004 Datenverarbeitung; Informatik
Refine
H-BRS Bibliography
- yes (57)
Departments, institutes and facilities
- Fachbereich Informatik (51)
- Institute of Visual Computing (IVC) (9)
- Institut für Cyber Security & Privacy (ICSP) (8)
- Graduierteninstitut (4)
- Institut für Technik, Ressourcenschonung und Energieeffizienz (TREE) (4)
- Fachbereich Wirtschaftswissenschaften (2)
- Institut für IT-Service (ITS) (1)
- Institut für Verbraucherinformatik (IVI) (1)
- Institut für funktionale Gen-Analytik (IFGA) (1)
Document Type
- Conference Object (29)
- Article (7)
- Preprint (5)
- Doctoral Thesis (4)
- Part of a Book (3)
- Research Data (3)
- Report (3)
- Book (monograph, edited volume) (2)
- Conference Proceedings (1)
Year of publication
- 2021 (57) (remove)
Keywords
- Cognitive robot control (2)
- Dimensionality reduction (2)
- Explainable robotics (2)
- Generative Models (2)
- Human-Computer Interaction (2)
- Learning from experience (2)
- LoRa (2)
- LoRaWAN (2)
- Low-Power Wide Area Network (LP-WAN) (2)
- Measurement (2)
Machine learning and neural networks are now ubiquitous in sonar perception, but it lags behind the computer vision field due to the lack of data and pre-trained models specifically for sonar images. In this paper we present the Marine Debris Turntable dataset and produce pre-trained neural networks trained on this dataset, meant to fill the gap of missing pre-trained models for sonar images. We train Resnet 20, MobileNets, DenseNet121, SqueezeNet, MiniXception, and an Autoencoder, over several input image sizes, from 32 x 32 to 96 x 96, on the Marine Debris turntable dataset. We evaluate these models using transfer learning for low-shot classification in the Marine Debris Watertank and another dataset captured using a Gemini 720i sonar. Our results show that in both datasets the pre-trained models produce good features that allow good classification accuracy with low samples (10-30 samples per class). The Gemini dataset validates that the features transfer to other kinds of sonar sensors. We expect that the community benefits from the public release of our pre-trained models and the turntable dataset.
The dataset contains the following data from successful and failed executions of the Toyota HSR robot placing a book on a shelf.
RGB images from the robot's head camera
Depth images from the robot's head camera
Rendered images of the robot's 3D model from the point of view of the robot's head camera
Force-torque readings from a wrist-mounted force-torque sensor
Joint efforts, velocities and positions
extrinsic and intrinsic camera calibration parameters
frame-level anomaly annotations
The anomalies that occur during execution include:
the manipulated book falling down
books on the shelf being disturbed significantly
camera occlusions
robot being disturbed by an external collision
The dataset is split into a train, validation and test set with the following number of trials:
Train: 48 successful trials
Validation: 6 successful trials
Test: 60 anomalous trials and 7 successful trials
Execution monitoring is essential for robots to detect and respond to failures. Since it is impossible to enumerate all failures for a given task, we learn from successful executions of the task to detect visual anomalies during runtime. Our method learns to predict the motions that occur during the nominal execution of a task, including camera and robot body motion. A probabilistic U-Net architecture is used to learn to predict optical flow, and the robot's kinematics and 3D model are used to model camera and body motion. The errors between the observed and predicted motion are used to calculate an anomaly score. We evaluate our method on a dataset of a robot placing a book on a shelf, which includes anomalies such as falling books, camera occlusions, and robot disturbances. We find that modeling camera and body motion, in addition to the learning-based optical flow prediction, results in an improvement of the area under the receiver operating characteristic curve from 0.752 to 0.804, and the area under the precision-recall curve from 0.467 to 0.549.
In Robot-Assisted Therapy for children with Autism Spectrum Disorder, the therapists’ workload is increased due to the necessity of controlling the robot manually. The solution for this problem is to increase the level of autonomy of the system, namely the robot should interpret and adapt to the behaviour of the child under therapy. The problem that we are adressing is to develop a behaviour model that will be used for the robot decision-making process, which will learn how to adequately react to certain child reactions. We propose the use of the reinforcement learning technique for this task, where feedback for learning is obtained from the therapist’s evaluation of a robot’s behaviour.
Property-Based Testing in Simulation for Verifying Robot Action Execution in Tabletop Manipulation
(2021)
An important prerequisite for the reliability and robustness of a service robot is ensuring the robot’s correct behavior when it performs various tasks of interest. Extensive testing is one established approach for ensuring behavioural correctness; this becomes even more important with the integration of learning-based methods into robot software architectures, as there are often no theoretical guarantees about the performance of such methods in varying scenarios. In this paper, we aim towards evaluating the correctness of robot behaviors in tabletop manipulation through automatic generation of simulated test scenarios in which a robot assesses its performance using property-based testing. In particular, key properties of interest for various robot actions are encoded in an action ontology and are then verified and validated within a simulated environment. We evaluate our framework with a Toyota Human Support Robot (HSR) which is tested in a Gazebo simulation. We show that our framework can correctly and consistently identify various failed actions in a variety of randomised tabletop manipulation scenarios, in addition to providing deeper insights into the type and location of failures for each designed property.
Characterization of Urban Radio Channels and Base Station Antenna Correlation in the 3.75 GHz Band
(2021)
Despite their age, ray-based rendering methods are still a very active field of research with many challenges when it comes to interactive visualization. In this thesis, we present our work on Guided High-Quality Rendering, Foveated Ray Tracing for Head Mounted Displays and Hash-based Hierarchical Caching and Layered Filtering. Our system for Guided High-Quality Rendering allows for guiding the sampling rate of ray-based rendering methods by a user-specified Region of Interest (RoI). We propose two interaction methods for setting such an RoI when using a large display system and a desktop display, respectively. This makes it possible to compute images with a heterogeneous sample distribution across the image plane. Using such a non-uniform sample distribution, the rendering performance inside the RoI can be significantly improved in order to judge specific image features. However, a modified scheduling method is required to achieve sufficient performance. To solve this issue, we developed a scheduling method based on sparse matrix compression, which has shown significant improvements in our benchmarks. By filtering the sparsely sampled image appropriately, large brightness variations in areas outside the RoI are avoided and the overall image brightness is similar to the ground truth early in the rendering process. When using ray-based methods in a VR environment on head-mounted display de vices, it is crucial to provide sufficient frame rates in order to reduce motion sickness. This is a challenging task when moving through highly complex environments and the full image has to be rendered for each frame. With our foveated rendering sys tem, we provide a perception-based method for adjusting the sample density to the user’s gaze, measured with an eye tracker integrated into the HMD. In order to avoid disturbances through visual artifacts from low sampling rates, we introduce a reprojection-based rendering pipeline that allows for fast rendering and temporal accumulation of the sparsely placed samples. In our user study, we analyse the im pact our system has on visual quality. We then take a closer look at the recorded eye tracking data in order to determine tracking accuracy and connections between different fixation modes and perceived quality, leading to surprising insights. For previewing global illumination of a scene interactively by allowing for free scene exploration, we present a hash-based caching system. Building upon the concept of linkless octrees, which allow for constant-time queries of spatial data, our frame work is suited for rendering such previews of static scenes. Non-diffuse surfaces are supported by our hybrid reconstruction approach that allows for the visualization of view-dependent effects. In addition to our caching and reconstruction technique, we introduce a novel layered filtering framework, acting as a hybrid method between path space and image space filtering, that allows for the high-quality denoising of non-diffuse materials. Also, being designed as a framework instead of a concrete filtering method, it is possible to adapt most available denoising methods to our layered approach instead of relying only on the filtering of primary hitpoints.
Urban LoRa networks promise to provide a cost-efficient and scalable communication backbone for smart cities. One core challenge in rolling out and operating these networks is radio network planning, i.e., precise predictions about possible new locations and their impact on network coverage. Path loss models aid in this task, but evaluating and comparing different models requires a sufficiently large set of high-quality received packet power samples. In this paper, we report on a corresponding large-scale measurement study covering an urban area of 200km2 over a period of 230 days using sensors deployed on garbage trucks, resulting in more than 112 thousand high-quality samples for received packet power. Using this data, we compare eleven previously proposed path loss models and additionally provide new coefficients for the Log-distance model. Our results reveal that the Log-distance model and other well-known empirical models such as Okumura or Winner+ provide reasonable estimations in an urban environment, and terrain based models such as ITM or ITWOM have no advantages. In addition, we derive estimations for the needed sample size in similar measurement campaigns. To stimulate further research in this direction, we make all our data publicly available.
It has been well proved that deep networks are efficient at extracting features from a given (source) labeled dataset. However, it is not always the case that they can generalize well to other (target) datasets which very often have a different underlying distribution. In this report, we evaluate four different domain adaptation techniques for image classification tasks: DeepCORAL, DeepDomainConfusion, CDAN and CDAN+E. These techniques are unsupervised given that the target dataset dopes not carry any labels during training phase. We evaluate model performance on the office-31 dataset. A link to the github repository of this report can be found here: https://github.com/agrija9/Deep-Unsupervised-Domain-Adaptation.
Ice accumulation in the blades of wind turbines can cause them to describe anomalous rotations or no rotations at all, thus affecting the generation of electricity and power output. In this work, we investigate the problem of ice accumulation in wind turbines by framing it as anomaly detection of multi-variate time series. Our approach focuses on two main parts: first, learning low-dimensional representations of time series using a Variational Recurrent Autoencoder (VRAE), and second, using unsupervised clustering algorithms to classify the learned representations as normal (no ice accumulated) or abnormal (ice accumulated). We have evaluated our approach on a custom wind turbine time series dataset, for the two-classes problem (one normal versus one abnormal class), we obtained a classification accuracy of up to 96$\%$ on test data. For the multiple-class problem (one normal versus multiple abnormal classes), we present a qualitative analysis of the low-dimensional learned latent space, providing insights into the capacities of our approach to tackle such problem. The code to reproduce this work can be found here https://github.com/agrija9/Wind-Turbines-VRAE-Paper.