Electronics Design And Algorithm Implementation Of Polarization Grating Sensor For Navigation

The bionic polarization sensor for navigation realizes autonomous navigationorientation, by imitating the polarization sensitive mechanism of insects’ compoundeyes and recognizing polarized skylight distribution pattern. Current prototype sensorshave disadvantages of limited channels, low spatial resolution and low angle accuracythanks to photodiode analyzer which detects polarization. A novel polarized navigationsensor using polarized photoreceptor array, obtains a higher spatial resolution andazimuth accuracy by detecting all-sky atmospheric polarization pattern, thus utilizing anew architecture of hardware-software partitioning to adapt to image processing. Thisnew one is known as polarization grating sensor for navigation.In order to verify the feasibility of this new architecture, the electronic system ofpolarization grating sensor for navigation is designed and implemented. The electronicsystem is not only a platform for algorithm research, which is expected to be extractingnavigation information from image, but also a significant part of prototype sensor. Theelectronic system acquires polarized image through Cameralink interface, and processesit through algorithm modules and Nios II CPU which is the kernel of SOPC systembased on FPGA chip. The azimuth angle is sent to host computer through serial interface.Firstly, an overall design of the sensor’s electronic system is proposed, on the basisof interpreting its principle and algorithm for polarized navigation. The system that madeup of power board, kernel board and interface board, is elaborated from aspects ofarchitecture and chip selection, signal interface and circuit module.Secondly, FPGA circuit design and SOPC system integration are analyzed in detailafter introducing system development environment. The hardware circuit elaborationincludes three aspects. The first one is timing diagram and logic of image acquisitionand restructuring pixels. The second one is computation of sine, cosine and arctangent,implemented by CORDIC circumference algorithm. Polarized angle solver and somecustom instructions for solving solar azimuth are based on it. The third one isimplementation of symmetry axis extraction module by threshold relaxation method,with a pipeline controller, a mean calculator and a relaxation filter. The software systemelaboration includes advanced customization of Nios II CPU to acquire a highperformance SOPC system, tightly coupled memory configuration, HAL memory section mapping and custom instructions integration. Development of datacommunications interrupt functions and solar azimuth program calling custominstructions is also included.Thirdly, verification system with flexible units is built to test image acquisition andcommand communication. Calculation accuracy of hardware and software algorithm isalso verified. The test indicates that the electronic system is able to achieve the task ofimage acquisition and processing with high performance and reliability. The update rateof output is up to20Hz, and the calculating accuracy of hardware module is within±0.01°. Furthermore, the calculating accuracy of solar azimuth program has reached thedesign requirement of max error0.02°3σ.