Autonomous orientation and geolocation via celestial objects

Cheng Liu, Purdue University

Abstract

Based on a hemispherical sensor geometry, a novel celestial navigation system is developed to use celestial objects to determine the absolute location and orientation information without the aid of satellites via two different approaches. The first approach employs a hemispherical arrangement of light intensity sensors to determine the vector to the dominant light source. We present the sensing system to measure the sun vector via least squares method and achieve the application of a low-cost, small-sized solar compass. The system is shown to work well under ideal conditions but is susceptible to noise and uncertainties in some situations. The second approach uses camera instead of light sensor, enabling the detection of celestial objects in a much more accurate and flexible fashion. An elaborate camera calibration was conducted to mitigate lens distortion and explore the transformation from image pixel coordinates to stationary world coordinates. With suitable image processing strategies, the system is able to use images of the sun and moon for the purpose of obtaining azimuth and zenith angles in spite of various disturbances. Given the results measured with our sensing systems, a generalized geolocation method is presented to estimate the absolute location on the earth. The approach, inspired by the traditional manual intercept method, automates all of its steps in an iterative fashion. It derives both the geolocation estimates and the error intervals based on measurement noise levels. This method is superior to most traditional approaches in that it derives the estimates even with lower quality sensors.

Degree

M.S.M.E.

Advisors

Ariyur, Purdue University.

Subject Area

Electrical engineering|Mechanical engineering

COinS