AI Image Detection Automatically Recognizes Pedestrians and Vehicles, Improving Wait Time at Major Intersections
Fixed Timer Controls on Traffic Lights Increase Waiting
Getting stuck at a red light on an otherwise empty street is a scenario familiar to people who drive at night on Taipei’s major roads. This phenomenon occurs when traffic lights are set to timers adjusted for peak daylight hours to accommodate for the city’s usual hustle and bustle. Taipei’s traffic lights’ lengthy countdowns exist to provide enough time for pedestrians to safely cross wide roads. Such arrangements are necessary during the day, but become burdensome around midnight when fewer pedestrians and vehicles traverse Taipei’s streets. Because of fixed timers, adjusted to daylight traffic, drivers on Taipei’s major arterial roads might find themselves waiting needlessly for extended periods of time.
In an effort to prevent this phenomenon and enhance traffic efficiency, Taipei City introduced an AI smart traffic light control system in 2019. This system, integrated by THI Consultants Inc., combines street cameras with Advantech’s MIC-720 industrial-edge AI host, and Yuan High-Tech’s AI algorithm. It provides flexible traffic light adjustments based on actual traffic to remedy efficiency issues.
Surveillance cameras are equipped with AI image detection that collects data on the status of cars and pedestrians. This system uploads images in real-time to the MIC-720. Next, MIC-720 conducts AI image detection and calculates the positions of people and cars. When necessary, this system can prolong green lights and activate responsive signals to adjust traffic lights in real-time. There are three justifications for traffic light adjustment:
- When arterial road lights are green and there are pedestrians or vehicles waiting at intersections, the timer is adjusted to its minimum value.
- When side road lights are green and there is traffic on it, the green light time is extended.
- When there is no traffic on side roads, their green lights are deactivated to maximize efficiency on arterial roads, which benefit from green wave.
In order to ensure the precision of AI image recognition algorithms, Yuan High-Tech adjusted the AI’s label and item recognition capacities. The adjustment, made in the initial stages of system development, yielded algorithm training and model construction adaptable to high-contrast scenarios found at night or on sunny days.
AI Smart Signal Control Yields 1.83 Million New Taiwan Dollars in Savings
This AI smart traffic signal control system can change specific traffic lights according to traffic flow–it gives traffic on side roads enough time to pass while increasing arterial road green light time when side roads are empty. In sum, it enhances traffic flows for arterial roads during the day, while reducing waiting times at red lights at night.
Statistical analysis conducted after the implementation of the AI smart traffic signal control system suggests a stark improvement in wait times. The average wait time for a vehicle across different time spans has decreased 15 ~ 78%. Night wait times at red lights on arterial roads decreased by 35% and arterial roads green light periods increased by 7 ~ 79%. In sum, traffic ran more efficiently.
Reducing the wait time at red lights corresponded with a reduction in vehicle idle time and gas consumption. When converted into an energy-saving and carbon-reducing data, this system reduces carbon dioxide emissions by around 23 tons every year and results in an annual economic benefits of around TWD 1.83 million.
Taipei City expanded its intersections in 2020 in the hope of using advanced AI smart signal control systems, in combination with reliable and accurate detection equipment, to effectively reduce unnecessary waiting and improve traffic efficiency at arterial road intersections. Additionally, this system reduces traffic jams and increases overall capacity.
Advantech’s Edge AI System Plays Key Role
Due to latency in data transmission, AI smart signal control systems use an edge computing structure in which hardware plays a significant role. This hardware must meet three criteria: First, it must deliver high-performance to conduct stream decoding and image processing. Second, as the success of image recognition is easily affected by environment changes, this hardware must wield enough processing power to facilitate deep learning and deduction. Finally, its hardware must be capable of withstanding the harsh environmental conditions found on street corners, like high temperatures, precipitation, and humidity.
Advantech’s MIC-720AI is equipped with NVIDIA® Jetson™ Tegra X2. This solution’s elegant fanless design features built-in image recognition modules and supports deep learning models and processes. Its wide range operating temperature makes it ideal for outdoor applications that require AI edge computing. By simply integrating a signal control system, MIC-720AI can be upgraded to an AI solution.
Advantechs’ solution meets specific demands and provides excellent after sales services. Outdoor electrical supplies and environmental conditions are less stable than those found in indoor control centers creating frequent power outages. In the event of an outage, the AI smart signal control systems will stop functioning, requiring engineers to reactivate it onsite. To deal with this issue, Advantech uses special circuit boards that automatically re-activate devices after power has returned, saving THI Consultants engineers’ time.
The Vice General Manager at THI predicts Advantech, as a major global industrial computer manufacturer, will aid their company with future global market development. In recent years, Advantech has invested significant resources towards developing edge computing equipment. THI Consultant’s extensive experience in integrating smart traffic systems, makes it an ideal partner in creating practical edge computing applications. Together, THI Consultant , Yuan High-Tech and Advantech can pursue the development of the smart traffic system market.
- Traffic ran more efficiently with average wait time decreased 15 ~ 78%.
- A reduction in vehicle idle time and gas consumption in an annual economic benefits of around TWD 1.83 million.
- Equipped with high performance, Advantech MIC-720AI is capable of withstanding the harsh environmental conditions found on street corners, like high temperatures, precipitation, and humidity.
- Advantechs’ solution meets customers' specific demands and provides excellent after sales services to save customer's engineers' time.
Smart Transportation Partners
AI Inference System based on NVIDIA® Jetson™ Tegra X2 256 CUDA Cores