Project Overview

Taichung City is a dynamic city with rapid growth in population, commerce, and industries. According to statistics, by the end of 2023, there will be 3.02 million motor vehicles in Taichung City, including approximately 1.84 million scooters, accounting for approximately 61%. Scooters should undergo regular exhaust inspections. However, there are so many scooters that it is not feasible to manage them solely through manpower. The Environmental Protection Bureau of Taichung City hopes to help manage the exhaust gas emissions from scooters by developing intelligent inspection tools. We have also developed intelligent inspection tools based on the IOT to ensure air quality in Taichung City.

Organisation

Environmental Protection Bureau of Taichung City Government

Team

The Environmental Protection Bureau of the Taichung City Government is the public agency responsible for implementing environmental protection and promoting environmental strategies in Taichung City. This project was entrusted by the Environmental Protection Bureau of Taichung City Government to implement Green Ideas Synergy Inc. It mainly develops Vehicles with AI license plates and smoke recognition systems to improve scooter exhaust gas and air quality.
Green Ideas Synergy Inc. is committed to environmental management, introducing environmental management and providing air quality improvement strategies through technology, the Internet of Things, artificial intelligence and other technologies. Green Ideas Synergy Inc. is an important partner of the Environmental Protection Bureau of the Taichung City Government.

Project Brief

Vehicles that are not properly maintained can easily produce smoke. To save manpower and facilitate management, the Environmental Protection Bureau commissioned Green Ideas Synergy Inc. to develop AI technology to identify smoke-emitting vehicles. When a smoke-emitting vehicle is seized, the owner will be further notified to improve the vehicle and have it inspected. The AI system uses photos of smoke-emitting scooters provided by the public to train the intelligent learning model to determine the smoke level of scooters. The training target is whether the smoke exhaust level of the scooter exceeds the opacity rate of 30% (the qualification standard is that the opacity rate is less than 30%), used to identify whether it is a high-polluting scooter. In addition to identifying scooters, the identification system can also identify car smoke conditions. To make the equipment highly mobile, the equipment is mounted on vehicles.

Project Innovation/Need

In the past, the judgment of smoke-emitting scooters was based on the visual judgment of personnel, which may have slight errors. To ensure consistency in execution, system judgment was used to ensure that the criteria for judging the smoke exhaust of a scooter were the same, thus reducing the risk of law enforcement caused by personnel misjudgment. Different from the fixed-point license plate recognition system used in the past, the Environmental Protection Bureau will cooperate with taxi operators in 2023 to recruit 70 taxis and install vehicles with AI license plate recognition system equipment; it’s mainly composed of two sets of photography lenses, which can automatically capture driving images and transmit them to the system management platform through the network equipment on the vehicle for automatic AI license plate image recognition and real-time positioning using GPS so that the license plate recognition photos can be accurately attached to the seat. The mark and time prove that the vehicle is driving on the road. In 2023, 13,257 scooters that had not completed regular inspections were seized through cooperation with taxis, with an average of 189 scooters per taxi. The Environmental Protection Bureau also notified scooter owners to complete the inspections within the deadline. This innovative work uses taxis to replace the Environmental Protection Bureau in conducting inspections, saving 780.6 metric tons of carbon dioxide annually.

Design Challenge

There is no quantitative standard for classifying vehicle smoke emission levels, and no relevant information and samples are available for reference. This was the problem we faced at the beginning. To solve this problem, photos of smoke-emitting vehicles reported by the public in the past few years were collected and classified, the characteristics of smoke-emitting vehicles were marked, and a feasible deep-learning model for smoke-emitting vehicle recognition was developed. After a long training period, it is successful in the laboratory and can be truly applied in actual environmental fields. Secondly, regarding vehicle selection, in order not to increase additional carbon emissions, we finally chose to cooperate with existing taxi operators to replace the additional vehicle inspections dispatched by the Environmental Protection Bureau.

Future Impact

When scooters drive on the road, it’s like small chimneys emitting exhaust gas. Driving in the city will affect the air quality. When the scooters in the city complete regular inspections as scheduled and comply with emission standards, it can help improve air quality. When air quality improves, air pollutants can also be reduced, which can positively impact human health. Many countries in South and Southeast Asia also use scooters as their main means of transportation, and there are also problems with scooter emissions. This project is highly mobile and replicable. If there is an opportunity to promote it to these countries in the future, it will help maintain air quality.