Technische Universität Darmstadt

About: Technische Universität Darmstadt is a education organization based out in Darmstadt, Germany. It is known for research contribution in the topics: Computer science & Context (language use). The organization has 17316 authors who have published 40619 publications receiving 937916 citations. The organization is also known as: Darmstadt University of Technology & University of Darmstadt.

Parallelization Strategies for Ant Colony Optimization

Abstract: Ant Colony Optimization (ACO) is a new population oriented search metaphor that has been successfully applied to NP-hard combinatorial optimization problems. In this paper we discuss parallelization strategies for Ant Colony Optimization algorithms. We empirically test the most simple strategy, that of executing parallel independent runs of an algorithm. The empirical tests are performed applying MAX-MIN Ant System, one of the most efficient ACO algorithms, to the Traveling Salesman Problem and show that using parallel independent runs is very effective.

Production of the entire range of r-process nuclides by black hole accretion disc outflows from neutron star mergers

Abstract: We consider $r$-process nucleosynthesis in outflows from black hole accretion discs formed in double neutron star and neutron star -- black hole mergers. These outflows, powered by angular momentum transport processes and nuclear recombination, represent an important -- and in some cases dominant -- contribution to the total mass ejected by the merger. Here we calculate the nucleosynthesis yields from disc outflows using thermodynamic trajectories from hydrodynamic simulations, coupled to a nuclear reaction network. We find that outflows produce a robust abundance pattern around the second $r$-process peak (mass number $A \sim 130$), independent of model parameters, with significant production of $A < 130$ nuclei. This implies that dynamical ejecta with high electron fraction may not be required to explain the observed abundances of $r$-process elements in metal poor stars. Disc outflows reach the third peak ($ A \sim 195$) in most of our simulations, although the amounts produced depend sensitively on the disc viscosity, initial mass or entropy of the torus, and nuclear physics inputs. Some of our models produce an abundance spike at $A = 132$ that is absent in the Solar system $r$-process distribution. The spike arises from convection in the disc and depends on the treatment of nuclear heating in the simulations. We conclude that disc outflows provide an important -- and perhaps dominant -- contribution to the $r$-process yields of compact binary mergers, and hence must be included when assessing the contribution of these systems to the inventory of $r$-process elements in the Galaxy.

MVE: a multi-view reconstruction environment

Abstract: We present MVE, the Multi-View Environment. MVE is an end-to-end multi-view geometry reconstruction software which takes photos of a scene as input and produces a surface triangle mesh as result. The system covers a structure-from-motion algorithm, multi-view stereo reconstruction, generation of extremely dense point clouds, and reconstruction of surfaces from point clouds. In contrast to most image-based geometry reconstruction approaches, our system is focused on reconstruction of multi-scale scenes, an important aspect in many areas such as cultural heritage. It allows to reconstruct large datasets containing some detailed regions with much higher resolution than the rest of the scene. Our system provides a graphical user interface for structure-from-motion reconstruction, visual inspection of images, depth maps, and rendering of scenes and meshes.

Reinforcement learning to adjust parametrized motor primitives to new situations

Abstract: Humans manage to adapt learned movements very quickly to new situations by generalizing learned behaviors from similar situations. In contrast, robots currently often need to re-learn the complete movement. In this chapter, we propose a method that learns to generalize parametrized motor plans by adapting a small set of global parameters, called meta-parameters. We employ reinforcement learning to learn the required meta-parameters to deal with the current situation, described by states. We introduce an appropriate reinforcement learning algorithm based on a kernelized version of the reward-weighted regression. To show its feasibility, we evaluate this algorithm on a toy example and compare it to several previous approaches. Subsequently, we apply the approach to three robot tasks, i.e., the generalization of throwing movements in darts, of hitting movements in table tennis, and of throwing balls where the tasks are learned on several different real physical robots, i.e., a Barrett WAM, a BioRob, the JST-ICORP/SARCOS CBi and a Kuka KR 6.

VARS: a vehicle ad-hoc network reputation system

Abstract: Using mobile ad-hoc networks in an automotive environment (VANET) opens a new set of applications, such as the distribution of information about local traffic or road conditions. This can increase traffic safety and improve mobility. One of the main challenges is to forward event related messages in such a way that the information can be trusted by receiving nodes. Authentication does not solve the problem as it does not target the quality of messages. One promising solution might be given by reputation systems. However, conventional centralized trust establishment approaches are not well suited for use within distributed networks such as those envisioned for automotive scenarios. Therefore, we present VARS, a completely distributed approach based on reputation. Our work is based on the following assumptions: cars move at a high average speed; VANETs may become very large, in the order of thousands or even millions of nodes, so that authenticated identities are not feasible; a solution has to be completely decentralized; available bandwidth for communication remains limited, while processing power and memory continue to increase. We introduce major architecture concepts that enable VARS to operate efficiently in the given environment, present the most relevant algorithms and provide some simulation results. First results show that our scheme works in general.