@misc{AppelhansKampmannMottoketal., author = {Appelhans, Marie-Luise and Kampmann, Matthias and Mottok, J{\"u}rgen and Riederer, Michael and Nagl, Klaus and Steffens, Oliver and D{\"u}nnweber, Jan and Wildgruber, Markus and Roth, Julius and Stadler, Timo and Palm, Christoph and Weiß, Martin Georg and Rochholz, Sandra and Bierl, Rudolf and Gschossmann, Andreas and Haug, Sonja and Schmidbauer, Simon and Koch, Anna and Westner, Markus and Bary, Benedikt von and Ellermeier, Andreas and V{\"o}gele, Daniel and Maiwald, Frederik and Hierl, Stefan and Schlamp, Matthias and Ehrlich, Ingo and Siegl, Marco and H{\"u}ntelmann, Sven and Wildfeuer, Matthias and Br{\"u}ckl, Oliver and Sterner, Michael and Hofrichter, Andreas and Eckert, Fabian and Bauer, Franz and Dawoud, Belal and Rabl, Hans-Peter and Gamisch, Bernd and Schmidt, Ottfried and Heberl, Michael and Thema, Martin and Mayer, Ulrike and Eller, Johannes and Sippenauer, Thomas and Adelt, Christian and Haslbeck, Matthias and Vogl, Bettina and Mauerer, Wolfgang and Ramsauer, Ralf and Lohmann, Daniel and Sax, Irmengard and Gabor, Thomas and Feld, Sebastian and Linnhoff-Popien, Claudia and Ławrowski, Robert Damian and Langer, Christoph and Schreiner, Rupert and Sellmair, Josef}, title = {Forschung 2019}, editor = {Baier, Wolfgang}, address = {Regensburg}, organization = {Ostbayerische Technische Hochschule Regensburg}, isbn = {978-3-9818209-7-3}, doi = {10.35096/othr/pub-789}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:898-opus4-7890}, pages = {72}, abstract = {Bericht mit Forschungsprojekten aus verschiedenen Bereichen der OTH Regensburg mit dem Schwerpunktthema "K{\"u}nstliche Intelligenz" und einem Gespr{\"a}ch zur "Medizin der Zukunft"}, subject = {Forschung}, language = {de} } @unpublished{BockWeiss, author = {Bock, Sebastian and Weiß, Martin Georg}, title = {Local Convergence of Adaptive Gradient Descent Optimizers}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:898-opus4-7546}, abstract = {Adaptive Moment Estimation (ADAM) is a very popular training algorithm for deep neural networks and belongs to the family of adaptive gradient descent optimizers. However to the best of the authors knowledge no complete convergence analysis exists for ADAM. The contribution of this paper is a method for the local convergence analysis in batch mode for a deterministic fixed training set, which gives necessary conditions for the hyperparameters of the ADAM algorithm. Due to the local nature of the arguments the objective function can be non-convex but must be at least twice continuously differentiable. Then we apply this procedure to other adaptive gradient descent algorithms and show for most of them local convergence with hyperparameter bounds.}, language = {en} } @unpublished{BockWeiss, author = {Bock, Sebastian and Weiß, Martin Georg}, title = {Rotation Detection of Components with Convolutional Neural Networks}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:898-opus4-4120}, abstract = {The main issues in many image processing applications are object recognition and detection of objects, which answers the questions whether an object is present and if it is present, where it is located. Popular object detection algorithms like YOLO use a regression formulation for the whole problem, especially for the bounding box parameters. In production industry the setting usually is different: One usually knows the object type and rather wants to know with high precision where the object is. We study a prototype application in this area where we identify the rotation of an object in a plane. To solve this problem use a regression approach with a CNN architecture as a function approximator. We compare our results to standard image processing algorithms, which do not use neural networks, and present quantitative results on the accuracy. CNNs seem at least competitive to classical image processing.}, language = {en} } @inproceedings{BockWeiss, author = {Bock, Sebastian and Weiß, Martin Georg}, title = {Non-Convergence and Limit Cycles in the Adam Optimizer}, series = {Proceedings of the 28th International Conference on Artificial Neural Networks, 2019, Munich, Germany, September 17-19}, booktitle = {Proceedings of the 28th International Conference on Artificial Neural Networks, 2019, Munich, Germany, September 17-19}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:898-opus4-490}, pages = {232 -- 243}, abstract = {One of the most popular training algorithms for deep neural networks is the Adaptive Moment Estimation (Adam) introduced by Kingma and Ba. Despite its success in many applications there is no satisfactory convergence analysis: only local convergence can be shown for batch mode under some restrictions on the hyperparameters, counterexamples exist for incremental mode. Recent results show that for simple quadratic objective functions limit cycles of period 2 exist in batch mode, but only for atypical hyperparameters, and only for the algorithm without bias correction. We extend the convergence analysis to all choices of the hyperparameters for quadratic functions. This finally answers the question of convergence for Adam in batch mode to the negative. We analyze the stability of these limit cycles and relate our analysis to other results where approximate convergence was shown, but under the additional assumption of bounded gradients which does not apply to quadratic functions. The investigation heavily relies on the use of computer algebra due to the complexity of the equations.}, subject = {Neuronales Netz}, language = {en} } @inproceedings{BockWeiss, author = {Bock, Sebastian and Weiß, Martin Georg}, title = {A Proof of Local Convergence for the Adam Optimizer}, series = {Proceedings of the 2019 International Joint Conference on Neural Networks (IJCNN), 2019, Budapest, Hungary, July 14-19}, volume = {2019}, booktitle = {Proceedings of the 2019 International Joint Conference on Neural Networks (IJCNN), 2019, Budapest, Hungary, July 14-19}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:898-opus4-501}, pages = {1 -- 8}, abstract = {Adaptive Moment Estimation (Adam) is a very popular training algorithm for deep neural networks, implemented in many machine learning frameworks. To the best of the authors knowledge no complete convergence analysis exists for Adam. The contribution of this paper is a method for the local convergence analysis in batch mode for a deterministic fixed training set, which gives necessary conditions for the hyperparameters of the Adam algorithm. Due to the local nature of the arguments the objective function can be non-convex but must be at least twice continuously differentiable.}, subject = {Neuronales Netz}, language = {en} } @unpublished{Weiss, author = {Weiß, Martin Georg}, title = {Optimization of Cartesian Tasks with Configuration Selection}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:898-opus4-20176}, abstract = {A basic task in the design of an industrial robot application is the relative placement of robot and workpiece. Process points are defined in Cartesian coordinates relative to the workpiece coordinate system, and the workpiece has to be located such that the robot can reach all points. Finding such a location is still an iterative procedure based on the developers' intuition. One difficulty is the choice of one of the several solutions of the backward transform of a typical 6R robot. We present a novel algorithm that simultaneously optimizes the workpiece location and the robot configuration at all process points using higher order optimization algorithms. A key ingredient is the extension of the robot with a virtual prismatic axis. The practical feasibility of the approach is shown with an example using a commercial industrial robot.}, subject = {Intralogistik}, language = {en} } @inproceedings{Weiss, author = {Weiß, Martin Georg}, title = {Optimization of Cartesian Tasks with Configuration Selection}, series = {2nd IMA Conference on Mathematics of Robotics: online September 8-10, 2021}, booktitle = {2nd IMA Conference on Mathematics of Robotics: online September 8-10, 2021}, editor = {Holderbaum, William and Selig, J. M.}, publisher = {Springer International Publishing}, address = {Cham}, isbn = {978-3-030-91351-9}, doi = {10.1007/978-3-030-91352-6_16}, pages = {153 -- 160}, abstract = {A basic task in the design of an industrial robot application is the relative placement of robot and workpiece. Process points are defined in Cartesian coordinates relative to the workpiece coordinate system, and the workpiece has to be located such that the robot can reach all points. Finding such a location is still an iterative procedure based on the developers' intuition. One difficulty is the choice of one of the several solutions of the backward transform of a typical 6R robot. We present a novel algorithm that simultaneously optimizes the workpiece location and the robot configuration at all process points using higher order optimization algorithms. A key ingredient is the extension of the robot with a virtual prismatic axis. The practical feasibility of the approach is shown with an example using a commercial industrial robot.}, language = {en} } @inproceedings{WeissVolbert, author = {Weiß, Martin Georg and Volbert, Klaus}, title = {Intelligente Steuerung von Industrierobotern}, series = {Zweite OTH-Clusterkonferenz 18. Januar 2017 Techbase, Regensburg}, booktitle = {Zweite OTH-Clusterkonferenz 18. Januar 2017 Techbase, Regensburg}, editor = {Falter, Thomas}, address = {Regensburg}, organization = {OTH Regensburg und OTH Amberg-Weiden}, language = {de} } @inproceedings{Weiss, author = {Weiß, Martin Georg}, title = {Optimal Object Placement Using a Virtual Axis}, series = {Advances in Robot Kinematics, ARK 2018}, volume = {8}, booktitle = {Advances in Robot Kinematics, ARK 2018}, editor = {Lenarcic, Jadran and Parenti-Castelli, Vincenzo}, publisher = {Springer International Publishing}, address = {Cham}, isbn = {978-3-319-93187-6}, doi = {10.1007/978-3-319-93188-3_14}, pages = {116 -- 123}, abstract = {A basic task in the design of a robotic production cell is the relative placement of robot and workpiece. The fundamental requirement is that the robot can reach all process positions; only then one can think further optimization. Therefore an algorithm that automatically places an object into the workspace is very desirable. However many iterative optimization algorithms cannot guarantee that all intermediate steps are reachable, resulting in complicated procedures. We present a novel approach which extends a robot by a virtual prismatic joint - which measures the distance to the workspace - such that any TCP frames are reachable. This allows higher order nonlinear programming algorithms to be used for placement of an object alone as well as the optimal placement under some differentiable criterion.}, language = {en} } @misc{MauererRexhepajMonkmanetal., author = {Mauerer, Wolfgang and Rexhepaj, Tanja and Monkman, Gareth J. and Sindersberger, Dirk and Diermeier, Andreas and Neidhart, Thomas and Wolfrum, Dominik and Sterner, Michael and Heberl, Michael and Nusko, Robert and Maier, Georg and Nagl, Klaus and Reuter, Monika and Hofrichter, Andreas and Lex, Thomas and Lesch, Florian and Kieninger, B{\"a}rbel and Szalo, Alexander Eduard and Zehner, Alexander and Palm, Christoph and Joblin, Mitchell and Apel, Sven and Ramsauer, Ralf and Lohmann, Daniel and Westner, Markus and Strasser, Artur and Munndi, Maximilian and Ebner, Lena and Elsner, Michael and Weiß, Nils and Segerer, Matthias and Hackenberg, Rudolf and Steger, Sebastian and Schmailzl, Anton and Dostalek, Michael and Armbruster, Dominik and Koch, Fabian and Hierl, Stefan and Thumann, Philipp and Swidergal, Krzysztof and Wagner, Marcus and Briem, Ulrich and Diermeier, Andreas and Spreitzer, Stefan and Beiderbeck, Sabrina and Hook, Christian and Zobel, Martin and Weber, Tim and Groß, Simon and Penzkofer, Rainer and Dendorfer, Sebastian and Schillitz, Ingo and Bauer, Thomas and Rudolph, Clarissa and Schmidt, Katja and Liebetruth, Thomas and Hamer, Markus and Haug, Sonja and Vernim, Matthias and Weber, Karsten and Saßmannshausen, Sean Patrick and Books, Sebastian and Neuleitner, Nikolaus and Rechenauer, Christian and Steffens, Oliver and Kusterle, Wolfgang and G{\"o}mmel, Roland and Wellnitz, Felix and Stierstorfer, Johannes and Stadler, Dominik and Hofmann, Matthias J. and Motschmann, Hubert and Shamonin (Chamonine), Mikhail and Bleicher, Veronika and Fischer, Sebastian and Hackenberg, Rudolf and Horn, Anton and Kawasch, Raphael and Petzenhauser, Michael and Probst, Tobias and Udalzow, Anton and Dams, Florian and Schreiner, Rupert and Langer, Christoph and Prommesberger, Christian and Ławrowski, Robert Damian}, title = {Forschungsbericht 2016}, editor = {Baier, Wolfgang}, address = {Regensburg}, organization = {Ostbayerische Technische Hochschule Regensburg}, doi = {10.35096/othr/pub-1384}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:898-opus4-13840}, language = {de} }