This paper describes our current work in developing a vision-based tracking and trajectory prediction system for an aerial robot based on low-cost digital cameras, image processing techniques, and a filtering and prediction algorithm. The system determines the pose (location and orientation) of a miniature airship, online during indoor flight, and will be used in a development framework for a future autonomous flight control system. Object localization is achieved by tracking an infra-red target array mounted to a model airship. Its pose in three-dimensional space can be computed from corresponding points in the images of two cameras which are calibrated in a global coordinate system. The calibration procedure and the localization, as well as some aspects of the measurement accuracy achieved, are discussed. Real-world applications provide an uncertain static or dynamic environment which complicates the tracking of a target. To overcome problems due to noisy data or even failed target detection in image frames, a filtering procedure is applied for estimating the airship's pose. In a first step, points in the two-dimensional image planes are directly tracked and propagated forward to the vehicle pose. In a second step, an adaptive noise Kalman filter is applied for estimating and predicting the flight trajectory. Its state is propagated back to points in the image planes to guide the detection algorithm by defining regions of confidence. Both approaches are combined in a tracking algorithm. In-flight measurements are used to validate the parameters of the adaption procedure. Some experimental results are shown.
Neural networks, especially in nonlinear system identification and control applications, are typically considered to be black-boxes which are difficult to analyze and understand mathematically. Due to this reason, an in- depth mathematical analysis offering insight into the different neural network transformation layers based on a theoretical transformation scheme is desired, but up to now neither available nor known. In previous works it has been shown how proven engineering methods such as dimensional analysis and the Laplace transform may be used to construct a neural controller topology for time-invariant systems. Using the knowledge of neural correspondences of these two classical methods, the internal nodes of the network could also be successfully interpreted after training. As further extension to these works, the paper describes the latest of a theoretical interpretation framework describing the neural network transformation sequences in nonlinear system identification and control. This can be achieved By incorporation of the method of exact input-output linearization in the above mentioned two transform sequences of dimensional analysis and the Laplace transformation. Based on these three theoretical considerations neural network topologies may be designed in special situations by pure translation in the sense of a structural compilation of the known classical solutions into their correspondent neural topology. Based on known exemplary results, the paper synthesizes the proposed approach into the visionary goals of a structural compiler for neural networks. This structural compiler for neural networks is intended to automatically convert classical control formulations into their equivalent neural network structure based on the principles of equivalence between formula and operator, and operator and structure which are discussed in detail in this work.