With Taiwan’s cities expanding and vehicle numbers climbing, effective traffic management is vital to ease congestion, cut pollution, and keep roads safe.
Artificial intelligence has become an essential tool for engineers looking to improve traffic flow and support Taiwan’s smart city ambitions. AI is now central to transportation policy, with officials emphasizing its role in improving efficiency, safety, and resilience.
In pursuit of practical applications, the Ministry of Transportation and Communications (MOTC) is focused on implementing AI-based traffic management solutions. Over the past two years, its Department of Transportation has applied AI to traffic signal control. Algorithms can adjust signals in real time, improving flow during peak hours or sudden congestion. The result: safer, faster commutes and reduced pollution.
The MOTC emphasizes that AI is already being deployed across multiple pilot projects, ranging from traffic violation detection and real-time alerts at intersections to warning systems at unsignalized junctions.
“Trials of AI-based traffic signal control are showing promise in improving road safety and reducing congestion,” the ministry told Taiwan Business TOPICS in a letter. Generative AI has also been adopted in highway customer services, enabling more intuitive and efficient public assistance for drivers.
The number of scooters on Taiwanese roads is staggering. Estimated at over 14 million, two-wheelers still dominate despite declines in sales and registrations. Their unpredictable movements make modeling and signal timing difficult.
Taiwan is one of the few countries to effectively model scooter behavior, accounting for lane weaving and rapid direction changes. The MOTC is using AI to monitor and predict scooter behavior in cities. Understanding how scooters interact with cars will be key to reducing accidents and improving traffic flow, an MOTC AI expert notes.
However, AI use in traffic planning is just one part of the government’s road-related initiatives to build a world-class AI industry ecosystem and turn Taiwan into an “AI island” — one of the Lai Ching-te administration’s main policy objectives.
AI-based pilot projects
The Software Technology Institute (STI), a research unit of the government-sponsored Institute for Information Industry, has helped reduce traffic incidents in Taichung by 39% by using AI to identify common violations such as illegal U-turns, red-light running, and unlawful turns.
Yet AI has limitations: rainy or foggy conditions and data latency can still cause errors. In clear weather, the system reliably classifies vehicles, buses, bicycles, and pedestrians.
AI algorithms are also being used to optimize bus scheduling. By analyzing traffic and ridership data, systems can predict congestion and dispatch standby buses. Prioritizing buses at intersections reduces delays and makes public transport more attractive.
The MOTC notes that AI technologies are also being applied well beyond roads, particularly in smart railways and infrastructure management. These projects include train operation monitoring, drone-based safety inspections, and AI-assisted bridge and port assessments to anticipate maintenance needs.
“We are actively developing AI-powered warning systems along coastal highways and ports, as well as tools for real-time ocean condition and thunderstorm forecasting,” the ministry said. Such applications aim to strengthen disaster response capacity for land, sea, and air transportation.
Another notable effort is the Intelligent Transportation System (ITS) Development and Construction Program, a four-year initiative running through 2028. It aims to integrate sensors into major roads to collect data on counts, density, and speed, while adding smart intersections along provincial roads to improve flow for commercial vehicles.
Pedestrian movement is also under study. A National Taipei University of Technology project proposes adjusting signals based on the number of people waiting to cross, enhancing both safety and flow.
In addition, the integration of drones and AI image analysis is being used to detect patterns at high-risk intersections. A two-year project launched in 2024 has assessed 12 sites using drones and AI-powered image recognition. What once took three years of data gathering can now be done in five months.
Dealing with a shrinking population, Taiwan is already experiencing worker shortages in the transport sector. The MOTC estimates a shortfall of about 5,000 bus drivers. Integrating AI, particularly with autonomous buses, could help. Eighteen trials are underway, including a TSMC shuttle that has already logged 50,000 km. Other pilots have taken place in Danhai and at the Hsinchu County-based Industrial Technology Research Institute.
New AI-driven systems, such as automatic inner-wheel braking and lane-keeping assist, are already in use, integrating connected-vehicle solutions to enhance road safety.
Being smart with data
To support AI systems, a variety of data types are collected and processed. Along with road-embedded sensors, vehicle detector data comes from traffic cameras. The Internet of Vehicles (IoV) adds communications between infrastructure and vehicles, while weather, accidents, and even social events enrich predictive models.
Ensuring accurate data is critical. “The key to developing AI technology is data,” says STI Director-General Meng I-heng. He notes that 4 million pieces of data were used to train STI’s Sardina recognition model, achieving 96% accuracy.
Another player is Chunghwa Telecom, which leverages AI modeling as part of its Traffic Big Data Analytics and Intelligent Transportation Initiative. Using anonymized cell phone signals, Chunghwa estimates traffic flow and parking availability. “All CCTV cameras are now AI-based,” said Vincent C. Chen, vice president of Chunghwa Telecom, at this year’s Smart City Summit and Expo in March. These systems can dynamically detect road conditions, improving traffic prediction and signal control.
Taiwan is collaborating with Singapore and Tokyo on AI traffic initiatives, sharing common problems and solutions. It is also working with smart cities in India on multi-camera tracking and AI governance standards.
On the policy front, the MOTC convened the first meeting of the Transportation AI Promotion Committee in August 2025 to translate AI’s potential into regulations and public services. The ministry outlined seven strategies, including strengthening AI data and infrastructure, promoting inclusive digital services, and cultivating diverse AI talent. Recognizing that transportation is considered a high-impact AI category globally, the ministry says it is drafting risk-classification guidance in line with the Executive Yuan’s AI Basic Act, now under legislative review.
“Moving forward, we will continue to place people-centered transportation at the core while driving innovative AI applications to create a smarter, safer, and more sustainable future,” the MOTC told TOPICS.
While there is a rich variety of data-gathering methods, the MOTC faces the task of balancing short-term enforcement with longer-term planning around governance, AI infrastructure, and predictive analytics. Globally, Taiwan remains at the forefront of AI traffic innovations, with public–private cooperation pushing solutions into smart city planning and future traffic management.
