Refine
Document Type
- Conference Object (44)
- Report (4)
- Article (3)
- Part of a Book (3)
- Preprint (3)
- Book (1)
- Doctoral Thesis (1)
Year of publication
Keywords
- Object recognition (2)
- Advanced Driver Assistance Systems (1)
- Dieseleinspritzanlage (1)
- Domestic robotics (1)
- Domestic service robots (1)
- Eingebettetes System (1)
- Elektrische Simulation (1)
- Entwurf (1)
- Fusion (1)
- Hardware (1)
Mobile manipulators are intended to be deployed in domestic and industrial environments where they will carry out tasks that require physical interaction with the surrounding world, for example, picking or handing over fragile objects. In-hand slippage, i.e. a grasped object moving within the robot’s grasp, is inherent to many of these tasks and thus, a robot’s ability to detect a slippage is vital for executing a manipulation task successfully. In this paper, we develop a slip detection approach which is based on the robot’s tactile sensors, a force/torque sensor and a combination thereof. The evaluation of our approach, carried out on the Care-O-bot 3 platform, highly suggests that the actions and motions performed by the robot during grasping should be taken into account during slip detection for improved performance. Based on this insight, we propose an in-hand slip detection architecture that is able to adapt to the current robot’s actions at run time.
With regard to performance well established SW-only design methodologies proceed by making the initial specification run first, then by enhancing its functionality and finally by optimizing it. When designing Embedded Systems (EbS) this approach is not viable since decisive design decisions like e.g. the estimation of required processing power or the identification of those parts of the specification which need to be delegated to dedicated HW depend on the fastness and fairness of the initial specification. We here propose a sequence of optimization steps embedded into the design flow, which enables a structured way to accelerate a given working EbS specification at different layers. This sequence of accelerations comprises algorithm selection, algorithm transformation, data transformation, implementation optimization and finally HW acceleration. It is analyzed how all acceleration steps are influenced by the specific attributes of the underlying EbS. The overall acceleration procedure is explained and quantified at hand of a real-life industrial example.
We describe the development of an architecture for the DESIRE technology demonstrator based on principles of classical component based software engineering. The architecture is directly derived from the project requirements and resides on the concept of an Autonomous Component utilizing a smart feedback value called WishLists. This return type is able to provide expert advice about the reasons of occurring failures and give hints for possible recovery strategies. This is of key importance to advance towards robustness. The integration of an AI task planner allows the realization of higher flexibility, dependability and capability during task execution and may resolve conflicts between occurring WishLists. Furthermore the necessity of a central system-state model (Eigenmodel), which represents the current state and configuration of the whole system at runtime, is explained and illustrated. We conclude with some lessons learned.
SISAL: User manual
(1990)
Robust text detection and recognition in arbitrarily distributed, unrestricted images is a difficult problem, e.g. when interpreting traffic panels outdoors during autonomous driving. Most previous work in text detection considers only a single script, usually Latin, and it is not able to detect text with multiple scripts. Our contribution combines an established technique -Maximum Stable Extremal Regions-with a histogram of stroke width (HSW) feature and a Support Vector Machine classifier. We combined characters into groups by raycasting and merged aligned groups into lines of text that can also be verified by using the HSW. We evaluated our detection pipeline on our own dataset of road scenes from Autobahn (German Highways), and show how the character classifier stage can be trained with one script and be successfully tested on a different one. While precision and recall match to state of the art solution. A unique characteristic of the HSW feature is that it can learn and detect multiple scripts, which we believe can yield script independence.
RoboCup projects can face a lack of progress and continuity. The teams change continuously and knowledge gets lost. The approach used in previous years is no longer valid due to rule changes and specialists leaving the team leave black boxes that no-one understands. This article presents the application of a recent software development technique called eXtreme Programming to the realm of RoboCup. Many common problems typical for teams of students seem to be solvable with this technique. It also gradually spreads out in professional software production companies. Students mastering it are of high use for their further career after having left the university. The strategy is being tested on a real RoboCup Mid-Size and an Aibo league project and produces very promising results. The approach makes it possible to modularize scientific knowledge into software that can be re-used. Both the scientist/expert, who has the knowledge, and the software development team benefit from this approach without much overhead on the project.
Cosynthesis in CASTLE
(1995)
Dual Dynamics (DD) is a mathematical model of a behavior control system for mobile autonomous robots. Behaviors are specified through differential equations; forming a global dynamical system made of behavior subsystems which interact in a number of ways. DD models can be directly compiled into executable code. The article (i) explains the model; (ii) sketches the Dual Dynamics Designer (DDD) environment that we use for the design; simulation; implementation and documentation; and (iii) illustrates our approach with the example of kicking a moving ball into a goal.
Co-design is concerned with the joint design of hardware and software making up an embedded computer system [Wol94]. A top down design flow for an embedded system begins with a system specification. If it is executable, it may be used for simulation, system verification or to identify algorithmical bottlenecks. In contrast to other chapters of this book, the specification is not developed in this case study, rather it is given from the beginning. Furthermore we are not concerned with partitioning or synthesis of dedicated HW. Instead we focus on the problem how to find an off-the-shelf micro-controller which implements the desired functionality and meets all specification constraints. If feasible, this is usually much cheaper then using dedicated hardware. This chapter will answer the question of feasibility for a real life problem from automobile industry.
CASTLE is a co-design platform developed at GMD SET institute. It provides a number of design tools for configuring application specific design flows. This paper presents a walk through the CASTLE co-design environment, following the design flow of a video processing system. The design methodology and the tool usage for this real life example are described, as seen from a designers point of view. The design flow starts with a C/C++ program and gradually derives a register-transfer level description of a processor hardware, as well as the corresponding compiler for generating the processor opcode. The main results of each design step are presented and the usage of the CASTLE tools at each step is explained.
Current object recognition methods fail on object sets that include both diffuse, reflective and transparent materials, although they are very common in domestic scenarios. We show that a combination of cues from multiple sensor modalities, including specular reflectance and unavailable depth information, allows us to capture a larger subset of household objects by extending a state of the art object recognition method. This leads to a significant increase in robustness of recognition over a larger set of commonly used objects.
Finding good Echo State Networks to control an underwater robot using evolutionary computations
(2004)
The control of an underwater robot is difficult due to non-linear effects. Echo State Networks (ESNs) provide a way of dealing with non-linearity. The quality of an Echo State Network (ESN) (H. Jaeger, 2001, 2002) depends strongly on its topology. Usually an educated guess combined with a brute force method is used to obtain an ESN that, after training, produces a low error. In this article the authors suggest another way to find good topologies by using evolution. Two heuristics, evolutionary algorithms and evolutionary strategies, are compared. The proposed method outperforms standard ones like ARX and backpropagation networks on the data from the Twin-Burger underwater robot.
In this paper we propose an implement a general convolutional neural network (CNN) building framework for designing real-time CNNs. We validate our models by creating a real-time vision system which accomplishes the tasks of face detection, gender classification and emotion classification simultaneously in one blended step using our proposed CNN architecture. After presenting the details of the training procedure setup we proceed to evaluate on standard benchmark sets. We report accuracies of 96% in the IMDB gender dataset and 66% in the FER-2013 emotion dataset. Along with this we also introduced the very recent real-time enabled guided back-propagation visualization technique. Guided back-propagation uncovers the dynamics of the weight changes and evaluates the learned features. We argue that the careful implementation of modern CNN architectures, the use of the current regularization methods and the visualization of previously hidden features are necessary in order to reduce the gap between slow performances and real-time architectures. Our system has been validated by its deployment on a Care-O-bot 3 robot used during RoboCup@Home competitions. All our code, demos and pre-trained architectures have been released under an open-source license in our public repository.
Current robot platforms are being employed to collaborate with humans in a wide range of domestic and industrial tasks. These environments require autonomous systems that are able to classify and communicate anomalous situations such as fires, injured persons, car accidents; or generally, any potentially dangerous situation for humans. In this paper we introduce an anomaly detection dataset for the purpose of robot applications as well as the design and implementation of a deep learning architecture that classifies and describes dangerous situations using only a single image as input. We report a classification accuracy of 97 % and METEOR score of 16.2. We will make the dataset publicly available after this paper is accepted.
We developed a scene text recognition system with active vision capabilities, namely: auto-focus, adaptive aperture control and auto-zoom. Our localization system is able to delimit text regions in images with complex backgrounds, and is based on an attentional cascade, asymmetric adaboost, decision trees and Gaussian mixture models. We think that text could become a valuable source of semantic information for robots, and we aim to raise interest in it within the robotics community. Moreover, thanks to the robot’s pan-tilt-zoom camera and to the active vision behaviors, the robot can use its affordances to overcome hindrances to the performance of the perceptual task. Detrimental conditions, such as poor illumination, blur, low resolution, etc. are very hard to deal with once an image has been captured and can often be prevented. We evaluated the localization algorithm on a public dataset and one of our own with encouraging results. Furthermore, we offer an interesting experiment in active vision, which makes us consider that active sensing in general should be considered early on when addressing complex perceptual problems in embodied agents.
The ability to track moving people is a key aspect of autonomous robot systems in real-world environments. Whilst for many tasks knowing the approximate positions of people may be sufficient, the ability to identify unique people is needed to accurately count people in the real world. To accomplish the people counting task, a robust system for people detection, tracking and identification is needed.
This paper presents a robust object tracking method using a sparse shape-based object model. Our approach consists of three ingredients namely shapes, a motion model and a sparse (non-binary) subsampling of colours in background and foreground parts based on the shape assumption. The tracking itself is inspired by the idea of having a short-term and a long-term memory. A lost object is ”missed” by the long-term memory when it is no longer recognized by the short-term memory. Moreover, the long-term memory allows to re-detect vanished objects and using their new position as a new initial position for object tracking. The short-term memory is implemented with a new Monte Carlo variant which provides a heuristic to cope with the loss-of-diversity problem. It enables simultaneous tracking of multiple (visually) identical objects. The long-term memory is implemented with a Bayesian Multiple Hypothesis filter. We demonstrate the robustness of our approach with respect to object occlusions and non-Gaussian/non-linear movements of the tracked object. We also show that tracking can be significantly improved by using compensating ego-motion. Our approach is very scalable since one can tune the parameters for a trade-off between precision and computational time.
