Evaluating and Validating Technology Options for Estimating Transit Vehicle Occupancy in Real Time
-
2023-10-01
-
Details:
-
Creators:
-
Corporate Creators:
-
Contributors:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Publication/ Report Number:
-
Resource Type:
-
Geographical Coverage:
-
Edition:Final Report May 2022 – January 2024
-
Corporate Publisher:
-
Abstract:The primary goal of this project was to evaluate and validate various technologies for collecting transit vehicle occupancy information in real time. The specific objectives include the following: (1) Identify a list of potential technology alternatives for transit vehicle occupancy estimation by scanning the academic literature, news, and technical reports, as well as interviewing transit practitioners; (2) Evaluate all potential technologies from a technical perspective (e.g., measurement accuracy, latency, reliability, level of automation, ease of implementation and use, and maintenance needs) involving both hardware and software, and a nontechnical perspective (e.g., cost efficiency, privacy impact, and user acceptance); (3) Develop detailed documentation of promising technologies, covering technical capabilities, privacy, barriers to implementation, risks, cost, and possible vendors; (4) Conduct pilot studies and validate selected technologies in three representative transit systems in Florida; (5) Complete a technical report on selecting and implementing vehicle occupancy estimation technology; (6) Develop and deliver a webinar to disseminate the project findings. After pilot studies in three transit systems in Florida, we noted that MAC (Media Access Control) address randomization in Wi-Fi probing presented a significant challenge for estimating vehicle occupancy accurately. However, a data-driven learning algorithm, after training and testing on substantial data, can achieve a satisfactory predictive performance, such as achieving an R2 value of 0.84 in the case study of Route Evergreen in Tallahassee, FL. The predictive performance may be even higher when continuous data are available, and noise in training data is better filtered.
-
Format:
-
Collection(s):
-
Main Document Checksum:
-
Download URL:
-
File Type: