The Effect of Primitive Shapes on NAI Values for a Simple Satellite Model

Paper number

IAC-08.B2.4.12

Author

Mr. Damian Rogers, Ryerson University, Canada

Year

2008

Abstract

Autonomous rendezvous and docking missions require extremely precise indications of the location, speed, and acceleration of a target in order for the attitude determination and control systems to precisely predict the motion of the objects and ensure successful docking. One such way of determining these factors with computer vision systems is via a LIDAR sensor. It is possible to simulate a LIDAR sensor using ray tracing in computer software programs on a computer aided design (CAD) model. Ray tracers provide a range for each individually scanned point, which can then be fed into an overall model of the object and compared to the original CAD model.

A method to compare the model created from the vision system to the original model is to use the noise amplification index (NAI) value. Information can be obtained about how well the sensor can accurately predict the pose of the satellite and thus how successful docking might be. The NAI value of a certain model in its pose is determined by finding the minimum and maximum Eigenvalues of a matrix resulting from principle component analysis.

The purpose of this paper is then to explore the effect that simple shapes have on the pose estimation accuracy of astronautical craft. Starting with a simple cube satellite and adding appendages made from single primitive shapes and combinations of these, the pose estimation accuracy is then calculated for three different poses. Results are tallied and compared with the accuracies of the other combinations in order to compare trends. NAI values are compared directly due to the relation that increasing NAI values correlate to decreasing pose estimation error.

It was found that the actual shape of the appendage had little effect on the overall pose estimation accuracy, but rather the number of visible edges and corners was the deciding factor. Thus, a rectangle and square had very similar values since the number of edges remained the same between the two shapes; however, comparing a rectangular shape to a spherical one yielded very different results, which led to the conclusion that shapes with more edges and corners resulted in better pose estimation. Obtained results will be used in future experimentation to increase the accuracy of pose estimation for future rendezvous and docking mission success.

Abstract document

IAC-08.B2.4.12.pdf

Manuscript document

IAC-08.B2.4.12.pdf (🔒 authorized access only).

To get the manuscript, please contact IAF Secretariat.