Of the many ways artificial intelligence (AI) and machine learning are poised to improve modern life, the promise of impacting mass transit is significant. The world is much different compared with the early days of the pandemic, and people around the world are again leveraging mobility and transit systems for work, leisure and more.
Across the U.S., traditional mass transit systems including buses, subways and personal vehicles have returned to struggling through gridlock, rider levels and congestion. However, advanced AI and machine learning solutions built on cloud-based platforms are being deployed to reduce these frustrations.
Exciting Transportation Opportunities with AI
Transportation is one of the most important areas where modern AI provides a significant advantage over conventional algorithms used in traditional transit system technology.
AI promises to streamline traffic flow and reduce congestion for many of today’s busiest roadways and thoroughfares. Smart traffic light systems and the cloud technology platforms they operate on are now designed to manage and predict traffic more efficiently, which can save a lot of money and create more efficiencies not only for the cities themselves, but also for individuals. AI and machine learning can process highly complex data and traffic trends and suggest optimum routing for drivers in real-time based on specific traffic conditions.
As a result of drastically improved processing power, transit system technologies are now used in various IoT (Internet of Things) devices to achieve real-time image recognition and prediction that took place in legacy data centers during the last half century. This new decentralized-focused architecture helps increase the implementation of machine learning and AI. Today’s recognition algorithms offer enhanced insight on the mix of density, traffic and overall rate of flow. Furthermore, these optimized algorithms can leverage data points by region, resulting in a streamlined pattern to reduce traffic problems while redistributing flow more optimally. Municipal transit systems can then make better decision-making power, and the control system has a much higher degree of failure tolerance as was previously demonstrated in legacy hub-and-spoke systems.
AI’s Current Impact
These technologies are already being deployed around the country. As one example, the Santa Clara Valley Transportation Authority (VTA) in partnership with the City of San José, California has been piloting a cloud-based, AI-powered transit signal priority (TSP) system that utilizes pre-existing bus fleet tracking sensors and city communication networks to dynamically adjust the phase and timing of traffic signals to provide sufficient green clearance time to buses while minimally impacting cross traffic.
Because the new platform leverages pre-existing infrastructure, it required no additional hardware installations inside traffic signal cabinets or buses. And unlike traditional, location-based check-in and check-out TSP solutions, the platform processes live bus location information through machine learning models and makes priority calls based on estimated times of arrival. The platform has so far improved travel times on VTA’s Route 77 by 18% to 20% overall, equating to a five- to six-minute reduction in signal delay.
The cloud-based transit signal priority system combines asset management and automation to produce a system capable of providing services to an entire region. Unlike hardware-based systems, this platform uses pre-existing equipment and leverages cloud technology to facilitate operations. This removes the need for vehicle detection hardware at the intersection because vehicle location is known through the CAD/AVL system. This enables both priority calls from greater distances away from signals and priority calls coordinated among a group of signals. Furthermore, the system provides real-time insights on which buses are currently receiving priority along with daily reports of performance metrics.
The advanced transit signal priority systems available today consist of two parts: a unit in the traffic cabinet and another unit placed on the vehicle. The transit priority logic is the same, regardless of the detection and communication medium. When a vehicle is within predetermined boundaries, the system places a request to the signal controller for prioritization. Since the original systems used fixed detection points, signal controllers were configured with static estimated travel times. Since travel times are dependent on several environmental factors, the industry implemented GPS-based, wireless communication systems. With this method, vehicles found within detection zones replace the static detection points and the vehicle’s speed is used to determine arrival time.
The platform allows cities to build upon current investments in infrastructure to deploy citywide TSP. To enable safe and secure connections with traffic signals, each city requires just one device for use that is a computer that resides at the “edge” and serves as the protective link between city traffic signals and the platform. It is designed to securely manage the information exchange between traffic lights and the cloud platform. It is the only additional hardware necessary, and depending on the existing city network configuration, the platform may receive vehicular data directly or via the city’s network using secure connections.
Sophisticated Process to Prioritize Traffic
The system’s method of placing priority calls to traffic signals is more sophisticated and is not constrained to fixed-point locations. Unlike the current state-of-the-art of placing priority calls from the detection of buses at specific locations that starts a pre-programmed time of arrival, this platform uses a “vectorized” approach. In mathematics, a vector is an arrow representing a magnitude and a direction. In this platform’s software, the arrow points in the direction of the traffic light and the magnitude is the travel time. When the system is set up, traffic signals, bus routes and bus stops all get a digital representation on this vector. This ends up producing a digital geospatial map where software is then able to track bus progression along bus routes. This results in a system that can dynamically place transit calls regardless of its location. Instead, the system makes precise priority calls based on the expected time of arrival which is the basis for all TSP check-in calls supported by all signal controller vendors. And due to the nature of the tracking algorithm, any significant changes to ETA can be adjusted. For example, if a bus was predicted to skip a bus stop but didn’t, the system will detect the change and adjust the priority call accordingly.
The combination of AI, machine learning and cloud-based technology all have great potential to not only improve the current mass transit system but reimagine it altogether. This advanced technology is already proving how it can improve coordination between GPS, navigational apps, connected autos and even taxi and ride-sharing services to efficiently combine into a single transit entity based on real-time data.
In the not-too-distant future, it is expected that connected self-driving cars and trucks will be more prevalent on the roads and highways, offering even greater potential for AI to reduce both the duration and risk of rapid mobility.