The promise of smart intersection technology goes beyond increased operational efficiency and encompasses its potential to improve safety for all road users, including those using the crosswalk. But it does not fulfill this promise if it is only used to reduce congestion and travel times for autos. Smart intersections depend on smart policy to realize the full range of benefits they offer. Traditionally, level of service for autos has been the prime motivator for intersection design. But the tide is turning toward designing with the needs of all users in mind. In addition, smart intersection technology can collect data about the behavior of drivers, pedestrians, and bicyclists to inform design.
By incorporating smart technology into five intersections along Larned Street in downtown Detroit, the city has 7 billion auto and 770 million bicycle and pedestrian data points, collected by Miovision over 10 years. Miovision is using this data innovatively to train its artificial intelligence (AI) systems to automatically identify, classify, and continuously count and record the positions of motor vehicles, bicycles, and pedestrians.
This installation creates a permanent, passive, automated data collector that can distinguish between autos, trucks, motorcycles, bicyclists, and pedestrians. The data can be used to analyze flow direction, volume, various congestion metrics, travel time, red light running, crosswalk blocking, and even pedestrian delay and behavior. A video record can validate the AI, and be used to analyze near-miss situations and other dangerous interactions that would be missed in traditional crash analyses.
Design changes can be implemented based on the needs shown by the analytics to reduce conflict between modes, reduce speeding and other dangerous driver behaviors, reduce spill-back that moves pedestrians out of crosswalks, and improve the overall level of service provided to bicyclists and pedestrians. Within days of any changes, smart technology analytics can provide evidence of the results and effectiveness.
The system in Detroit can also add upgrades without requiring multiple systems and new hardware, making it both flexible and less expensive to add functionality. In addition to giving planners and engineers ready access to intersection analytics, remote diagnostics, and alerts, this system is also being used to give emergency and freight vehicles signal priority.
Smart intersection technology represents a shift in the way that public works agencies are notified about the condition of transportation infrastructure. No longer dependent on notification by concerned citizens or first responders, agencies can be alerted automatically by the technology. IT developers can leverage the technology’s open data to develop apps to better inform the public about infrastructure conditions.
While the technology can currently provide a host of metrics and detection algorithms, the promise of improved safety in real time lies in the adoption of this technology by transportation agencies, as well as the proliferation of connected vehicles. Widespread use of dynamic traffic control to extend green lights for pedestrians and cyclists in need, for example, is perhaps a ways off, as are in-vehicle pedestrian or cyclist warning systems linked to intersection tech. But these smart systems are capable of providing the necessary data.
Michael Brenneis is an Associate Researcher at SSTI.