Tobias Grüning

Feedforward control design for a four-rotor UAV using direct and indirect methods

To calculate feedforward control strategies by means of dynamic optimization procedures, the alternatives of direct and indirect methods are compared in this paper. The direct optimization approaches that are considered in this paper make use of the Hermite-Simpson method and of the Legendre Pseudospectral method as underlying discretization strategies. Their description is followed by a short introduction to the maximum principle of Pontryagin, i.e., the indirect method. All procedures are employed to compute feedforward control sequences for the flight control of a four-rotor UAV, which is an unstable, nonlinear multi-input multi-output system. The optimization results are compared with respect to their accuracy and applicability.

Optimization of the longitudinal dynamics of parallel hybrid railway vehicles

In this paper, a basic simulation model of the longitudinal dynamics is developed for a diesel hybrid railway vehicle with a battery as energy storage system. This model represents an inverse system description that employs a given duty cycle to compute all system variables in a purely algebraic manner. Based on the developed simulation model, three different optimization strategies aiming at the minimization of fuel consumption are investigated. The optimization parameters that are used in these approaches allow for a modification of certain phases of the operating strategy. The optimization results indicate possible reductions of the fuel consumption of more than 20 percent in comparison with a pure diesel operation.

Regular expressions for decoding of neural network outputs

This article proposes a convenient tool for decoding the output of neural networks trained by Connectionist Temporal Classification (CTC) for handwritten text recognition. We use regular expressions to describe the complex structures expected in the writing. The corresponding finite automata are employed to build a decoder. We analyze theoretically which calculations are relevant and which can be avoided. A great speed-up results from an approximation. We conclude that the approximation most likely fails if the regular expression does not match the ground truth which is not harmful for many applications since the low probability will be even underestimated. The proposed decoder is very efficient compared to other decoding methods. The variety of applications reaches from information retrieval to full text recognition. We refer to applications where we integrated the proposed decoder successfully.