An Object Recognition Robot Vision System Design Based On Embedded System

As the modules of embedded system getting smaller and more integrated, the CPU's speed getting higher, embedded system will be applied more and more in robot vision, not just in low level application area. Also, embedded system has the good features such as simple and easy to use, small volume, low cost and so on, it is very adaptable to be used in simple robot vision system.This paper presents the design of the hardware platform of robot vision system. It deeply analyses the format of standard PAL rule TV signal (the source video signal), introduced the principle and performance of 32bit RISC CPU ARM S3C2410A and video decoder chip TVP5145. It explains the TVP5145's initialization and I2C technology, discusses the buffer storage's implementation method which is needed in video signal's transfer and gives a brief introduction of ping-pong buffer storage. It presents the total system's design concept and each module's circuit design and drive software, such as NAND flash, SDRAM, LCD, UART interface circuit and FPGA's configuration by ARM. Flexible setting of interrupt and DMA is adopted in the software design, which improves the efficiency of data transfer and performance of ARM.It successfully realizes high speed image signal's acquisition and display, providing the hardware platform to implement robot vision algorithm (such as object's location and object's tracking).