Skip to main content
eScholarship
Open Access Publications from the University of California

Improved VLSI architecture for attitude determination computations

  • Author(s): Arrigo, Jeanette Fay Freauf
  • et al.
Abstract

Microelectromechanical sensor (MEMS) technology has produced angular rate sensors that are dramatically smaller in size and lower in cost than in the past. Although the sensor size has decreased significantly, the algorithms that sensor based systems use to compute the attitude of an object have not sustained similar improvements. The primary goal of this dissertation is to develop a new method for computing the attitude of an object that can be incorporated into an application specific integrated circuit (ASIC) or field programmable gate array (FPGA). Determination of the attitude of an object under conditions of varying angular body rate is one of the most complex computations in motion detection. The kinematic equation for rotation, which is used for this computation, is a differential equation that is solved using approximation methods. Since the closed form approximation for the solution of the kinematic equation involves trigonometric functions, it seems natural to consider use of the CORDIC algorithm to perform these calculations. The CORDIC algorithm can also be used to perform arithmetic operations such as multiplication and division. A CORDIC processor can therefore represent a general processing element that is useful throughout the kinematic computation. This dissertation describes the development of a system that uses CORDIC processing, in either a single or parallel processing element configuration, to perform computations needed to solve the kinematic equation for rotation of a solid body. Results of the CORDIC based computation are compared to other more popular techniques such as the Taylor Series approximation method in the thesis. One of the major advantages to using the CORDIC method is the accuracy that is achieved in the attitude computation, particularly at high angular rate inputs. Other methods that approximate the closed form solution either incur truncation errors significant enough to call their use into question, or require more system complexity to overcome difficulties. Extensive simulations and comparisons of the results to other methods have been performed, and they demonstrate the computational accuracy and effectiveness of the new method. Results of this research can be useful in aerospace applications as well as in motion capture aspects of computer graphics. Examples in these areas are presented

Main Content
Current View