If you’ve been driving around Phoenix recently, you might have experienced a double-take if the car next to you is without a driver — and on occasion, any occupants in the vehicle. The surge of artificial intelligence (AI) and machine learning (ML) technologies has allowed for the proliferation of autonomous vehicles in Phoenix — perhaps most recognizably Waymo — and across the country.
And this trend is showing no signs of slowing down. The Insurance Institute for Highway Safety anticipates that there will be 3.5 million self-driving vehicles on U.S. roads by 2025, and up to 4.5 million by 2030.
Because of this acceleration into a world with autonomous vehicles, safety and security are key priorities for municipalities and local cities. Traffic monitoring, powered by emerging technologies, will allow local governments to keep a careful eye on the community to maintain the highest standards of care for those on the road.
Yezhou ‘YZ’ Yang, associate professor at School of Computing and Augmented Intelligence (SCAI) at ASU, researches computer vision and robot perception with his team. Key to this research is understanding situational awareness, which uses sensor data to understand the surrounding environment and what’s taking place.
Automating traffic monitoring smart technology, such as sensors and cameras, can make roads safer for drivers. Yang, for example, has observed how city cameras around the Traffic Operations Center (TOC) of Arizona and Phoenix are able to effectively monitor and detect when someone is in need of assistance. However, because they are manually monitored, they lack the ability to process real-time feedback. “The ultimate goal is to make these processes autonomous,” said Yang.
“We can realize and recognize situations with cameras and automate the cameras to do the best situation monitoring — that will reduce response time,” Yang said. “In urgent situations, one second — or one-tenth of a second — shortened response time may lead to life and death.”
But how will local governments and private companies know the best technologies to monitor the roads? What are the advantages and disadvantages of each solution?
To tackle these questions, Yang is working with the Cox Connected Environments Collaboratory at ASU to obtain an unbiased evaluation of technology effectiveness and performance. Entering its fifth year, the Cox Collaboratory — located at ASU Skysong — has thrived by combining the industry leadership of Cox Communications with the research capability of ASU.
Through the collaboration between Yang and the Cox Collaboratory, this traffic monitoring and management systems’ evaluation will determine the strongest detection technology and automated traffic signal management technology to make local roads — while shared with autonomous vehicles — as safe as possible.
What solutions are being measured?
In all aspects of the evaluation, the team determines if the smart technology solutions preserve the privacy of those being monitored. While cameras monitor traffic and safety, Yang shares that it’s important to consider how to utilize solutions that provide enough metadata needed while protecting the public’s private information.
“As a researcher, we keep social good in mind when we are looking at solutions,” he said. “We are trying to reduce the opportunity for the data to be breached.”
To rank technology solutions, Yang and his team look at three dimensions:
- Hardware: This includes looking at the types of sensors (i.e., high-definition or 360-degree cameras), lidar radar and sensor fusion, or fused sensor input to create a sensory image.
- Software: The team is reviewing the capabilities of the technology — some can track and detect vehicles, some can detect acceleration and speed of vehicles, some can evaluate environmental conditions and others can identify pedestrians.
- Energy consumption: While safety is the top priority, sustainability is also measured by evaluating how energy efficient solutions are. In doing so, solar or locally powered technology can be considered. This is beneficial to know because the solutions won’t drain infrastructure power and are easier to deploy, especially in rural and underserved areas.
How is success measured?
While unbiased evaluation is always the challenge, said Yang, the research team is utilizing objective metrics — such as accuracy, energy used and estimation errors — to evaluate the hardware, software and energy consumption of smart traffic technology, like cameras and sensors. Then, the team compares vendor solutions’ outputs with data collected from previous research.
The goal? For the ASU Collaboratory to gain an impartial evaluation of technology effectiveness and performance.
What’s next for the research project
Yang and his team have completed the first round of vendor surveying and will continue into the next phase of the project, in which they will further identify vendors and acquire devices for on-site lab testing. This data collection and collaboration of validating software and testing hardware will be completed by January 2024.
Once the research is finished, most of it will be publicly accessible, available as open source and included in a white paper that shares the evaluation of in-market products for intelligent transportation system infrastructures. “Hopefully, with our efforts, we can directly apply, get feedback, refine our work and then scale our research worldwide,” said Yang. “It’s very exciting.”