Making Cities More Safe & Efficient May Be Easier Than You’d Think.
GridMatrix’s ML Software May Be the Solution
Written by David Von Dollen
Introduction
At GridMatrix, we provide a software platform for agencies to integrate data into a cloud environment, where we run Machine Learning (ML) algorithms to automatically detect and track objects and generate metrics to provide insights at the intersection related to Safety, Congestion, Signal Performance, and Emissions. Sensor data may include streams from cameras, LiDAR, inductive loop, or connected vehicle data. The platform also allows for automated alerts to be sent to users when certain conditions are met.
In this article we will investigate questions related to the platform such as:
- How does the automation of data collection at the intersection benefit the traffic engineering domain, and other agencies?
- What kind of actionable insights can our customers glean from the platform?
- How can alerting be set up to help customers detect events in real time? What are some caveats with such alerts?
Pain Points
GridMatrix is currently partnering with the agencies to understand how the platform may be used to enable traffic engineers, first responders, and transportation planners with actionable insights. In conversations with customers, some common scenarios and pain points that arise are:
- No detection system in place for monitoring pedestrians in restricted areas on highways, on/off ramps and approaches
- No automated system in place to monitor traffic congestion on approaches to bridges or tunnels
- No automated system in place for monitoring safety for mixed-modality traffic at key merges at approaches
Developing solutions towards these scenarios can have downstream impacts. In 2020 over 55,000 pedestrian injuries occurred on roadways in the United States [1]. Additionally in 2020, the transportation industry accounted for 27% of CO2 emissions in the United States [2].
The GridMatrix Platform
The GridMatrix platform streams data sources in real-time, presenting metrics to customers related to congestion, signal performance, emissions and safety. The platform features a variety of methods for visualizing this data, both live and historical. But how is this helpful for customers?
Consider a four-way intersection that is being monitored by a camera. Camera data is streamed into the cloud, over which the platform continuously runs machine learning algorithms for object detection and tracking. Output data from the machine learning algorithms in the form of metrics are then presented in global views to allow for quick actionable insights. It’s also important to note that no personally identifiable data is stored during this process according to GridMatrix’s privacy policies.
To give an example, let’s say an engineer working on emissions reduction initiatives wants to find intersections with emission levels above the average intersection. The platform allows for a user to quickly scan a heat map of locations to find points where emissions may be higher than average. Upon further investigation, the user may be able to drill down into an intersection view on the platform and determine the source of the increased emissions to be vehicle congestion at a given approach. Other metrics such as platoon ratio and average arrival on green are provided by the platform to aid further analysis. The engineer may then alter the signal timing to adjust given this analysis.
Let’s take another example. Let’s say a first responder wants to be able to set up alerts when certain conditions are met or anomalies are detected, such as a pedestrian walking onto a highway offramp, or identify a disabled vehicle at an intersection. The platform can allow for alerts to be set so that the first responder is notified when these events occur and are detected by the ML algorithms on the platform.
In addition to these streams the platform allows for integration with other data streams, such as weather or connected fleet data. These may be further analyzed and monitored to create alerts for anomalous events and monitor the overall health of vehicle fleets.
Summary
Enabling traffic engineers, first responders, and agencies with artificial intelligence tools for automating surveys and detecting anomalies presents opportunities for increasing efficiency and safety. There are also opportunities beyond the traffic intersection. Agencies can use the platform for planning, optimization or other operations at ports, airports and facilities as well.
For more information, please visit www.gridmatrix.com.
References
CrashStats — NHTSA — DOT. Accessed February 15, 2023. https://crashstats.nhtsa.dot.gov.
“Sources of Greenhouse Gas Emissions | US EPA.” Environmental Protection Agency, 5 August 2022, https://www.epa.gov/ghgemissions/sources-greenhouse-gas-emissions. Accessed 15 February 2023
