How It Works | I-24 MOTION

Infrastructure

110-135ft road-side poles and multiple high-resolution cameras

Steel road-side poles are spaced along the roadway at 550ft intervals to maintain a continuous view of vehicles across the 4mi length of I-24 MOTION. Pole height is 110-135ft to provide adequate vantage point for cameras to observe all vehicles on the roadway with minimal occlusion (e.g., large trucks blocking small cars). Each pole hosts six high-resolution (4K) cameras to cover the area around the pole.

There are three interchanges with surface streets along I-24 MOTION, all of which are critical residential and industrial links. Camera poles surrounding these interchanges each hold an additional six cameras, for a total of 26 cameras to monitor the entrance ramps, exit ramps, and adjacent surface street intersections.

Map of I-24 MOTION roadway with pole locations — Layout of camera poles and other landmarks across the I-24 MOTION testbed.

Video feeds from the 294 cameras on the system are transmitted over a fiber optic network for processing off-site at the Vanderbilt University datacenter. The full resolution data stream runs at approximately 20Gbps, requiring dedicated infrastructure for system operation. Dual GPS-based network time servers are hosted at network buildings close to the testbed for the purpose of camera time synchronization.

Map of I-24 MOTION system infrastructure, showing area of testbed poles, fiber network, and datacenter.

Data Processing

Anonymous vehicle space-time data, driven by deep learning

The nearly 300 video feeds from I-24 MOTION are processed by a custom software suite that includes state-of-the art computer vision algorithms. These techniques use deep learning (a form of artificial intelligence) to identify vehicles and determine their position on the roadway. The type of vehicle is also collected, but all vehicle data is anonymous and contains no personal information.

This processing can deliver vehicle positions at over 30 measurements per second. Cutting-edge 3D image processing discerns the footprint of each vehicle to achieve high positional accuracy.

Assembling the time-space data for all vehicles on the roadway allows us to study vehicle-to-vehicle interactions, the evolution of large-scale traffic phenomena, and full roadway efficiency. Synergies with the I-24 SMART Corridor will enable studies on active traffic managements strategies and their impacts.