An Edge Computing System with AMD Xilinx FPGA AI Customer Platform for Advanced Driver Assistance System.

Area of Science:

• Edge Computing
• Field-Programmable Gate Arrays (FPGAs)
• Advanced Driver-Assistance Systems (ADAS)

Background:

• Real-time applications benefit from edge computing and FPGAs.
• Pavement defect detection is crucial for vehicle safety and performance.

Purpose of the Study:

• To design an innovative edge computing system for pavement defect detection in ADAS.
• To integrate AMD Xilinx AI platform and YOLOv3 model for efficient defect identification.

Main Methods:

• Utilized cameras for road scene capture.
• Employed a Deep Learning Processing Unit (DPU) with the YOLOv3 model for defect detection.
• Integrated FPGA technology for customized circuit configuration and real-time processing.

Main Results:

• Achieved high accuracy and efficiency in identifying three types of pavement defects.
• Enabled real-time transmission of defect information via Controller Area Network (CAN) interface.
• Transformed ADAS into an edge computing device for informed decision-making.

Conclusions:

• FPGA-based edge computing significantly enhances ADAS capabilities for automotive safety.
• The system facilitates real-time communication and decision-making, improving driving experience.
• This synergy paves the way for future innovations in vehicle assistance systems.