Deep learning in robotics: a review of recent research

Advances in deep learning over the last decade have led to a flurry of research in the application of deep artificial neural networks to robotic systems, with at least 30 papers published on the subject between 2014 and the present. This review discusses the applications, benefits, and limitations of deep learning vis-à-vis physical robotic systems, using contemporary research as exemplars. It is intended to communicate recent advances to the wider robotics community and inspire additional interest in and application of deep learning in robotics.     

Who’s my audience again? Understanding audience management strategies for designing privacy management technologies

Social network site users are often confronted with invisible audiences. Although various settings for managing audiences are available, we argue that these do not always match the users’ interpretations. This study explores the audience-management strategies of 18 young adults when categorizing their (invisible) audiences in Facebook, using card sorting as a research method. Approximately 1254 out of 1800 people (cards) were categorized based on the shared-community strategy, in which the participants referred to multiple community roles. The theoretical framework of Symbolic Interactionism and the Communication Privacy Management Theory are used to frame the problematic nature of invisible audiences. Implications for designing privacy-management technologies are discussed.     

Effect of didecyl dimethyl ammonium chloride on nitrate reduction in a mixed methanogenic culture.

The effect of the quaternary ammonium compound, didecyl dimethyl ammonium chloride (DDAC), on nitrate reduction was investigated at concentrations up to 100 mg/L in a batch assay using a mixed, mesophilic (35 degrees C) methanogenic culture. Glucose was used as the carbon and energy source and the initial nitrate concentration was 70 mg N/L. Dissimilatory nitrate reduction to ammonia (DNRA) and to dinitrogen (denitrification) were observed at DDAC concentrations up to 25 mg/L. At and above 50 mg DDAC/L, DNRA was inhibited and denitrification was incomplete resulting in accumulation of nitrous oxide. At DDAC concentrations above 10 mg/L, production of nitrous oxide, even transiently, resulted in complete, long-term inhibition of methanogenesis and accumulation of volatile fatty acids. Fermentation was inhibited at and above 75 mg DDAC/L. DDAC suppressed microbial growth and caused cell lysis at a concentration 50 mg/L or higher. Most of the added DDAC was adsorbed on the biomass. Over 96% of the added DDAC was recovered from all cultures at the end of the 100-days incubation period, indicating that DDAC did not degrade in the mixed methanogenic culture under the conditions of this study.     

Application of sputtered ruthenium nitride thin films as electrode material for energy-storage devices

1. Download : Download high-res image (80KB)
2. Download : Download full-size image

     

Origin of fecal contamination in waters from contrasted areas: stanols as Microbial Source Tracking markers

Improving the microbiological quality of coastal and river waters relies on the development of reliable markers that are capable of determining sources of fecal pollution. Recently, a principal component analysis (PCA) method based on six stanol compounds (i.e. 5β-cholestan-3β-ol (coprostanol), 5β-cholestan-3α-ol (epicoprostanol), 24-methyl-5α-cholestan-3β-ol (campestanol), 24-ethyl-5α-cholestan-3β-ol (sitostanol), 24-ethyl-5β-cholestan-3β-ol (24-ethylcoprostanol) and 24-ethyl-5β-cholestan-3α-ol (24-ethylepicoprostanol)) was shown to be suitable for distinguishing between porcine and bovine feces. In this study, we tested if this PCA method, using the above six stanols, could be used as a tool in “Microbial Source Tracking (MST)” methods in water from areas of intensive agriculture where diffuse fecal contamination is often marked by the co-existence of human and animal sources. In particular, well-defined and stable clusters were found in PCA score plots clustering samples of “pure” human, bovine and porcine feces along with runoff and diluted waters in which the source of contamination is known. A good consistency was also observed between the source assignments made by the 6-stanol-based PCA method and the microbial markers for river waters contaminated by fecal matter of unknown origin. More generally, the tests conducted in this study argue for the addition of the PCA method based on six stanols in the MST toolbox to help identify fecal contamination sources. The data presented in this study show that this addition would improve the determination of fecal contamination sources when the contamination levels are low to moderate.     

步入DirectX的圣殿（下）

本文继续介绍了3D编程中的一些术语和概念，并对前两篇文章中的示例进行了进一步的充实。自此，读者朋友就能进入DirectX游戏开发的精彩世界了。     

走进游戏开发的精彩世界

“工欲善其事，必先利其器。”本文为大家介绍了在应用DirectX开发游戏时所需的工具，并带领大家构造了一个游戏的基本代码框架。     

Transistor and circuit design for 100-200-GHz ICs

Compared to SiGe, InP HBTs offer superior electron transport properties but inferior scaling and parasitic reduction. Figures of merit for mixed-signal ICs are developed and HBT scaling laws introduced. Device and circuit results are summarized, including a simultaneous 450 GHz f/sub /spl tau// and 490 GHz f/sub max/ DHBT, 172-GHz amplifiers with 8.3-dBm output power and 4.5-dB associated power gain, and 150-GHz static frequency dividers (a digital circuit figure-of-merit for a device technology). To compete with advanced 100-nm SiGe processes, InP HBTs must be similarly scaled and high process yields are imperative. Described are several process modules in development: these include an emitter-base dielectric sidewall spacer for increased yield, a collector pedestal implant for reduced extrinsic C/sub cb/, and emitter junction regrowth for reduced base and emitter resistances.     

Influence of environmental stress on the kinetics and strength of attachment of Listeria monocytogenes Scott A to Buna-N rubber and stainless steel

Attachment and detachment of Listeria monocytogenes Scott A to Buna-N rubber and stainless steel under varying conditions of temperature and pH were investigated using model systems. Numbers of attached cells increased with increasing attachment temperature (10 to 45 degrees C) and time (up to 120 min) for both test surfaces. Compared to Buna-N rubber, the rate of attachment to stainless steel was markedly more rapid for all temperature and pH conditions studied and could not be calculated. Rate of attachment to Buna-N rubber was found to be significantly lower when cells were attached at 10 degrees C. Growth temperature did not significantly affect rates of adhesion to Buna-N rubber. Altering the medium pH during attachment between 4 and 9 demonstrated that rates of adhesion were slower under alkaline conditions. Growth pH was also found to significantly affect rates of attachment to Buna-N rubber. Detachment of cells adhered to Buna-N rubber was significantly affected by growth temperature but not growth pH. Significant differences in detachment were also found between Buna-N rubber and stainless steel, inferring stronger attachment to Buna-N rubber. Cell surface hydrophobicity was found to be affected by both growth temperature and growth pH. However, changes in hydrophobicity could not be correlated to differences in rates of attachment. Addition of 0.01% trypsin to the attachment medium during cell exposure to either test surface resulted in a 99.9% reduction in the adhered cell population when compared to controls. This would suggest that proteins play a role in the initial attachment process of L. monocytogenes.