3-D sensing of object orientation using back propagation neural network

This thesis is about the application of Neural Networks in the sensing of an object position relative to a reference point. An object is rotated a certain angle in space and a Neural Network was then used to estimate its orientation. Previous methods had always been to transform 3-dimensional data into 2-dimensional data before presenting it to the network. In this thesis 3-dimensional data is used as the input to the Neural Network. Training data was obtained via two methods, the first one was through a Pascal program and the other through a commercial CAD software. The input to the Backpropagation Network are coordinates of the object vertices and the target output are the rotation parameters for each particular position.