If you would like to be notified of future Safe-D webinars, please fill out this form.
| Title/Project/Date | Speaker | Webinar Overview |
|---|---|---|
| Title: Evaluation of a Narrow Automated Vehicle-Exclusive Reversible Lane on an Existing Smart Freeway
Project: Safe-D 04-101 Date: September 29, 2021, 14:00 EST Zoom Link: Register Here |
|
AVs are dependent on several sensors to recognize the surrounding environment and navigate the roadway. The operational features and logic of AVs are different from human-driven vehicles where operational decisions are made based on driver capabilities and behavioral characteristics. AVs’ lane-keeping capabilities could allow for infrastructure standard adjustments, such as narrower lanes, fewer lanes, and smaller and less signage, which could result in more efficient mobility. A full infrastructure adaptation to AVs will not take place quickly, especially given that the transportation system will be serving both AVs and human-driven vehicles for quite some time. Therefore, a mix of dedicated AV lanes and normal vehicle lanes seems to be a viable solution. This webinar shares results of recent research that evaluates implications of an innovative infrastructure solution, exclusive AV lanes, for safe and efficient integration of AVs into an existing transportation system. Examining a real-world case study, this project investigates implications of adding a narrow reversible AV exclusive lane to the existing configuration of the I-15 expressway in San Diego, resulting in a 9-foot AV reversible lane, and in both directions of travel, two 12-feet lanes for HOV and HOT vehicles. A series of tasks were completed, including a literature review, an AV manufacturers product review, expert interviews, a consumer questionnaire review, a crash data analysis, and a traffic simulation analysis. This webinar details these tasks with the main focus being on the traffic simulation part of the study. |
| Title: Reference Machine Vision for ADAS Functions
Project: Safe-D 04-115 Date: October 15, 2021, 14:00 EST Zoom Link: Register Here |
|
The objective of this project is to develop a reference system for evaluating different lane markings and perception algorithms. This project validates the effectiveness of different types of lane markings for detectability on state-of-the-art lane detection (LD) algorithms. An in-depth study into the different parameters affecting the performance of LD algorithms was conducted by incorporating pavement marking material characteristics into the evaluation framework. The effect of environmental factors (Day vs Night), driving direction, lane marking material characteristics (reflective properties like Qd/RL, marking quality), lane making layouts (30ft gap vs 40ft gap, 4inch wide vs 6 inches wide), and LD evaluation characteristics (Type of LD algorithm, Near Field-of-view (FOV) vs Far FOV) were studied. Observations were made on how these different factors interact with each other and affect LD performance. Three different annotated image datasets were also generated which includes the (1) College Station Dataset (On-road with Material data), (2) 3M panel dataset (Closed course with material data), and (3) US290 Dataset (On-road special type of markings without material data). These datasets can be used as a reference/benchmark system by researchers to evaluate their LD algorithms and infer how their performance relates to different types of lane markings and their material characteristics. In this presentation, we will present an overview of the project, the methods used and the results obtained in this project. |
| Title: Data Fusion for Non-Motorized Safety Analysis
Project: Safe-D 03-049 Date: November 18, 2021, 14:00 EST Zoom Link: Register Here |
|
Nonmotorized activity, despite sharing a low percentage of total trips in the United States, contributes to a disproportionate share of total fatal and serious injury crashes. The exigency of the issue has alarmed safety advocates in multiple areas, resulting in their persistent efforts to explore and develop evidence-based, data-driven strategies to reduce nonmotorized crashes. However, such efforts are often stonewalled due to a lack of robust and reliable exposure measures. This webinar presents a SAFE-D research study, which explored an emerging research territory, a fusion of nonmotorized traffic data for estimating reliable and robust exposure measures. The research was divided into three sequential stages. The first stage involved developing and applying a guideline to process and homogenize available data sources to estimate annual average daily bike volume at intersections. The research team selected the City of Austin as the study area, gathered five bike data sources (using both traditional and crowdsourced data sources) and developed bike demand models to be used as inputs into the fusion mechanisms developed as part of the study. The second stage was focused on developing and applying the fusion framework—demonstrating the efficacy of multiple fusion algorithms, including two novel mechanisms, suited to the data characteristics and based on the availability of actual counts. The analysis of actual and simulated data illustrated that the fusion methods outperformed the individual estimates in most cases. In the third stage, the fused data were applied in both macro (hot-spot analysis in block group level) and micro (individual safety-related perception) models in Austin to ascertain the significance of incorporating exposure in safety analysis. While the fusion framework contributes to the research in the field of decision fusion, the demand and crash models provide insights to help stakeholders formulate policies to encourage bike activity and reduce crashes. |
