Research On Real-Time Signal Processing Of Robotics Sensitive Skin Based On DSP+FPGA

Intelligence is one of the main developing directions in the robotics research. The intelligence of robot is mainly represented with ability of working autonomously in unkown environment. Robotics sensitive skin can improve the ability of robot to sense its surroundings greatyly, and it has very important study significance. According to the requirement of the National Natural Science Foundation"Research of the self-operation and collision avoidance fuzzy theory for robots based on sensitive skin", a novel hardware system structure based on DSP+FPGA is designed to process the sensors signal of robotic sensitive skin, and spectrum analysis method based on Fast Fourier Transform is used for analyzing the signal.Firstly, the outputs of the infrared sensors are analog signals, which can be disturbed by noise in optical path and circuit. It has a low precision and stability when using traditional signal processing method, so in this paper a novel spectrum analysis method based on the FFT is discussed and some methods are taken to rectify errors from the limitation of the FFT.Secondly, it discusses the signal processing hardware system of sensitive skin. In order to meet the requirement of real time of system, modularization configuration and DSP+FPGA is used for realizing parallel processing of signal in system hardware. The implementation of FFT algorithms is performed by Field Programmable Gate Arrays (FPGA), and spectrum analysis is done by DSP, and at the same time, sequence control of entire system is performed by CPLD. It takes full advantage of DSP and FPGA, and gives attention to both rapidity and flexibly.Thirdly, it discusses the implementation of a high speed and real-time FFT processor with FPGA. According to the characteristic of our system, data decoding, windowing, saving, FFT, PSD, and result output is performed by FPGA. Radix 2 FFT algorithms is analyzed, and fixed geometry configuration and blocking floating format of data is used to implement 256 points 12 bit complex numbers FFT.At last, it discusses the simulation and experiment of the entire system. The...