-
CNN-based Game State Detection for a Foosball Table
Authors:
David Hagens,
Jan M. Knaup,
Elke Hergenröther,
Andreas Weinmann
Abstract:
The automation of games using Deep Reinforcement Learning Strategies (DRL) is a well-known challenge in AI research. While for feature extraction in a video game typically the whole image is used, this is hardly practical for many real world games. Instead, using a smaller game state reducing the dimension of the parameter space to include essential parameters only seems to be a promising approach…
▽ More
The automation of games using Deep Reinforcement Learning Strategies (DRL) is a well-known challenge in AI research. While for feature extraction in a video game typically the whole image is used, this is hardly practical for many real world games. Instead, using a smaller game state reducing the dimension of the parameter space to include essential parameters only seems to be a promising approach. In the game of Foosball, a compact and comprehensive game state description consists of the positional shifts and rotations of the figures and the position of the ball over time. In particular, velocities and accelerations can be derived from consecutive time samples of the game state. In this paper, a figure detection system to determine the game state in Foosball is presented. We capture a dataset containing the rotations of the rods which were measured using accelerometers and the positional shifts were derived using traditional Computer Vision techniques (in a laboratory setting). This dataset is utilized to train Convolutional Neural Network (CNN) based end-to-end regression models to predict the rotations and shifts of each rod. We present an evaluation of our system using different state-of-the-art CNNs as base architectures for the regression model. We show that our system is able to predict the game state with high accuracy. By providing data for both black and white teams, the presented system is intended to provide the required data for future developments of Imitation Learning techniques w.r.t. to observing human players.
△ Less
Submitted 16 May, 2024; v1 submitted 8 April, 2024;
originally announced April 2024.
-
Deep Learning-Based Reconstruction of Interventional Tools from Four X-Ray Projections for Tomographic Interventional Guidance
Authors:
Elias Eulig,
Joscha Maier,
Michael Knaup,
N. Robert Bennett,
Klaus Hörndler,
Adam S. Wang,
Marc Kachelrieß
Abstract:
Image guidance for minimally invasive interventions is usually performed by acquiring fluoroscopic images using a C-arm system. However, the projective data provide only limited information about the spatial structure and position of interventional tools such as stents, guide wires or coils. In this work we propose a deep learning-based pipeline for real-time tomographic (four-dimensional) interve…
▽ More
Image guidance for minimally invasive interventions is usually performed by acquiring fluoroscopic images using a C-arm system. However, the projective data provide only limited information about the spatial structure and position of interventional tools such as stents, guide wires or coils. In this work we propose a deep learning-based pipeline for real-time tomographic (four-dimensional) interventional guidance at acceptable dose levels. In the first step, interventional tools are extracted from four cone-beam CT projections using a deep convolutional neural network (CNN). These projections are then reconstructed and fed into a second CNN, which maps this highly undersampled reconstruction to a segmentation of the interventional tools. Our pipeline is capable of reconstructing interventional tools from only four x-ray projections without the need for a patient prior with very high accuracy. Therefore, the proposed approach is capable of overcoming the drawbacks of today's interventional guidance and could enable the development of new minimally invasive radiological interventions by providing full spatiotemporal information about the interventional tools.
△ Less
Submitted 24 November, 2021; v1 submitted 23 September, 2020;
originally announced September 2020.
-
How the aggregation of oxygen vacancies in rutile based TiO$_{2-δ}$ phases causes memristive behavior
Authors:
Wolfgang Heckel,
Michael Wehlau,
Sascha B. Maisel,
Thomas Frauenheim,
Jan M. Knaup,
Stefan Müller
Abstract:
The results of a comprehensive and systematic ab-initio based ground-state search for the structural arrangement of oxygen vacancies in rutile phase TiO$_2$ provide new insights into their memristive properties. We find that O vacancies tend to form planar arrangements which relax into structures exhibiting metallic behavior. These meta-stable arrangements are structurally akin to, yet distinguish…
▽ More
The results of a comprehensive and systematic ab-initio based ground-state search for the structural arrangement of oxygen vacancies in rutile phase TiO$_2$ provide new insights into their memristive properties. We find that O vacancies tend to form planar arrangements which relax into structures exhibiting metallic behavior. These meta-stable arrangements are structurally akin to, yet distinguishable from the Magnéli phase. They exhibit a more pronounced metallic nature, but are energetically less favorable. Our results confirm a clear structure-property relationship between segregated oxygen vacancy arrangement and metallic behavior in reduced oxides.
△ Less
Submitted 20 November, 2015; v1 submitted 28 September, 2015;
originally announced September 2015.
-
A permutation invariant collective variable to track and drive vacancy dynamics in simulations of solids
Authors:
Jan M. Knaup,
Michael Wehlau,
Thomas Frauenheim
Abstract:
Vacancy dynamics in oxides are vital for understanding redox reactions and resulting memristive effects or catalytic activity. We present a method to track and drive vacancies which we apply to metadynamics simulation of oxygen vacancies (V$_{\mathrm{O}}$) in rutile, demonstrating its effectiveness. Using the density functional based tight binding method, it is possible to explore the free energy…
▽ More
Vacancy dynamics in oxides are vital for understanding redox reactions and resulting memristive effects or catalytic activity. We present a method to track and drive vacancies which we apply to metadynamics simulation of oxygen vacancies (V$_{\mathrm{O}}$) in rutile, demonstrating its effectiveness. Using the density functional based tight binding method, it is possible to explore the free energy hyperplane of oxygen vacancies in TiO$_{2}$. We show that the migration of V$_{\mathrm{O}}$ in TiO$_{2}$ is governed by the jump with the higherst number of realizations. Free energy profiles are consistent with minimum energy paths.
△ Less
Submitted 18 December, 2013; v1 submitted 15 July, 2013;
originally announced July 2013.
