Refine
H-BRS Bibliography
- yes (979)
Departments, institutes and facilities
- Fachbereich Informatik (979) (remove)
Document Type
- Conference Object (571)
- Article (221)
- Preprint (50)
- Report (41)
- Part of a Book (31)
- Doctoral Thesis (20)
- Conference Proceedings (14)
- Research Data (11)
- Book (monograph, edited volume) (9)
- Master's Thesis (6)
Year of publication
Language
- English (979) (remove)
Keywords
- Virtual Reality (13)
- Robotics (12)
- Machine Learning (10)
- virtual reality (10)
- 3D user interface (7)
- Quality diversity (7)
- Augmented Reality (6)
- Robotik (6)
- Usable Security (6)
- Benchmarking (5)
Is It Really You Who Forgot the Password? When Account Recovery Meets Risk-Based Authentication
(2024)
Force field (FF) based molecular modeling is an often used method to investigate and study structural and dynamic properties of (bio-)chemical substances and systems. When such a system is modeled or refined, the force field parameters need to be adjusted. This force field parameter optimization can be a tedious task and is always a trade-off in terms of errors regarding the targeted properties. To better control the balance of various properties’ errors, in this study we introduce weighting factors for the optimization objectives. Different weighting strategies are compared to fine-tune the balance between bulk-phase density and relative conformational energies (RCE), using n-octane as a representative system. Additionally, a non-linear projection of the individual property-specific parts of the optimized loss function is deployed to further improve the balance between them. The results show that the overall error is reduced. One interesting outcome is a large variety in the resulting optimized force field parameters (FFParams) and corresponding errors, suggesting that the optimization landscape is multi-modal and very dependent on the weighting factor setup. We conclude that adjusting the weighting factors can be a very important feature to lower the overall error in the FF optimization procedure, giving researchers the possibility to fine-tune their FFs.
Self-motion perception is a multi-sensory process that involves visual, vestibular, and other cues. When perception of self-motion is induced using only visual motion, vestibular cues indicate that the body remains stationary, which may bias an observer’s perception. When lowering the precision of the vestibular cue by for example, lying down or by adapting to microgravity, these biases may decrease, accompanied by a decrease in precision. To test this hypothesis, we used a move-to-target task in virtual reality. Astronauts and Earth-based controls were shown a target at a range of simulated distances. After the target disappeared, forward self-motion was induced by optic flow. Participants indicated when they thought they had arrived at the target’s previously seen location. Astronauts completed the task on Earth (supine and sitting upright) prior to space travel, early and late in space, and early and late after landing. Controls completed the experiment on Earth using a similar regime with a supine posture used to simulate being in space. While variability was similar across all conditions, the supine posture led to significantly higher gains (target distance/perceived travel distance) than the sitting posture for the astronauts pre-flight and early post-flight but not late post-flight. No difference was detected between the astronauts’ performance on Earth and onboard the ISS, indicating that judgments of traveled distance were largely unaffected by long-term exposure to microgravity. Overall, this constitutes mixed evidence as to whether non-visual cues to travel distance are integrated with relevant visual cues when self-motion is simulated using optic flow alone.
During robot-assisted therapy, a robot typically needs to be partially or fully controlled by therapists, for instance using a Wizard-of-Oz protocol; this makes therapeutic sessions tedious to conduct, as therapists cannot fully focus on the interaction with the person under therapy. In this work, we develop a learning-based behaviour model that can be used to increase the autonomy of a robot’s decision-making process. We investigate reinforcement learning as a model training technique and compare different reward functions that consider a user’s engagement and activity performance. We also analyse various strategies that aim to make the learning process more tractable, namely i) behaviour model training with a learned user model, ii) policy transfer between user groups, and iii) policy learning from expert feedback. We demonstrate that policy transfer can significantly speed up the policy learning process, although the reward function has an important effect on the actions that a robot can choose. Although the main focus of this paper is the personalisation pipeline itself, we further evaluate the learned behaviour models in a small-scale real-world feasibility study in which six users participated in a sequence learning game with an assistive robot. The results of this study seem to suggest that learning from guidance may result in the most adequate policies in terms of increasing the engagement and game performance of users, but a large-scale user study is needed to verify the validity of that observation.
Selection Performance and Reliability of Eye and Head Gaze Tracking Under Varying Light Conditions
(2024)
Due to their user-friendliness and reliability, biometric systems have taken a central role in everyday digital identity management for all kinds of private, financial and governmental applications with increasing security requirements. A central security aspect of unsupervised biometric authentication systems is the presentation attack detection (PAD) mechanism, which defines the robustness to fake or altered biometric features. Artifacts like photos, artificial fingers, face masks and fake iris contact lenses are a general security threat for all biometric modalities. The Biometric Evaluation Center of the Institute of Safety and Security Research (ISF) at the University of Applied Sciences Bonn-Rhein-Sieg has specialized in the development of a near-infrared (NIR)-based contact-less detection technology that can distinguish between human skin and most artifact materials. This technology is highly adaptable and has already been successfully integrated into fingerprint scanners, face recognition devices and hand vein scanners. In this work, we introduce a cutting-edge, miniaturized near-infrared presentation attack detection (NIR-PAD) device. It includes an innovative signal processing chain and an integrated distance measurement feature to boost both reliability and resilience. We detail the device’s modular configuration and conceptual decisions, highlighting its suitability as a versatile platform for sensor fusion and seamless integration into future biometric systems. This paper elucidates the technological foundations and conceptual framework of the NIR-PAD reference platform, alongside an exploration of its potential applications and prospective enhancements.
This work proposes a novel approach for probabilistic end-to-end all-sky imager-based nowcasting with horizons of up to 30 min using an ImageNet pre-trained deep neural network. The method involves a two-stage approach. First, a backbone model is trained to estimate the irradiance from all-sky imager (ASI) images. The model is then extended and retrained on image and parameter sequences for forecasting. An open access data set is used for training and evaluation. We investigated the impact of simultaneously considering global horizontal (GHI), direct normal (DNI), and diffuse horizontal irradiance (DHI) on training time and forecast performance as well as the effect of adding parameters describing the irradiance variability proposed in the literature. The backbone model estimates current GHI with an RMSE and MAE of 58.06 and 29.33 W m−2, respectively. When extended for forecasting, the model achieves an overall positive skill score reaching 18.6 % compared to a smart persistence forecast. Minor modifications to the deterministic backbone and forecasting models enables the architecture to output an asymmetrical probability distribution and reduces training time while leading to similar errors for the backbone models. Investigating the impact of variability parameters shows that they reduce training time but have no significant impact on the GHI forecasting performance for both deterministic and probabilistic forecasting while simultaneously forecasting GHI, DNI, and DHI reduces the forecast performance.
In vision tasks, a larger effective receptive field (ERF) is associated with better performance. While attention natively supports global context, convolution requires multiple stacked layers and a hierarchical structure for large context. In this work, we extend Hyena, a convolution-based attention replacement, from causal sequences to the non-causal two-dimensional image space. We scale the Hyena convolution kernels beyond the feature map size up to 191$\times$191 to maximize the ERF while maintaining sub-quadratic complexity in the number of pixels. We integrate our two-dimensional Hyena, HyenaPixel, and bidirectional Hyena into the MetaFormer framework. For image categorization, HyenaPixel and bidirectional Hyena achieve a competitive ImageNet-1k top-1 accuracy of 83.0% and 83.5%, respectively, while outperforming other large-kernel networks. Combining HyenaPixel with attention further increases accuracy to 83.6%. We attribute the success of attention to the lack of spatial bias in later stages and support this finding with bidirectional Hyena.
While humans can effortlessly pick a view from multiple streams, automatically choosing the best view is a challenge. Choosing the best view from multi-camera streams poses a problem regarding which objective metrics should be considered. Existing works on view selection lack consensus about which metrics should be considered to select the best view. The literature on view selection describes diverse possible metrics. And strategies such as information-theoretic, instructional design, or aesthetics-motivated fail to incorporate all approaches. In this work, we postulate a strategy incorporating information-theoretic and instructional design-based objective metrics to select the best view from a set of views. Traditionally, information-theoretic measures have been used to find the goodness of a view, such as in 3D rendering. We adapted a similar measure known as the viewpoint entropy for real-world 2D images. Additionally, we incorporated similarity penalization to get a more accurate measure of the entropy of a view, which is one of the metrics for the best view selection. Since the choice of the best view is domain-dependent, we chose demonstration-based training scenarios as our use case. The limitation of our chosen scenarios is that they do not include collaborative training and solely feature a single trainer. To incorporate instructional design considerations, we included the trainer’s body pose, face, face when instructing, and hands visibility as metrics. To incorporate domain knowledge we included predetermined regions’ visibility as another metric. All of those metrics are taken into account to produce a parameterized view recommendation approach for demonstration-based training. An online study using recorded multi-camera video streams from a simulation environment was used to validate those metrics. Furthermore, the responses from the online study were used to optimize the view recommendation performance with a normalized discounted cumulative gain (NDCG) value of 0.912, which shows good performance with respect to matching user choices.
Although climate-induced liquidity risks can cause significant disruptions and instabilities in the financial sector, they are frequently overlooked in current debates and policy discussions. This paper proposes a macro-financial agent-based integrated assessment model to investigate the transmission channels of climate risks to financial instability and study the emergence of liquidity crises through interbank market dynamics. Our simulations show that the financial system could experience serious funding and market liquidity shortages due to climate-induced liquidity crises. Our investigation contributes to our understanding of the impact - and possible solutions - to climate-induced liquidity crises, besides the issue of asset stranding related to transition risks usually considered in the existing studies.
A PM2.5 concentration prediction framework with vehicle tracking system: From cause to effect
(2023)
Representing 3D surfaces as level sets of continuous functions over R3 is the common denominator of neural implicit representations, which recently enabled remarkable progress in geometric deep learning and computer vision tasks. In order to represent 3D motion within this framework, it is often assumed (either explicitly or implicitly) that the transformations which a surface may undergo are homeomorphic: this is not necessarily true, for instance, in the case of fluid dynamics. In order to represent more general classes of deformations, we propose to apply this theoretical framework as regularizers for the optimization of simple 4D implicit functions (such as signed distance fields). We show that our representation is capable of capturing both homeomorphic and topology-changing deformations, while also defining correspondences over the continuously-reconstructed surfaces.
Neuromorphic computing aims to mimic the computational principles of the brain in silico and has motivated research into event-based vision and spiking neural networks (SNNs). Event cameras (ECs) capture local, independent changes in brightness, and offer superior power consumption, response latencies, and dynamic ranges compared to frame-based cameras. SNNs replicate neuronal dynamics observed in biological neurons and propagate information in sparse sequences of ”spikes”. Apart from biological fidelity, SNNs have demonstrated potential as an alternative to conventional artificial neural networks (ANNs), such as in reducing energy expenditure and inference time in visual classification. Although potentially beneficial for robotics, the novel event-driven and spike-based paradigms remain scarcely explored outside the domain of aerial robots.
To investigate the utility of brain-inspired sensing and data processing in a robotics application, we developed a neuromorphic approach to real-time, online obstacle avoidance on a manipulator with an onboard camera. Our approach adapts high-level trajectory plans with reactive maneuvers by processing emulated event data in a convolutional SNN, decoding neural activations into avoidance motions, and adjusting plans in a dynamic motion primitive formulation. We conducted simulated and real experiments with a Kinova Gen3 arm performing simple reaching tasks involving static and dynamic obstacles. Our implementation was systematically tuned, validated, and tested in sets of distinct task scenarios, and compared to a non-adaptive baseline through formalized quantitative metrics and qualitative criteria.
The neuromorphic implementation facilitated reliable avoidance of imminent collisions in most scenarios, with 84% and 92% median success rates in simulated and real experiments, where the baseline consistently failed. Adapted trajectories were qualitatively similar to baseline trajectories, indicating low impacts on safety, predictability and smoothness criteria. Among notable properties of the SNN were the correlation of processing time with the magnitude of perceived motions (captured in events) and robustness to different event emulation methods. Preliminary tests with a DAVIS346 EC showed similar performance, validating our experimental event emulation method. These results motivate future efforts to incorporate SNN learning, utilize neuromorphic processors, and target other robot tasks to further explore this approach.
Skill generalisation and experience acquisition for predicting and avoiding execution failures
(2023)
For performing tasks in their target environments, autonomous robots usually execute and combine skills. Robot skills in general and learning-based skills in particular are usually designed so that flexible skill acquisition is possible, but without an explicit consideration of execution failures, the impact that failure analysis can have on the skill learning process, or the benefits of introspection for effective coexistence with humans. Particularly in human-centered environments, the ability to understand, explain, and appropriately react to failures can affect a robot's trustworthiness and, consequently, its overall acceptability. Thus, in this dissertation, we study the questions of how parameterised skills can be designed so that execution-level decisions are associated with semantic knowledge about the execution process, and how such knowledge can be utilised for avoiding and analysing execution failures. The first major segment of this work is dedicated to developing a representation for skill parameterisation whose objective is to improve the transparency of the skill parameterisation process and enable a semantic analysis of execution failures. We particularly develop a hybrid learning-based representation for parameterising skills, called an execution model, which combines qualitative success preconditions with a function that maps parameters to predicted execution success. The second major part of this work focuses on applications of the execution model representation to address different types of execution failures. We first present a diagnosis algorithm that, given parameters that have resulted in a failure, finds a failure hypothesis by searching for violations of the qualitative model, as well as an experience correction algorithm that uses the found hypothesis to identify parameters that are likely to correct the failure. Furthermore, we present an extension of execution models that allows multiple qualitative execution contexts to be considered so that context-specific execution failures can be avoided. Finally, to enable the avoidance of model generalisation failures, we propose an adaptive ontology-assisted strategy for execution model generalisation between object categories that aims to combine the benefits of model-based and data-driven methods; for this, information about category similarities as encoded in an ontology is integrated with outcomes of model generalisation attempts performed by a robot. The proposed methods are exemplified in terms of various use cases - object and handle grasping, object stowing, pulling, and hand-over - and evaluated in multiple experiments performed with a physical robot. The main contributions of this work include a formalisation of the skill parameterisation problem by considering execution failures as an integral part of the skill design and learning process, a demonstration of how a hybrid representation for parameterising skills can contribute towards improving the introspective properties of robot skills, as well as an extensive evaluation of the proposed methods in various experiments. We believe that this work constitutes a small first step towards more failure-aware robots that are suitable to be used in human-centered environments.
Loading of shipping containers for dairy products often includes a press-fit task, which involves manually stacking milk cartons in a container without using pallets or packaging. Automating this task with a mobile manipulator can reduce worker strain, and also enhance the efficiency and safety of the container loading process. This paper proposes an approach called Adaptive Compliant Control with Integrated Failure Recovery (ACCIFR), which enables a mobile manipulator to reliably perform the press-fit task. We base the approach on a demonstration learning-based compliant control framework, such that we integrate a monitoring and failure recovery mechanism for successful task execution. Concretely, we monitor the execution through distance and force feedback, detect collisions while the robot is performing the press-fit task, and use wrench measurements to classify the direction of collision; this information informs the subsequent recovery process. We evaluate the method on a miniature container setup, considering variations in the (i) starting position of the end effector, (ii) goal configuration, and (iii) object grasping position. The results demonstrate that the proposed approach outperforms the baseline demonstration-based learning framework regarding adaptability to environmental variations and the ability to recover from collision failures, making it a promising solution for practical press-fit applications.
In the design of robot skills, the focus generally lies on increasing the flexibility and reliability of the robot execution process; however, typical skill representations are not designed for analysing execution failures if they occur or for explicitly learning from failures. In this paper, we describe a learning-based hybrid representation for skill parameterisation called an execution model, which considers execution failures to be a natural part of the execution process. We then (i) demonstrate how execution contexts can be included in execution models, (ii) introduce a technique for generalising models between object categories by combining generalisation attempts performed by a robot with knowledge about object similarities represented in an ontology, and (iii) describe a procedure that uses an execution model for identifying a likely hypothesis of a parameterisation failure. The feasibility of the proposed methods is evaluated in multiple experiments performed with a physical robot in the context of handle grasping, object grasping, and object pulling. The experimental results suggest that execution models contribute towards avoiding execution failures, but also represent a first step towards more introspective robots that are able to analyse some of their execution failures in an explicit manner.
Saliency methods are frequently used to explain Deep Neural Network-based models. Adebayo et al.'s work on evaluating saliency methods for classification models illustrate certain explanation methods fail the model and data randomization tests. However, on extending the tests for various state of the art object detectors we illustrate that the ability to explain a model is more dependent on the model itself than the explanation method. We perform sanity checks for object detection and define new qualitative criteria to evaluate the saliency explanations, both for object classification and bounding box decisions, using Guided Backpropagation, Integrated Gradients, and their Smoothgrad versions, together with Faster R-CNN, SSD, and EfficientDet-D0, trained on COCO. In addition, the sensitivity of the explanation method to model parameters and data labels varies class-wise motivating to perform the sanity checks for each class. We find that EfficientDet-D0 is the most interpretable method independent of the saliency method, which passes the sanity checks with little problems.