Search for: All records

Microscopic simulation-based approaches are extensively used for determining good signal timing plans on traffic intersections. Measures of Effectiveness (MOEs) such as wait time, throughput, fuel consumption, emission, and delays can be derived for variable signal timing parameters, traffic flow patterns, etc. However, these techniques are computationally intensive, especially when the number of signal timing scenarios to be simulated are large. In this paper, we propose InterTwin, a Deep Neural Network architecture based on Spatial Graph Convolution and Encoder-Decoder Recurrent networks that can predict the MOEs efficiently and accurately for a wide variety of signal timing and traffic patterns. Our methods can generate probability distributions of MOEs and are not limited to mean and standard deviation. Additionally, GPU implementations using InterTwin can derive MOEs, at least four to five orders of magnitude faster than microscopic simulations on a conventional 32 core CPU machine. 

As a part of road safety initiatives, surrogate road safety approaches have gained popularity due to the rapid advancement of video collection and processing technologies. This paper presents an end-to-end software pipeline for processing traffic videos and running a safety analysis based on surrogate safety measures. We developed algorithms and software to determine trajectory movement and phases that, when combined with signal timing data, enable us to perform accurate event detection and categorization in terms of the type of conflict for both pedestrian-vehicle and vehicle-vehicle interactions. Using this information, we introduce a new surrogate safety measure, “severe event,” which is quantified by multiple existing metrics such as time-to-collision (TTC) and post-encroachment time (PET) as recorded in the event, deceleration, and speed. We present an efficient multistage event filtering approach followed by a multi-attribute decision tree algorithm that prunes the extensive set of conflicting interactions to a robust set of severe events. The above pipeline was used to process traffic videos from several intersections in multiple cities to measure and compare pedestrian and vehicle safety. Detailed experimental results are presented to demonstrate the effectiveness of this pipeline. 

Older citizens experience a large number of falls and hospitalizations per year throughout the world. These intervening health events (IHEs) such as falls/injuries, illnesses, hospitalizations, are strong precipitants of disability in older adults. They are episodic in nature, extremely difficult to study, and require continuous and long-term monitoring. Wearable technologies with remote capabilities are an ideal solution for capturing information before and after such events. This work presents the ROAMM campaign platform for harnessing sensor and interface capabilities on smart wearables to provide customizable, affordable, and versatile health monitoring that leads to practical remote-based interventions. The platform is flexible, efficient, and scalable for concurrently running multiple studies, each of which consists of patient-reported outcomes, ecological momentary assessments and mental health-related patient responses. Additionally, the system is able to capture and derive ecological, momentary assessments of pain with concurrent mobility tracking that allows life-space mobility ascertainment. The platform supports multiple watches, and we show implementations on both the Samsung Galaxy and Apple series of smartwatches.