When the Cox Connected Environments Collaboratory collaborates with local partners to introduce smart technologies into the community, copious amounts of testing and data collection must be done to ensure which technology and products are the best solutions.
One current example of this is the Collaboratory’s smart traffic research project, in which the Collaboratory team is collaborating with Yezhou ‘YZ’ Yang, associate professor at School of Computing and Augmented Intelligence (SCAI) at Arizona State University, and his team. Together, they are comparing technologies to best monitor safety for vehicles and pedestrians in populated areas.
The purpose of the current smart traffic research project — which is part of a larger research effort from Yang and the team — is to collect raw video data from devices that can monitor the amount of vehicles and pedestrians on the roads and intersections with precision and accuracy. The data will be readily available to share with municipalities and others at scale to help monitor roads to ensure safety for pedestrians and drivers.
A look into recent data collection project in Tempe
In late April, the Collaboratory (including their spring 2023 interns) and ASU research teams were able to start testing different technologies and devices and gathering data that they are now sorting and labeling. While the goal of testing was not to compare vendors specifically, but more the technology and devices being used.
The team began with 15 devices for tests, and during this event, tested two devices. The two types of software — the predictive capability of the device — that were tested include:
- Computer-vision cameras (which syncs data in the cloud)
- Event-based cameras (which tracks motion)
During testing, the team was able to detect cars, buses, pedestrians, trucks and bikes. They also plan to test radar solutions in addition to the two software types above.
While the metrics which they’ll use to evaluate the data haven’t been fully defined, areas to consider will include:
- Accuracy of the technology
- Predictive capabilities of the software and hardware
- Limitations within the hardware
- Heat affects
- Power dissipation
Setting up on a rooftop in downtown Tempe allowed the research team to collect the quantity of data needed thanks to the high-volume of traffic for both vehicles and pedestrians in the area. Testing day was a hot one in Tempe, so the team had to get creative so that their set-up wouldn’t be harmed by the heat.
Blake Harrison, a graduate student researcher on professor Yang’s team (studying Computer Science at the Ira A. Fulton School of Engineering) and leader of the data collection, was pleased with the raw video data that they were able to gather during the testing event, sourcing a variety of sights and sounds that will prove to be helpful in evaluation. It’s important to note that the data collected was anonymous — the software blurs out faces or captures figures versus detailed images.
Next steps for the traffic research project
The end goal is to package the raw data to share which devices shared the most accurate data. The research team will continue to gather devices to monitor and test to make sure they are collecting data on a wide variety of products over the next few weeks. Then, they will develop a white paper sharing their findings that will be accessible to all this summer.
“It’s been great working with the Collaboratory,” said Harrison. “They’ve been with us every step of the way, offering help and assistance. It’s been really good.”