Vision-based Development Of Advanced Driver Assistance System Hardware Platform

Advanced assisted driving system,referred to as ADAS,has been extensively studied in recent years as a technical stage for the transition to unmanned driving systems and a technical means to ensure the personal safety of drivers.Whether from the perspective of development costs or technology accumulation,vision-based ADAS solutions are currently the mainstream research direction.The visual ADAS hardware platform is essentially an image processing platform,but compared to general image processing platforms,its performance requirements are higher and peripheral interfaces are richer.With the rapid development of the computer and semiconductor industries in recent years,more and more high-performance chips have been developed,which provides favorable conditions for the development of such hardware platforms.This paper mainly researches the existing visual ADAS schemes and image processing platform hardware design schemes,and completes the development of the visual ADAS hardware platform based on this.The platform uses the high-performance system-level chip BCM2837 as the core processor to take charge of the ADAS algorithm and some video image acquisition tasks.In order to implement the function of data interaction between the platform and the car and expand more functional modules and sensor modules,micro-control is used.XC2234 L is used as a coprocessor to expand CAN-based external devices,UART-based GPS modules,and I2C-based sensor devices.The programmable device XC6SLX9 is used as a coprocessor to expand four-way wide-angle cameras.At the same time,in order to ensure the core processor's control of the coprocessor and the data interaction between the processors,this article formulates the upper and lower layers with BCM2837 as the upper processor,XC2234 L and XC6SLX9 as the lower processor,and SPI as the transmission medium.letter of agreement.In addition,this article also transplants a customized Linux system to the hardware platform,and on this basis further completes the testing of the hardware platform and verifies the platform's functions.The platform supports the collection of video data from up to 7 cameras.It can implement a variety of ADAS solutions including front view,surround view,and driver status monitoring.It supports CAN communication and Wi Fi wireless communication,and has the capability of data interaction with cars.And the ability to network.In addition,the platform adopts a modular design and a combination of stacking methods.It has the advantages of small size,tailorable functions and expandability.It is very suitable for ADAS related research,and provides new ideas for the development of similar platforms.