A learning architecture based on reinforcement learning for adaptive control of the walking machine LAURON

The learning of complex control behaviour of autonomous mobile robots is one of the actual research topics. In this article an intelligent control architecture is presented which integrates learning methods and available domain knowledge. This control architecture is based on Reinforcement Learning and allows continuous input and output parameters, hierarchical learning, multiple goals, self-organized topology of the used networks and online learning. As a testbed this architecture is applied to the six-legged walking machine LAURON to learn leg control and leg coordination.     

Insufficient incentives for investment in electricity generations

In theory, competitive electricity markets provide incentives for efficient investment in generation capacity. We show that if consumers and investors are risk averse, investment is efficient only if investors in generation capacity can sign long-term contracts with consumers. Otherwise the uncovered price risk increases financing costs, reduces equilibrium investment levels, distorts technology choice towards less capital-intensive generation and reduces consumer utility. We observe insufficient levels of long-term contracts in existing markets, possibly because retail companies are not credible counter-parties if their final customers can switch easily between them. With a consumer franchise, retailers can sign long-term contracts, but this solution comes at the expense of retail competition. Alternative capacity mechanisms to stimulate investment are discussed.     

UppDMC: A Distributed Model Checker for Fragments of the μ-Calculus

We present UppDMC, a distributed model-checking tool. It is tailored for checking finite-state systems and μ-calculus specifications with at most one alternation of minimal and maximal fixed-point operators. This fragment is also known as . Recently, efficient game-based algorithms for this logic have been outlined.We describe the implementation of these algorithms within UppDMC and study their performance on practical examples. Running UppDMC on a simple workstation cluster, we were able to check liveness properties of the largest examples given in the VLTS Benchmark Suite, for which no answers were previously known.     

Structure analysis of acetylated and non-acetylated O-linked MUC1-glycopeptides by post-source decay matrix-assisted laser desorption/ionization mass spectrometry

We have investigated the potential of structural elucidation of O-linked glycopeptides by post-source decay matrix-assisted laser desorption ionization mass spectrometry (PSD-MALDI-MS). In order to establish detailed fragmentation patterns and to dissect fragment ions with and without carbohydrate content, the same O-linked MUC1-derived glycopeptides with acetylated and non-acetylated sugars were analysed and compared. Furthermore, we were interested to examine possible differences in the fragmentation between glycopeptides with acetylated and non-acetylated sugars. The obtained PSD-MALDI-MS spectra showed a rather complete set of fragmentation data which allows to localize the glycan on the peptide, in parallel with sequencing a short glycan and the backbone peptide. Fragmentations of the sugars were dominated by inter-ring cleavages at the glycosidic bond. Intra-ring cleavage did also occur from the non-reducing end, but to a much lower extent. The fragmentation of the peptide backbone was not changed either by acetylated or non-acetylated sugars. Glycosylated peptide fragments occurred as fully glycosylated fragment ions, partially deglycosylated ions and completely deglycosylated ions, and was not influenced by the acetylation of sugars. However, differences occurred in the quality and quantity of fragment ions from the non-reducing end of the glycan part when comparing acetylated with non-acetylated glycopeptides.     

The use of video technology for providing feedback to students: Can it enhance the feedback experience for staff and students?

There are numerous issues surrounding the provision of assessment-related feedback in Higher Education, which in recent years have been highlighted in the National Student Survey. In this paper questionnaire data from staff and students at the University of Reading are used to confirm the main issues encountered with feedback, namely problems of time efficiency for staff, lack of engagement by students with feedback and issues with the timeliness and quality of feedback received. Therefore we explored the potential of technology, specifically video, to address these issues by enabling staff to produce brief feedback videos for students. The videos were housed within a new online resource, ‘ASSET’, and were used to investigate whether use of this technology could enhance the feedback experience for both staff and students. A pilot of the ASSET resource for generic feedback provision found that it was considered advantageous by staff and students. Moreover, the use of video was also shown to resolve many of the common problems of feedback in relation to quality and engagement of students.     

The simulator TOUGH2/EWASG for modelling geothermal reservoirs with brines and non-condensible gas

An equation-of-state (EOS) module has been developed for the TOUGH2 simulator, belonging to the MULKOM family of computer codes developed at Lawrence Berkeley National Laboratory. This module, named EWASG (Equation-of-State for Water, Salt and Gas), is able to handle three-component mixtures of water, sodium chloride, and a slightly soluble non-condensible gas (NCG). At present the NCG can be chosen to be air, CO₂, CH₄, H₂, or N₂. EWASG can describe liquid and gas phases, and includes precipitation and dissolution of solid salt. The dependence of density, viscosity, enthalpy, and vapour pressure of brine on salt concentration is taken into account, as well as the effects of salinity on gas solubility in the liquid phase and related heat of solution. The reduction of rock porosity because of salt precipitation is also considered, as well as the related decrease of formation permeability. Vapour pressure lowering (VPL) due to suction pressure is represented by Kelvin's equation, in which the effects of salt are considered whereas those of NCG have currently been neglected.The main assumptions made in developing the EWASG module are described, together with the correlations employed to calculate the thermophysical properties of multiphase H₂ONaClCO₂ mixtures, which can be used to simulate the thermodynamic behaviour of commonly exploited geothermal reservoirs.     

A taxonomy of interaction models for Internet and ATM quality of service architectures

In communication systems there are two worlds at the moment: Internet and ATM. Both possess Quality of Service (QoS) architectures which shall allow them to integrate services of data and telecommunications formerly performed by separate infrastructures. We believe that none of them will be able to oust the other. That means both will exist for at least the middle-term future. Therefore, an interaction between both appears to be necessary, especially in the field of distributed multimedia applications where both worlds meet first. In order to perform gracefully, distributed multimedia applications require a certain QoS provision, in particular from the communication system. Thus, for such applications the existence of heterogeneous IP/ATM networks makes the interaction between Internet and ATM QoS architectures an important issue. In this article a taxonomy of interaction models for the Internet and ATM QoS architectures is developed. We do not let our view be restricted by existing approaches for the interaction between ATM and Internet. Instead we will derive more unconventional models by regarding the possible communication patterns based on different topological variants for heterogeneous IP/ATM networks. The investigation is driven by applications' communication requirements. This is accomplished by examining possibly interacting applications and their communication patterns. The interaction models are contrasted and compared to each other and their assumptions and implications are shown. The derived taxonomy of models allows us to classify proposed approaches for the interaction of Internet and ATM QoS architectures. Thereby we are able to identify the basic assumptions of these approaches and their corresponding restrictions.     

How to improve collaborative learning with video tools in the classroom? Social vs. cognitive guidance for student teams

Digital video technologies offer a variety of functions for supporting collaborative learning in classrooms. Yet, for novice learners, such as school students, positive learning outcomes also depend centrally on effective social interactions. We present empirical evidence for the positive effects of instructive guidance on performance and on learning of students who use web-based video tools during a short collaborative-design task in their history lesson. In an experiment with 16-year old learners (N?=?148) working on a history topic, we compared two contrasting types of guidance for student teams?? collaboration processes (social-interaction-related vs. cognitive-task-related guidance). We also compared two types of advanced video tools. Both types of guidance and tools were aimed at supporting students?? active, meaningful learning and critical analysis of a historical newsreel. Results indicated that social-interaction-related guidance was more effective in terms of learning outcomes (e.g., the students?? history skills) than cognitive-task-related guidance. The different tools did not yield consistent results. The implications of these findings are discussed.     

Electrochemical etch-stop control for silicon structures containing electronic components

A predictive analysis of the feasibility of electrochemical etch-stop control in the fabrication of some silicon sensors was validated experimentally. The application is for sensors with thin silicon structures containing electronic components, such as cantilever accelerometers and diaphragm pressure gauges with strain gauges in the surface to detect deflection. Such structures are formed by deep anisotropic etching of silicon. Since the depth of the etch determines the thickness of the cantilever or diaphragm, the ability to stop the etching precisely is critical. Accurate etch-stop control can be accomplished by electrochemical passivation of an n-type epitaxial layer on a p-type silicon wafer, where the epi layer thickness becomes that of the diaphragm or cantilever. The analysis shows that passivation of the epi layer can be maintained even underneath the electronic components for conditions which allow etching of the p-type silicon substrate. Therefore, electrochemical etch-stop control appears feasible in most practical sensor designs.