A Guide to Using AI to Improve Transportation
A new briefing from the Eno Center walks practitioners through AI processes and showcases practical applications.
By Angie Schmitt
Artificial Intelligence has the potential to change the way we work and profoundly disrupt many industries. A recent policy brief by the Eno Center for Transportation explores how it could change the way we use and deliver transportation. Most famously, AI could upend the way we drive. AI companies operating in the vehicle automation space have received $95 billion in venture capital, says report author Renee Autumn Ray. While fully autonomous cars remain far from wide circulation, artificial intelligence is already changing driving. By contributing to the development of advanced safety features, AI has potential benefits for vehicle safety.
AI also has practical applications for those who build and operate our transportation infrastructure. For example, machine learning can be used to forecast demand for micro-transit systems, Ray says. In addition, a category of AI called “supervised learning” can be used to analyze weather, road conditions, and demographic info to help predict where crashes may occur. A complicated machine learning algorithm called a “convolutional neural network” can be used, for example, to help identify illegally parked vehicles and aid in enforcement.
One of the more exciting frontiers for AI in transportation is its use in asset management. AI can be used to gather data. For example, an AI method called “computer vision” processes and analyzes data from video as well as still images. State DOTs, like Texas, are already using LiDAR technology and AI to inventory assets and gather data about their conditions. As a result, they’ve seen significant efficiency gains.
Not only can AI help agencies gather data, but it can also be helpful in data analysis, says Ray. U.S. transportation agencies have lots of information. However, it is often unorganized and inaccessible for analysis. To remedy this, agencies can use the AI method known as “deep learning” to extract specific information from huge data stores.