Traffic AI: A Real-Life Use Case
It is already commonplace to see cities collect big data from sensors and the Internet of Things, and using it to improve city services. The next step is using Artificial Intelligence (AI) to derive insights at a speed and scale that would be impossible for human city officials.
Read on to learn how traffic AI systems work, how they can help identify traffic patterns, guide traffic lights, and improve safety, and read a real-world case study of traffic AI in the world’s most congested city, New Delhi.
In this article you will learn:
What Is Traffic AI?
The term traffic AI refers to applications of Artificial Intelligence (AI) and Machine Learning (ML) in traffic systems. Traffic AI systems collect and analyze traffic data, generate solutions, and apply them to the traffic infrastructure.
The field of traffic AI is still experimental, with organizations, government bodies, and universities contributing to solving the traffic optimization problem. Presently, traffic AI systems give cities the power to improve traffic monitoring and transport data analytics. In the future, a reliable, stable traffic AI may be capable of autonomously controlling traffic flow.
Traffic AI works by collecting data from connected traffic systems, which provide input about live traffic or years of historical traffic behavior. In order to understand this unstructured data, it uses machine learning models to process, analyze, and learn about traffic infrastructure. The AI then uses these insights to solve traffic problems.
Object Detection and Event Prediction in Traffic Analysis
Traffic AI systems are capable of processing and analyzing a variety of data types, including images and videos. Aided by computer vision techniques, traffic AI systems can recognize objects in images and video from traffic cameras, by breaking them down into identifying features, and matching them to classes, such as “car”, “motorcycle” or “pedestrian”.
Object detection and classification can help distinguish between pedestrians and vehicles, between different types of vehicles, and between buildings and living creatures. Traffic AI can count vehicles, determine if a road is congested, and re-route traffic accordingly.
At a certain point, after accumulating enough data, traffic AI can start detecting patterns to better understand the traffic ecosystem. It can learn to differentiate between events that may cause congestion, such as a car accident or rush hour. When traffic AI learns that rush hour happens at certain times, it can perform predictive analysis and use the data to improve traffic flow.
Common Use Cases of Traffic AI
The following are three ways cities use traffic AI to improve traffic management and benefit drivers.
1. Traffic Lights
Traffic AI systems can optimize traffic lights and reduce waiting time at intersections. The AI detects vehicles in images from traffic cameras. The information is sent to a control center, where algorithms analyze traffic density. If the system detects congestion, it can direct traffic lights to re-route traffic, based on real-time data.
2. Traffic Patterns
Using predictive analytics, traffic AI systems can identify traffic patterns and prevent or alleviate road congestion before it occurs. Smart cities can integrate a traffic AI system with their Intelligent Transport System (ITS), or build the AI into their Advanced Traffic Management System (ATMS).
3. Improved Safety
Traffic AI systems can make emergency services and public transportation safer and more efficient. Traffic AI differentiates between types of road users and can prioritize traffic flow accordingly. Once the AI detects an emergency vehicle, it can re-route traffic to help emergency personnel reach their destination faster. If a bus is stuck in traffic, the AI can help it make a stop on time.
New Delhi: A Traffic AI Use Case
The city of New Delhi was recently ranked among the world’s worst cities by traffic congestion. Delhi drivers spend approximately 58% more of their time in traffic than drivers in any other city in the world. A 2017 study revealed that 1 out of 7 traffic policemen suffer from respiratory issues due to prolonged exposure to air pollution.
Officials from the Delhi Traffic Police were determined to find a solution to their ever-increasing traffic problem, and decided to implement a new AI-driven traffic management system (ITMS).
Delhi’s ITMS infrastructure is spread across the city, with more than 7,500 CCTV cameras, automated traffic lights, and 1,000 LED signs that provide drivers with real-time data. The AI analyzes data generated by cameras and sensors, and provides city officials with real-time traffic insights. Authorities can make real-time decisions to balance traffic flow, and can identify trends to build long term plans for alleviating traffic problems.
In addition, Delhi’s ITMS reduced manual intervention in traffic control to a minimum. With traffic AI monitoring the flow in congested roads, traffic policemen and government officials can manage and direct traffic from a safe distance.
Traffic AI—a Superhuman Traffic Controller
What if we could have a police officer or city official watch every road and junction, and take real-time decisions to improve traffic conditions? This would be infeasible to do with humans, but it is a reality today with traffic AI.
The ability to monitor huge data volumes in real-time and take intelligent decisions can have a major impact on congested cities, above and beyond what they have already accomplished with traditional analytics.
The more AI is allowed to operate freely, re-routing traffic according to data-driven insights, the bigger impact it can have on traffic efficiency. However, AI can make mistakes that have unintended consequences. Policymakers will need to take a balanced approach to introducing AI that allows the system to learn at scale while leaving in place controls for human intervention.