies the impacts, reflected by attention weights, of recent trajectories and neighboring vehicles on the movements of the target vehicle. We further provide in-depth analysis of various attention weights found using the hierarchical attention mechanism according to multiple vehicle and environment factors including types of vehicles (e.g., auto vehicle and truck), locations of the target vehicle (e.g., innermost, middle, outermost lanes and ramp), and assorted traffic conditions (e.g., mild and congested). Lastly, we have conducted a fine-grained analysis of the attention weights associated with specific driving behaviors of the target vehicle and found that the learned attention weights can be adopted to explain its lane-changing behaviors.
Transportation and big data professionals from academia, industry and government gather annually for the Symposium on Transportation Informatics, hosted by Transportation Informatics Tier I University Transportation Center (TransInfo) at its lead institution, the University at Buffalo. The focus of this year’s conference is on Research at the Intersection of Big Data and Connected & Automated Transportation Systems.
You can find more information about this year’s 2019 TransInfo Symposium here.