Technical Report / Hochschule Bonn-Rhein-Sieg University of Applied Sciences. Department of Computer Science
Publisher: Dean Prof. Dr. Sascha Alda
Hochschule Bonn-Rhein-Sieg University of Applied Sciences, Department of Computer Science
Sankt Augustin, Germany
ISSN 1869-5272
Hochschule Bonn-Rhein-Sieg University of Applied Sciences, Department of Computer Science
Sankt Augustin, Germany
ISSN 1869-5272
Refine
H-BRS Bibliography
- yes (39)
Departments, institutes and facilities
Document Type
- Report (32)
- Master's Thesis (7)
Year of publication
Language
- English (39) (remove)
Has Fulltext
- yes (39)
Keywords
- Robotik (6)
- 3D-Scanner (2)
- Active Learning (2)
- Computer Vision (2)
- Deep Learning (2)
- Machine Learning (2)
- Object Detection (2)
- Virtuelle Realität (2)
- computer vision (2)
- 3D Segmentation (1)
02-2023
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.
01-2023
This thesis investigates the benefit of rubrics for grading short answers using an active learning mechanism. Automating short answer grading using Natural Language Processing (NLP) is one of the active research areas in the education domain. This could save time for the evaluator and invest more time in preparing for the lecture. Most of the research on short answer grading was treated as a similarity task between reference and student answers. However, grading based on reference answers does not account for partial grades and does not provide feedback. Also, the grading is automatic that tries to replace the evaluator. Hence, using rubrics for short answer grading with active learning eliminates the drawbacks mentioned earlier.
Initially, the proposed approach is evaluated on the Mohler dataset, popularly used to benchmark the methodology. This phase is used to determine the parameters for the proposed approach. Therefore, the approach with the selected parameter exceeds the performance of current State-Of-The-Art (SOTA) methods resulting in the Pearson correlation value of 0.63 and Root Mean Square Error (RMSE) of 0.85. The proposed approach has surpassed the SOTA methods by almost 4%.
Finally, the benchmarked approach is used to grade the short answer based on rubrics instead of reference answers. The proposed approach evaluates short answers from Autonomous Mobile Robot (AMR) dataset to provide scores and feedback (formative assessment) based on the rubrics. The average performance of the dataset results in the Pearson correlation value of 0.61 and RMSE of 0.83. Thus, this research has proven that rubrics-based grading achieves formative assessment without compromising performance. In addition, the rubrics have the advantage of generalizability to all answers.
02-2022
Recent advances in Natural Language Processing have substantially improved contextualized representations of language. However, the inclusion of factual knowledge, particularly in the biomedical domain, remains challenging. Hence, many Language Models (LMs) are extended by Knowledge Graphs (KGs), but most approaches require entity linking (i.e., explicit alignment between text and KG entities). Inspired by single-stream multimodal Transformers operating on text, image and video data, this thesis proposes the Sophisticated Transformer trained on biomedical text and Knowledge Graphs (STonKGs). STonKGs incorporates a novel multimodal architecture based on a cross encoder that uses the attention mechanism on a concatenation of input sequences derived from text and KG triples, respectively. Over 13 million so-called text-triple pairs, coming from PubMed and assembled using the Integrated Network and Dynamical Reasoning Assembler (INDRA), were used in an unsupervised pre-training procedure to learn representations of biomedical knowledge in STonKGs. By comparing STonKGs to an NLP- and a KG-baseline (operating on either text or KG data) on a benchmark consisting of eight fine-tuning tasks, the proposed knowledge integration method applied in STonKGs was empirically validated. Specifically, on tasks with a comparatively small dataset size and a larger number of classes, STonKGs resulted in considerable performance gains, beating the F1-score of the best baseline by up to 0.083. Both the source code as well as the code used to implement STonKGs are made publicly available so that the proposed method of this thesis can be extended to many other biomedical applications.
01-2022
Effective Neighborhood Feature Exploitation in Graph CNNs for Point Cloud Object-Part Segmentation
(2022)
Part segmentation is the task of semantic segmentation applied on objects and carries a wide range of applications from robotic manipulation to medical imaging. This work deals with the problem of part segmentation on raw, unordered point clouds of 3D objects. While pioneering works on deep learning for point clouds typically ignore taking advantage of local geometric structure around individual points, the subsequent methods proposed to extract features by exploiting local geometry have not yielded significant improvements either. In order to investigate further, a graph convolutional network (GCN) is used in this work in an attempt to increase the effectiveness of such neighborhood feature exploitation approaches. Most of the previous works also focus only on segmenting complete point cloud data. Considering the impracticality of such approaches, taking into consideration the real world scenarios where complete point clouds are scarcely available, this work proposes approaches to deal with partial point cloud segmentation.
In the attempt to better capture neighborhood features, this work proposes a novel method to learn regional part descriptors which guide and refine the segmentation predictions. The proposed approach helps the network achieve state-of-the-art performance of 86.4% mIoU on the ShapeNetPart dataset for methods which do not use any preprocessing techniques or voting strategies. In order to better deal with partial point clouds, this work also proposes new strategies to train and test on partial data. While achieving significant improvements compared to the baseline performance, the problem of partial point cloud segmentation is also viewed through an alternate lens of semantic shape completion.
Semantic shape completion networks not only help deal with partial point cloud segmentation but also enrich the information captured by the system by predicting complete point clouds with corresponding semantic labels for each point. To this end, a new network architecture for semantic shape completion is also proposed based on point completion network (PCN) which takes advantage of a graph convolution based hierarchical decoder for completion as well as segmentation. In addition to predicting complete point clouds, results indicate that the network is capable of reaching within a margin of 5% to the mIoU performance of dedicated segmentation networks for partial point cloud segmentation.
03-2022
As cameras are ubiquitous in autonomous systems, object detection is a crucial task. Object detectors are widely used in applications such as autonomous driving, healthcare, and robotics. Given an image, an object detector outputs both the bounding box coordinates as well as classification probabilities for each object detected. The state-of-the-art detectors are treated as black boxes due to their highly non-linear internal computations. Even with unprecedented advancements in detector performance, the inability to explain how their outputs are generated limits their use in safety-critical applications in particular. It is therefore crucial to explain the reason behind each detector decision in order to gain user trust, enhance detector performance, and analyze their failure.
Previous work fails to explain as well as evaluate both bounding box and classification decisions individually for various detectors. Moreover, no tools explain each detector decision, evaluate the explanations, and also identify the reasons for detector failures. This restricts the flexibility to analyze detectors. The main contribution presented here is an open-source Detector Explanation Toolkit (DExT). It is used to explain the detector decisions, evaluate the explanations, and analyze detector errors. The detector decisions are explained visually by highlighting the image pixels that most influence a particular decision. The toolkit implements the proposed approach to generate a holistic explanation for all detector decisions using certain gradient-based explanation methods. To the author’s knowledge, this is the first work to conduct extensive qualitative and novel quantitative evaluations of different explanation methods across various detectors. The qualitative evaluation incorporates a visual analysis of the explanations carried out by the author as well as a human-centric evaluation. The human-centric evaluation includes a user study to understand user trust in the explanations generated across various explanation methods for different detectors. Four multi-object visualization methods are provided to merge the explanations of multiple objects detected in an image as well as the corresponding detector outputs in a single image. Finally, DExT implements the procedure to analyze detector failures using the formulated approach.
The visual analysis illustrates that the ability to explain a model is more dependent on the model itself than the actual ability of the explanation method. In addition, the explanations are affected by the object explained, the decision explained, detector architecture, training data labels, and model parameters. The results of the quantitative evaluation show that the Single Shot MultiBox Detector (SSD) is more faithfully explained compared to other detectors regardless of the explanation methods. In addition, a single explanation method cannot generate more faithful explanations than other methods for both the bounding box and the classification decision across different detectors. Both the quantitative and human-centric evaluations identify that SmoothGrad with Guided Backpropagation (GBP) provides more trustworthy explanations among selected methods across all detectors. Finally, a convex polygon-based multi-object visualization method provides more human-understandable visualization than other methods.
The author expects that DExT will motivate practitioners to evaluate object detectors from the interpretability perspective by explaining both bounding box and classification decisions.
04-2022
In the field of automatic music generation, one of the greatest challenges is the consistent generation of pieces continuously perceived positively by the majority of the audience since there is no objective method to determine the quality of a musical composition. However, composing principles, which have been refined for millennia, have shaped the core characteristics of today's music. A hybrid music generation system, mlmusic, that incorporates various static, music-theory-based methods, as well as data-driven, subsystems, is implemented to automatically generate pieces considered acceptable by the average listener. Initially, a MIDI dataset, consisting of over 100 hand-picked pieces of various styles and complexities, is analysed using basic music theory principles, and the abstracted information is fed into explicitly constrained LSTM networks. For chord progressions, each individual network is specifically trained on a given sequence length, while phrases are created by consecutively predicting the notes' offset, pitch and duration. Using these outputs as a composition's foundation, additional musical elements, along with constrained recurrent rhythmic and tonal patterns, are statically generated. Although no survey regarding the pieces' reception could be carried out, the successful generation of numerous compositions of varying complexities suggests that the integration of these fundamentally distinctive approaches might lead to success in other branches.
01-2020
An essential measure of autonomy in service robots designed to assist humans is adaptivity to the various contexts of human-oriented tasks. These robots may have to frequently execute the same action, but subject to subtle variations in task parameters that determine optimal behaviour. Such actions are traditionally executed by robots using pre-determined, generic motions, but a better approach could utilize robot arm maneuverability to learn and execute different trajectories that work best in each context.
In this project, we explore a robot skill acquisition procedure that allows incorporating contextual knowledge, adjusting executions according to context, and improvement through experience, as a step towards more adaptive service robots. We propose an apprenticeship learning approach to achieving context-aware action generalisation on the task of robot-to-human object hand-over. The procedure combines learning from demonstration, with which a robot learns to imitate a demonstrator’s execution of the task, and a reinforcement learning strategy, which enables subsequent experiential learning of contextualized policies, guided by information about context that is integrated into the learning process. By extending the initial, static hand-over policy to a contextually adaptive one, the robot derives and executes variants of the demonstrated action that most appropriately suit the current context. We use dynamic movement primitives (DMPs) as compact motion representations, and a model-based Contextual Relative Entropy Policy Search (C-REPS) algorithm for learning policies that can specify hand-over position, trajectory shape, and execution speed, conditioned on context variables. Policies are learned using simulated task executions, before transferring them to the robot and evaluating emergent behaviours.
We demonstrate the algorithm’s ability to learn context-dependent hand-over positions, and new trajectories, guided by suitable reward functions, and show that the current DMP implementation limits learning context-dependent execution speeds. We additionally conduct a user study involving participants assuming different postures and receiving an object from the robot, which executes hand-overs by either exclusively imitating a demonstrated motion, or selecting hand-over positions based on learned contextual policies and adapting its motion accordingly. The results confirm the hypothesized improvements in the robot’s perceived behaviour when it is context-aware and adaptive, and provide useful insights that can inform future developments.
05-2020
The ability to finely segment different instances of various objects in an environment forms a critical tool in the perception tool-box of any autonomous agent. Traditionally instance segmentation is treated as a multi-label pixel-wise classification problem. This formulation has resulted in networks that are capable of producing high-quality instance masks but are extremely slow for real-world usage, especially on platforms with limited computational capabilities. This thesis investigates an alternate regression-based formulation of instance segmentation to achieve a good trade-off between mask precision and run-time. Particularly the instance masks are parameterized and a CNN is trained to regress to these parameters, analogous to bounding box regression performed by an object detection network.
In this investigation, the instance segmentation masks in the Cityscape dataset are approximated using irregular octagons and an existing object detector network (i.e., SqueezeDet) is modified to regresses to the parameters of these octagonal approximations. The resulting network is referred to as SqueezeDetOcta. At the image boundaries, object instances are only partially visible. Due to the convolutional nature of most object detection networks, special handling of the boundary adhering object instances is warranted. However, the current object detection techniques seem to be unaffected by this and handle all the object instances alike. To this end, this work proposes selectively learning only partial, untainted parameters of the bounding box approximation of the boundary adhering object instances. Anchor-based object detection networks like SqueezeDet and YOLOv2 have a discrepancy between the ground-truth encoding/decoding scheme and the coordinate space used for clustering, to generate the prior anchor shapes. To resolve this disagreement, this work proposes clustering in a space defined by two coordinate axes representing the natural log transformations of the width and height of the ground-truth bounding boxes.
When both SqueezeDet and SqueezeDetOcta were trained from scratch, SqueezeDetOcta lagged behind the SqueezeDet network by a massive ≈ 6.19 mAP. Further analysis revealed that the sparsity of the annotated data was the reason for this lackluster performance of the SqueezeDetOcta network. To mitigate this issue transfer-learning was used to fine-tune the SqueezeDetOcta network starting from the trained weights of the SqueezeDet network. When all the layers of the SqueezeDetOcta were fine-tuned, it outperformed the SqueezeDet network paired with logarithmically extracted anchors by ≈ 0.77 mAP. In addition to this, the forward pass latencies of both SqueezeDet and SqueezeDetOcta are close to ≈ 19ms. Boundary adhesion considerations, during training, resulted in an improvement of ≈ 2.62 mAP of the baseline SqueezeDet network. A SqueezeDet network paired with logarithmically extracted anchors improved the performance of the baseline SqueezeDet network by ≈ 1.85 mAP.
In summary, this work demonstrates that if given sufficient fine instance annotated data, an existing object detection network can be modified to predict much finer approximations (i.e., irregular octagons) of the instance annotations, whilst having the same forward pass latency as that of the bounding box predicting network. The results justify the merits of logarithmically extracted anchors to boost the performance of any anchor-based object detection network. The results also showed that the special handling of image boundary adhering object instances produces more performant object detectors.
02-2020
Object detectors have improved considerably in the last years by using advanced Convolutional Neural Networks (CNNs) architectures. However, many detector hyper-parameters are not generally tuned, and they are used with values set by the detector authors. Blackbox optimization methods have gained more attention in recent years because of its ability to optimize the hyper-parameters of various machine learning algorithms and deep learning models. However, these methods are not explored in improving CNN-based object detector's hyper-parameters. In this research work, we propose the use of blackbox optimization methods such as Gaussian Process based Bayesian Optimization (BOGP), Sequential Model-based Algorithm Configuration (SMAC), and Covariance Matrix Adaptation Evolution Strategy (CMA-ES) to tune the hyper-parameters in Faster R-CNN and Single Shot MultiBox Detector (SSD). In Faster R-CNN, tuning the input image size, prior box anchor scales and ratios using BOGP, SMAC, and CMA-ES has increased the performance around 1.5% in terms of Mean Average Precision (mAP) on PASCAL VOC. Tuning the anchor scales of SSD has increased the mAP by 3% on PASCAL VOC and marine debris datasets. On the COCO dataset with SSD, mAP improvement is observed in the medium and large objects, but mAP decreases by 1% in small objects. The experimental results show that the blackbox optimization methods have proved to increase the mAP performance by optimizing the object detectors. Moreover, it has achieved better results than the hand-tuned configurations in most of the cases.
03-2020
Human and robot tasks in household environments include actions such as carrying an object, cleaning a surface, etc. These tasks are performed by means of dexterous manipulation, and for humans, they are straightforward to accomplish. Moreover, humans perform these actions with reasonable accuracy and precision but with much less energy and stress on the actuators (muscles) than the robots do. The high agility in controlling their forces and motions is actually due to "laziness", i.e. humans exploit the existing natural forces and constraints to execute the tasks.
The above-mentioned properties of the human lazy strategy motivate us to relax the problem of controlling robot motions and forces, and solve it with the help of the environment. Therefore, in this work, we developed a lazy control strategy, i.e. task specification models and control architectures that relax several aspects of robot control by exploiting prior knowledge about the task and environment. The developed control strategy is realized in four different robotics use cases. In this work, the Popov-Vereshchagin hybrid dynamics solver is used as one of the building blocks in the proposed control architectures. An extension of the solver’s interface with the artificial Cartesian force and feed-forward joint torque task-drivers is proposed in this thesis.
To validate the proposed lazy control approach, an experimental evaluation was performed in a simulation environment and on a real robot platform.
04-2020
A Comparative Study of Uncertainty Estimation Methods in Deep Learning Based Classification Models
(2020)
Deep learning models produce overconfident predictions even for misclassified data. This work aims to improve the safety guarantees of software-intensive systems that use deep learning based classification models for decision making by performing comparative evaluation of different uncertainty estimation methods to identify possible misclassifications.
In this work, uncertainty estimation methods applicable to deep learning models are reviewed and those which can be seamlessly integrated to existing deployed deep learning architectures are selected for evaluation. The different uncertainty estimation methods, deep ensembles, test-time data augmentation and Monte Carlo dropout with its variants, are empirically evaluated on two standard datasets (CIFAR-10 and CIFAR-100) and two custom classification datasets (optical inspection and RoboCup@Work dataset). A relative ranking between the methods is provided by evaluating the deep learning classifiers on various aspects such as uncertainty quality, classifier performance and calibration. Standard metrics like entropy, cross-entropy, mutual information, and variance, combined with a rank histogram based method to identify uncertain predictions by thresholding on these metrics, are used to evaluate uncertainty quality.
The results indicate that Monte Carlo dropout combined with test-time data augmentation outperforms all other methods by identifying more than 95% of the misclassifications and representing uncertainty in the highest number of samples in the test set. It also yields a better classifier performance and calibration in terms of higher accuracy and lower Expected Calibration Error (ECE), respectively. A python based uncertainty estimation library for training and real-time uncertainty estimation of deep learning based classification models is also developed.
02-2019
Multi-robot systems (MRS) are capable of performing a set of tasks by dividing them among the robots in the fleet. One of the challenges of working with multirobot systems is deciding which robot should execute each task. Multi-robot task allocation (MRTA) algorithms address this problem by explicitly assigning tasks to robots with the goal of maximizing the overall performance of the system. The indoor transportation of goods is a practical application of multi-robot systems in the area of logistics. The ROPOD project works on developing multi-robot system solutions for logistics in hospital facilities. The correct selection of an MRTA algorithm is crucial for enhancing transportation tasks. Several multi-robot task allocation algorithms exist in the literature, but just few experimental comparative analysis have been performed. This project analyzes and assesses the performance of MRTA algorithms for allocating supply cart transportation tasks to a fleet of robots. We conducted a qualitative analysis of MRTA algorithms, selected the most suitable ones based on the ROPOD requirements, implemented four of them (MURDOCH, SSI, TeSSI, and TeSSIduo), and evaluated the quality of their allocations using a common experimental setup and 10 experiments. Our experiments include off-line and semi on-line allocation of tasks as well as scalability tests and use virtual robots implemented as Docker containers. This design should facilitate deployment of the system on the physical robots. Our experiments conclude that TeSSI and TeSSIduo suit best the ROPOD requirements. Both use temporal constraints to build task schedules and run in polynomial time, which allow them to scale well with the number of tasks and robots. TeSSI distributes the tasks among more robots in the fleet, while TeSSIduo tends to use a lower percentage of the available robots.
Subsequently, we have integrated TeSSI and TeSSIduo to perform multi-robot task allocation for the ROPOD project.
03-2019
Currently, a variety of methods exist for creating different types of spatio-temporal world models. Despite the numerous methods for this type of modeling, there exists no methodology for comparing the different approaches or their suitability for a given application e.g. logistics robots. In order to establish a means for comparing and selecting the best-fitting spatio-temporal world modeling technique, a methodology and standard set of criteria must be established. To that end, state-of-the-art methods for this type of modeling will be collected, listed, and described. Existing methods used for evaluation will also be collected where possible.
Using the collected methods, new criteria and techniques will be devised to enable the comparison of various methods in a qualitative manner. Experiments will be proposed to further narrow and ultimately select a spatio-temporal model for a given purpose. An example network of autonomous logistic robots, ROPOD, will serve as a case study used to demonstrate the use of the new criteria. This will also serve to guide the design of future experiments that aim to select a spatio-temporal world modeling technique for a given task. ROPOD was specifically selected as it operates in a real-world, human shared environment. This type of environment is desirable for experiments as it provides a unique combination of common and novel problems that arise when selecting an appropriate spatio-temporal world model. Using the developed criteria, a qualitative analysis will be applied to the selected methods to remove unfit options.
Then, experiments will be run on the remaining methods to provide comparative benchmarks. Finally, the results will be analyzed and recommendations to ROPOD will be made.
01-2019
Interactive Object Detection
(2019)
The success of state-of-the-art object detection methods depend heavily on the availability of a large amount of annotated image data. The raw image data available from various sources are abundant but non-annotated. Annotating image data is often costly, time-consuming or needs expert help. In this work, a new paradigm of learning called Active Learning is explored which uses user interaction to obtain annotations for a subset of the dataset. The goal of active learning is to achieve superior object detection performance with images that are annotated on demand. To realize active learning method, the trade-off between the effort to annotate (annotation cost) unlabeled data and the performance of object detection model is minimised.
Random Forests based method called Hough Forest is chosen as the object detection model and the annotation cost is calculated as the predicted false positive and false negative rate. The framework is successfully evaluated on two Computer Vision benchmark and two Carl Zeiss custom datasets. Also, an evaluation of RGB, HoG and Deep features for the task is presented.
Experimental results show that using Deep features with Hough Forest achieves the maximum performance. By employing Active Learning, it is demonstrated that performance comparable to the fully supervised setting can be achieved by annotating just 2.5% of the images. To this end, an annotation tool is developed for user interaction during Active Learning.
03-2018
Friction effects impose a requirement for the supplementary amount of torque to be produced in actuators for a robot to move, which in turn increases energy consumption. We cannot eliminate friction, but we can optimize motions to make them more energy efficient, by considering friction effects in motion computations. Optimizing motions means computing efficient joint torques/accelerations based on different friction torques imposed in each joint. Existing friction forces can be used for supporting certain types of arm motions, e.g standing still.
Reducing energy consumption of robot's arms will provide many benefits, such as longer battery life of mobile robots, reducing heat in motor systems, etc.
The aim of this project is extending an already available constrained hybrid dynamic solver, by including static friction effects in the computations of energy optimal motions. When the algorithm is extended to account for static friction factors, a convex optimization (maximization) problem must be solved.
The author of this hybrid dynamic solver has briefly outlined the approach for including static friction forces in computations of motions, but without providing a detailed derivation of the approach and elaboration that will show its correctness. Additionally, the author has outlined the idea for improving the computational efficiency of the approach, but without providing its derivation.
In this project, the proposed approach for extending the originally formulated algorithm has been completely derived and evaluated in order to show its feasibility. The evaluation is conducted in simulation environment with one DOF robot arm, and it shows correct results from the computation of motions. Furthermore, this project presents the derivation of the outlined method for improving the computational efficiency of the extended solver.
01-2018
Motion capture, often abbreviated mocap, generally aims at recording any kind of motion -- be it from a person or an object -- and to transform it to a computer-readable format. Especially the data recorded from (professional and non-professional) human actors are typically used for analysis in e.g. medicine, sport sciences, or biomechanics for evaluation of human motion across various factors. Motion capture is also widely used in the entertainment industry: In video games and films realistic motion sequences and animations are generated through data-driven motion synthesis based on recorded motion (capture) data.
Although the amount of publicly available full-body-motion capture data is growing, the research community still lacks a comparable corpus of specialty motion data such as, e.g. prehensile movements for everyday actions. On the one hand, such data can be used to enrich (hand-over animation) full-body motion capture data - usually captured without hand motion data due to the drastic dimensional difference in articulation detail. On the other hand, it provides means to classify and analyse prehensile movements with or without respect to the concrete object manipulated and to transfer the acquired knowledge to other fields of research (e.g. from 'pure' motion analysis to robotics or biomechanics).
Therefore, the objective of this motion capture database is to provide well-documented, free motion capture data for research purposes.
The presented database GraspDB14 in sum contains over 2000 prehensile movements of ten different non-professional actors interacting with 15 different objects. Each grasp was realised five times by each actor. The motions are systematically named containing an (anonymous) identifier for each actor as well as one for the object grasped or interacted with.
The data were recorded as joint angles (and raw 8-bit sensor data) which can be transformed into positional 3D data (3D trajectories of each joint).
In this document, we provide a detailed description on the GraspDB14-database as well as on its creation (for reproducibility).
Chapter 2 gives a brief overview of motion capture techniques, freely available motion capture databases for both, full body motions and hand motions, and a short section on how such data is made useful and re-used. Chapter 3 describes the database recording process and details the recording setup and the recorded scenarios. It includes a list of objects and performed types of interaction. Chapter 4 covers used file formats, contents, and naming patterns. We provide various tools for parsing, conversion, and visualisation of the recorded motion sequences and document their usage in chapter 5.
04-2017
Emotional communication is a key element of habilitation care of persons with dementia. It is, therefore, highly preferable for assistive robots that are used to supplement human care provided to persons with dementia, to possess the ability to recognize and respond to emotions expressed by those who are being cared-for. Facial expressions are one of the key modalities through which emotions are conveyed. This work focuses on computer vision-based recognition of facial expressions of emotions conveyed by the elderly.
Although there has been much work on automatic facial expression recognition, the algorithms have been experimentally validated primarily on young faces. The facial expressions on older faces has been totally excluded. This is due to the fact that the facial expression databases that were available and that have been used in facial expression recognition research so far do not contain images of facial expressions of people above the age of 65 years. To overcome this problem, we adopt a recently published database, namely, the FACES database, which was developed to address exactly the same problem in the area of human behavioural research. The FACES database contains 2052 images of six different facial expressions, with almost identical and systematic representation of the young, middle-aged and older age-groups.
In this work, we evaluate and compare the performance of two of the existing imagebased approaches for facial expression recognition, over a broad spectrum of age ranging from 19 to 80 years. The evaluated systems use Gabor filters and uniform local binary patterns (LBP) for feature extraction, and AdaBoost.MH with multi-threshold stump learner for expression classification. We have experimentally validated the hypotheses that facial expression recognition systems trained only on young faces perform poorly on middle-aged and older faces, and that such systems confuse ageing-related facial features on neutral faces with other expressions of emotions. We also identified that, among the three age-groups, the middle-aged group provides the best generalization performance across the entire age spectrum. The performance of the systems was also compared to the performance of humans in recognizing facial expressions of emotions. Some similarities were observed, such as, difficulty in recognizing the expressions on older faces, and difficulty in recognizing the expression of sadness.
The findings of our work establish the need for developing approaches for facial expression recognition that are robust to the effects of ageing on the face. The scientific results of our work can be used as a basis to guide future research in this direction.
03-2017
Population ageing and growing prevalence of disability have resulted in a growing need for personal care and assistance. The insufficient supply of personal care workers and the rising costs of long-term care have turned this phenomenon into a greater social concern. This has resulted in a growing interest in assistive technology in general, and assistive robots in particular, as a means of substituting or supplementing the care provided by humans, and as a means of increasing the independence and overall quality of life of persons with special needs. Although many assistive robots have been developed in research labs world-wide, very few are commercially available. One of the reasons for this, is the cost. One way of optimising cost is to develop solutions that address specific needs of users. As a precursor to this, it is important to identify gaps between what the users need and what the technology (assistive robots) currently provides. This information is obtained through technology mapping.
The current literature lacks a mapping between user needs and assistive robots, at the level of individual systems. The user needs are not expressed in uniform terminology across studies, which makes comparison of results difficult. In this research work, we have illustrated the technology mapping of assistive robots using the International Classification of Functioning, Disability and Health (ICF). ICF provides standard terminology for expressing user needs in detail. Expressing the assistive functions of robots also in ICF terminology facilitates communication between different stakeholders (rehabilitation professionals, robotics researchers, etc.).
We also investigated existing taxonomies for assistive robots. It was observed that there is no widely accepted taxonomy for classifying assistive robots. However, there exists an international standard, ISO 9999, which classifies commercially available assistive products. The applicability of the latest revision of ISO 9999 standard for classifying mobility assistance robots has been studied. A partial classification of assistive robots based on ISO 9999 is suggested. The taxonomy and technology mapping are illustrated with the help of four robots that have the potential to provide mobility assistance. These are the SmartCane, the SmartWalker, MAid and Care-O-bot (R) 3. SmartCane, SmartWalker and MAid provide assistance by supporting physical movement. Care-O-bot (R) 3 provides assistance by reducing the need to move.
01-2017
This paper describes the security mechanisms of several wireless building automation technologies, namely ZigBee, EnOcean, ZWave, KNX, FS20, and Home-Matic. It is shown that none of the technologies provides the necessary measure ofsecurity that should be expected in building automation systems. One of the conclusions drawn is that software embedded in systems that are build for a lifetime of twenty years or more needs to be updatable.
02-2017
Recent work in image captioning and scene-segmentation has shown significant results in the context of scene-understanding. However, most of these developments have not been extrapolated to research areas such as robotics. In this work we review the current state-ofthe- art models, datasets and metrics in image captioning and scenesegmentation. We introduce an anomaly detection dataset for the purpose of robotic applications, and we present a deep learning architecture that describes and classifies anomalous situations. We report a METEOR score of 16.2 and a classification accuracy of 97 %.
04-2015
Advanced driver assistance systems (ADAS) are technology systems and devices designed as an aid to the driver of a vehicle. One of the critical components of any ADAS is the traffic sign recognition module. For this module to achieve real-time performance, some preprocessing of input images must be done, which consists of a traffic sign detection (TSD) algorithm to reduce the possible hypothesis space. Performance of TSD algorithm is critical.
One of the best algorithms used for TSD is the Radial Symmetry Detector (RSD), which can detect both Circular [7] and Polygonal traffic signs [5]. This algorithm runs in real-time on high end personal computers, but computational performance of must be improved in order to be able to run in real-time in embedded computer platforms.
To improve the computational performance of the RSD, we propose a multiscale approach and the removal of a gaussian smoothing filter used in this algorithm. We evaluate the performance on both computation times, detection and false positive rates on a synthetic image dataset and on the german traffic sign detection benchmark [29].
We observed significant speedups compared to the original algorithm. Our Improved Radial Symmetry Detector is up to 5.8 times faster than the original on detecting Circles, up to 3.8 times faster on Triangle detection, 2.9 times faster on Square detection and 2.4 times faster on Octagon detection. All of this measurements were observed with better detection and false positive rates than the original RSD.
When evaluated on the GTSDB, we observed smaller speedups, in the range of 1.6 to 2.3 times faster for Circle and Regular Polygon detection, but for Circle detection we observed a decreased detection rate than the original algorithm, while for Regular Polygon detection we always observed better detection rates. False positive rates were high, in the range of 80% to 90%.
We conclude that our Improved Radial Symmetry Detector is a significant improvement of the Radial Symmetry Detector, both for Circle and Regular polygon detection. We expect that our improved algorithm will lead the way to obtain real-time traffic sign detection and recognition in embedded computer platforms.
05-2015
Extraction of text information from visual sources is an important component of many modern applications, for example, extracting the text from traffic signs on a road scene in an autonomous vehicle. For natural images or road scenes this is a unsolved problem. In this thesis the use of histogram of stroke widths (HSW) for character and noncharacter region classification is presented. Stroke widths are extracted using two methods. One is based on the Stroke Width Transform and another based on run lengths. The HSW is combined with two simple region features– aspect and occupancy ratios– and then a linear SVM is used as classifier. One advantage of our method over the state of the art is that it is script-independent and can also be used to verify detected text regions with the purpose of reducing false positives. Our experiments on generated datasets of Latin, CJK, Hiragana and Katakana characters show that the HSW is able to correctly classify at least 90% of the character regions, a similar figure is obtained for non-character regions. This performance is also obtained when training the HSW with one script and testing with a different one, and even when characters are rotated. On the English and Kannada portions of the Chars74K dataset we obtained over 95% correctly classified character regions. The use of raycasting for text line grouping is also proposed. By combining it with our HSW-based character classifier, a text detector based on Maximally Stable Extremal Regions (MSER) was implemented. The text detector was evaluated on our own dataset of road scenes from the German Autobahn, where 65% precision, 72% recall with a f-score of 69% was obtained. Using the HSW as a text verifier increases precision while slightly reducing recall. Our HSW feature allows the building of a script-independent and low parameter count classifier for character and non-character regions.
01-2015
Rural areas often lack affordable broadband Internet connectivity, mainly due to the CAPEX and especially OPEX of traditional operator equipment [HEKN11]. This digital divide limits the access to knowledge, health care and other services for billions of people. Different approaches to close this gap were discussed in the last decade [SPNB08]. In most rural areas satellite bandwidth is expensive and cellular networks (3G,4G) as well as WiMAX suffer from the usually low population density making it hard to amortize the costs of a base station [SPNB08].
02-2014
A principal step towards solving diverse perception problems is segmentation. Many algorithms benefit from initially partitioning input point clouds into objects and their parts. In accordance with cognitive sciences, segmentation goal may be formulated as to split point clouds into locally smooth convex areas, enclosed by sharp concave boundaries. This goal is based on purely geometrical considerations and does not incorporate any constraints, or semantics, of the scene and objects being segmented, which makes it very general and widely applicable. In this work we perform geometrical segmentation of point cloud data according to the stated goal. The data is mapped onto a graph and the task of graph partitioning is considered. We formulate an objective function and derive a discrete optimization problem based on it. Finding the globally optimal solution is an NP-complete problem; in order to circumvent this, spectral methods are applied. Two algorithms that implement the divisive hierarchical clustering scheme are proposed. They derive graph partition by analyzing the eigenvectors obtained through spectral relaxation. The specifics of our application domain are used to automatically introduce cannot-link constraints in the clustering problem. The algorithms function in completely unsupervised manner and make no assumptions about shapes of objects and structures that they segment. Three publicly available datasets with cluttered real-world scenes and an abundance of box-like, cylindrical, and free-form objects are used to demonstrate convincing performance. Preliminary results of this thesis have been contributed to the International Conference on Autonomous Intelligent Systems (IAS-13).