Crowd-sourced data has tremendous value to the DOT which provides citizen reports of slowdowns and accidents. This information can typically be received significantly quicker than through other sensors, cameras or other data sources. The problem with crowd-sourced data such as Waze is that there are often duplicate reports. Operators do not need to receive these duplicate reports of information they already have so a procedure must be developed to eliminate these reports. In addition, crowd-sourced data are point sources and do not accurately display the extent of a slow down.
This project has two phases. The first phase is intended to quantify the value of Waze data to the DOT by determining the number of incidents which would not otherwise be detected by the TMC. The second phase involves creating a clean data feed of the Waze data which can be used by operators which eliminated duplicate events as well as shows the geographical extents of the reports rather than a single point.
Anticipated Project Outcomes:
Waze stream enhancement completed is provided within the table (to the right).
The first phase quantifying the benefits of Waze has been completed and is available in this presentation: https://iastate.box.com/s/yv58j3q4fnwtmme4ms172uni80mcc605
The clustering procedure to identify duplicate events has been completed and tested. Currently working on creating the data feed which processes the data in real-time and updates the feed to show the waze events. The feed will also be conflated to the LRS and will show a from and to measure to display the geographical extents of the waze event.
Cluster validation:
o Using video and sensor data to validate and fine-tune clusters
o Done by: Jan 31
Cluster reliability:
o Develop methods to evaluate reliability based on cluster validations
o Done by: Feb 15
Set up server for live Waze feed to IDOT
o Currently being tested
o Done by: Jan 31
Recurring vs non-recurring event detection
o Target: End of February