AI-Powered Surveillance System for Unmanned Gas Stations in Japan
2/20/2024
According to the Agency for Natural Resources and Energy in Japan, the country has seen a significant decrease in the number of gas stations in the past twenty years, with over 40% closing down. This trend poses a significant challenge regarding gas station accessibility. Moreover, upkeep and maintenance of the existing fuel distribution infrastructure is both costly and laborintensive, often requiring trained personnel who are in short supply. This human resource shortage is compounded by Japan’s declining birthrate, population aging, and depopulation in rural areas, all of which will adversely affect the fuel distribution network’s efficiency. In light of these challenges, a leading Japanese gas station brand turned to Advantech’s cutting-edge AI and IoT innovations for a transformative solution to optimize its fuel distribution network.
Challenge
Moving away from traditional full-service gas stations, the brand pursued a novel approach – establishing small-scale unmanned self-service gas stations empowered by AI and computer vision. Given the shortage of field service personnel, the brand required a surveillance system that was robust, durable, resilient in extreme outdoor conditions, and required minimal maintenance or replacement. The system would not only need to perform real-time AI analysis on both vehicles and drivers, but also interface with the fuel dispensers to execute corresponding actions based on the analysis results. Thus, key requirements included having a substantial number of I/O and serial ports to ensure seamless integration with the fuel dispensers, and outstanding computing performance to handle multiple high-resolution cameras. To support their 24/7 operation into the future, securing replacement parts and local after-sales services were also key considerations.
Solutions
Leveraging the NVIDIA Jetson AGX Orin 64GB module, the AIR-030 delivers up to 275 trillion operations per second
(TOPS), supporting multiple concurrent AI inference pipelines and high-speed data transmission for detecting vehicle
and driver behavior, as well as controlling fuel dispensers. Additionally, it boasts an Ethernet speed of 2.5GbE to
support high-resolution camera connections in vision AI processing. The solution includes the Linux Ubuntu 20.04
OS environment and supports NVIDIA JetPack™ SDK 5.0 and above, as well as Advantech’s Edge AI SDK. This
not only ensures superior AI inferencing, computer vision, and image processing, but also reduces testing and
implementation times.
By utilizing Advantech’s edge AI SDK, the brand’s SI partner helped verify the performance of the AIR-030, AI
containers, AI models, and libraries during project planning, significantly reducing the time spent on trial and error.
Furthermore, the abundant resources from the NVIDIA® Jetson Ecosystem greatly reduced development time while
optimizing the AI applications. This facilitated the smooth integration of cutting-edge AI technology into their AI
software at minimal expense.
The AIR-030’s compact size makes it easy to fit into virtually any gas station design, and its rich set of interface options, including PoE, LAN, and USB, allows for connecting to various types of cameras. By integrating the fuel dispensers and surveillance system into the AIR-030 via a COM port, the brand is able to perform AI analysis upon fuel nozzle pickup and return while also monitoring for potential hazards, such as cigarette smoking or improper fuel nozzle usage. Upon detection of a hazardous event, the fuel supply is suspended automatically, and an alarm LED connected to the AIR-030 also lights up to alert the driver.
Results
Benefits
• Employs a Linux Ubuntu 20.04 OS environment and is compatible with NVIDIA JetPack™ SDK 5.0, guaranteeing exceptional performance in AI inferencing, and image processing.
• Leverages Advantech’s edge AI SDK for seamless
verification of AI containers, models, and
libraries, streamlining the project development
process.
• Achieves enhanced safety through integrating the
fuel dispensers and AIR-030, enabling automatic
fuel supply cut-off upon hazard detection.
Results
The brand not only simplified the process of establishing small unmanned stations but also facilitated the efficient realization of their goal to rapidly expand their gas station coverage.
-25/AIR-030_04--20230601175550.jpg)
-25/AIR-020_Front-right1-_Banner2021082715073020210827164439.jpg)
