AI-Enabled Traffic Control, where are you?
It seems like the effects of climate change are appearing everywhere. Record levels of moisture, drought, and temperature extremes are being recorded nearly every day. Even skeptics are beginning to accept that our climate is changing in detrimental ways. The only issue remaining is if it is a result of a natural cycle or due to human contamination of our atmosphere. Intelligent traffic control and monitoring is one way to lessen our carbon impact.
It’s time to implement intelligent traffic control systems. Billions of dollars are being spent in efforts to reduce the carbon dioxide humans and our warming planet inject into the atmosphere every day. Incandescent lighting is being replaced with LEDs, coal fired electrical generation plants are being decommissioned, and forests are being replanted to reduce our human carbon footprint. Unfortunately, a significant source of greenhouse gas in the form of exhaust from idling vehicles at intersections continues to go unaddressed. Automotive engineers have made significant progress in increasing the efficiency of cars but braking to a stop, idling and accelerating at a stop light wastes fuel and increases pollution.
Driving along in the rolling computer we used to call a car, we seem to encounter an inordinate number of red traffic lights. Certainly, there are many more vehicles on the roads these days, but how many times have you waited for what seemed like eternity at a stop light where there was no cross traffic? Expansion of our road and highway infrastructure always seems to lag demand created by increased population. The only practical short-term solution is to increase the efficiency of existing roads.
Upgrading traffic control and management systems with artificial intelligence could make a significant dent in the atmospheric overload we have created. Reducing the amount of time drivers spend needlessly waiting at stop lights could also lower the amount of frustration and road rage that seems to be on the rise.
The origins of automated traffic control in the U.S. go back to 1914, when the first electric traffic light was installed in Cleveland, Ohio. It has been estimated that over 90% of traffic lights in use today still use the same basic clock technology that times their operation. We seem to be stuck with technology from another era. There are many practical reasons why clock-based technology persists, including:
- Simplicity and proven reliability with software that requires little technical expertise
- Compatibility and familiarity with the installed base of traffic control systems
- Relative low cost
Intersections with low volume and predictable traffic patterns work well with clock management. As traffic levels increase, newer technology that optimizes flow is required.
The rate of change in this industry appears to be best described as glacial. In 1980, the Split Cycle Offset Optimization Technique (SCOOT) introduced the ability to automatically adjust timing cycles to local traffic conditions. In 1987, the National Cooperative Highway Research Program was created under the U.S. Department of Transportation. Its objective was to study ways to relieve urban traffic congestion. More recently the Intelligent Transportation System (ITS) software platform combines the power of computational, sensing, and communication technologies to more efficiently manage traffic. The Advanced Traffic Management System (ATMS) enhances control capabilities by sending real-time traffic data from cameras, speed sensors, and volume flow to a transportation management center where signaling cycles can be adjusted. Adaptive traffic signal control and signal priority functions are part of an intelligent transportation system. Implementation of these advances has been limited.
Advanced traffic sensors can generate the data necessary to optimize traffic flow in real time. Inductive loop, infrared, radar, video cameras, and doppler sensors are capable of reporting current conditions, including traffic count, vehicle speeds, weather conditions, emergency vehicles, pedestrians, bikers, and road incidents.
Many intersections are monitored by live closed-circuit video to allow human managers to assess traffic conditions as well as record accidents. Data is transmitted to a traffic control center where it is analyzed, saved, and used to drive digital highway alert signs. Some individual intersections utilize sensor output to detect waiting cross traffic and adjust cycle time accordingly. In other cases, traffic lights at every corner of an urban street are coordinated to allow uninterrupted blocks of traffic to flow. Relatively few systems today actually use the data to automatically and proactively manage urban traffic in real time.
Application of artificial intelligence and machine learning to enable traffic pattern recognition introduces an entirely new capability into the quest for increased road efficiency. Systems driven by AI can make split second traffic management decisions that adaptively modify signal light times to facilitate maximum traffic flow by analyzing the total number and type of vehicles approaching from each direction. Dynamic traffic signal control uses algorithms that continuously monitor and calculate the optimal green signal light time to maximize the number of vehicles passing through the intersection. Additional factors include historical traffic data, weekday rush hours, emergency vehicles, road hazards, local events, road construction, and weather conditions. Enormous amounts of collected data can allow an AI system to predict future traffic loads, further optimizing traffic flow. A hard limit is built in to ensure vehicles in a low volume turn or cross lane are not ignored. It offers additional functions, from identifying where and when to make road repairs to evaluating crash history. Using real-time data, AI traffic control systems could preemptively redirect high levels of traffic to alternate routes. The number of vehicles that are running a red light can be interpreted as the level of frustration due to excessive congestion.
Most conventional embedded road sensors detect one to four cars and often only in turn lanes. Powerful microwave sensors can collect data not only at the intersection but also as much as thousands of feet back in multiple lanes to factor in the number of approaching vehicles. Current sensors are capable of detecting at distances of up to 500 meters. Advanced computer vision and forward firing radar systems are capable of reliably detecting and identifying up to several hundred vehicles in adverse conditions including rain, snow, and fog.
Historically, cities have been reluctant to implement new traffic control equipment and software, often preferring to stay with familiar legacy systems that are compatible with the installed infrastructure. Creating AI models and training them is a costly process that requires experienced personnel with new skillsets. Current iterations of AI suffer from occasional “hallucinations,” introducing potential reliability issues. Although reducing traffic congestion can have a positive economic impact, cities may have limited financial incentive to make major investments in new technology for no benefit other than the appreciation of its citizens.
It appears that all the elements — including access to AI computing power, ruggedized sensors, high-definition video cameras, copper or fiber optic network communication links, and customized AI algorithms — are available. Now only the will to utilize them remains.
Adding and upgrading existing high-performance sensors and cameras will create an expanding market for advanced sensors and many types of related interconnects, including ruggedized and environmentally sealed connectors. Fiber optic links to AI computing resources could become another aspect of the current build-out of fiber to the home. Gen 5 and future Gen 6 cellular communications protocol could deliver high-speed secure links and offers another channel by which remote traffic sensors can transmit data and receive traffic light changes. With an extensive network of sensors installed throughout major urban roadways, data collection and analysis will evolve to a cloud-based computing architecture requiring minimal human intervention.
All these advances will result in demand for computing and communication systems with greater capacity, speed, reliability, security, and low-latency networking resources. At some point in the future, vehicles will communicate directly with the road infrastructure as well as with other cars to manage autonomous transport. Eliminating human error is the ultimate solution to minimizing automobile accidents.
Traffic management systems utilizing AI could reduce wasted fuel, air pollution, travel time, frustration, and collisions, while making car travel a more pleasant experience. Seems like a no-brainer to me.
Learn about more interconnect solutions for smart traffic infrastructure.
See Bob Hult’s archive and follow the development of new technologies.
To learn more about the companies mentioned in this article, visit the Preferred Supplier pages for Avnet, PEI Genesis, Phoenix Contact, and TE Connectivity.
Like this article? Check out our other Sensors and Smart Cities articles, our Transportation Market Page, and our 2024 Article Archives.
Subscribe to our weekly e-newsletters, follow us on LinkedIn, Twitter, and Facebook, and check out our eBook archives for more applicable, expert-informed connectivity content.
- AI-Enabled Traffic Control, where are you? - November 5, 2024
- Heat in Data Centers: The Conundrum of Thermal Management - October 22, 2024
- Artificial Intelligence: Impact on the Connector Industry - August 13, 2024