Towards vision-based deep reinforcement learning for robotic motion control

This paper introduces a machine learning based system for controlling a robotic manipulator with visual perception only. The capability to autonomously learn robot controllers solely from raw-pixel images and without any prior knowledge of configuration is shown for the first time. We build upon the success of recent deep reinforcement learning and develop a system for learning target reaching with a three-joint robot manipulator using external visual observation. A Deep Q Network (DQN) was demonstrated to perform target reaching after training in simulation. Transferring the network to real hardware and real observation in a naive approach failed, but experiments show that the network works when replacing camera images with synthetic images.

Turbulent Wall-Jets in Conical Diffusers

Abstract is not available.

Jerusalem - City Wall

Digital image

Jerusalem - City Wall

Digital image

Molecular motion in plastic crystals

Abstract is not available.

Unsteady Motion of a Stratified Rotating Viscous Fluid between Two Disks

Abstract is not available.

Uncanny nature: Hybrid stop motion using Blender

This talk gives an overview of the project "Uncanny Nature", which incoporates a style of animation called Hybrid Stop Motion, that combines physical object armatures with virtual copies. The development of the production pipeline (using a mix of Blender, Dragonframe, Photoscan and Arduino) is discussed, as well as the way that Blender was used throughout the production to visualise, model, animate and composite the elements together.

Inverse Problem for Fractional Brownian Motion with Discrete Data

The problem of recovering information from measurement data has already been studied for a long time. In the beginning, the methods were mostly empirical, but already towards the end of the sixties Backus and Gilbert started the development of mathematical methods for the interpretation of geophysical data. The problem of recovering information about a physical phenomenon from measurement data is an inverse problem. Throughout this work, the statistical inversion method is used to obtain a solution. Assuming that the measurement vector is a realization of fractional Brownian motion, the goal is to retrieve the amplitude and the Hurst parameter. We prove that under some conditions, the solution of the discretized problem coincides with the solution of the corresponding continuous problem as the number of observations tends to infinity. The measurement data is usually noisy, and we assume the data to be the sum of two vectors: the trend and the noise. Both vectors are supposed to be realizations of fractional Brownian motions, and the goal is to retrieve their parameters using the statistical inversion method. We prove a partial uniqueness of the solution. Moreover, with the support of numerical simulations, we show that in certain cases the solution is reliable and the reconstruction of the trend vector is quite accurate.

Non-cellulosic cell wall polysaccharides are subject to genotype × environment effects in sorghum (Sorghum bicolor) grain

Sorghum is a staple food for half a billion people and, through growth on marginal land with minimal inputs, is an important source of feed, forage and increasingly, biofuel feedstock. Here we present information about non-cellulosic cell wall polysaccharides in a diverse set of cultivated and wild Sorghum bicolor grains. Sorghum grain contains predominantly starch (64–76) but is relatively deficient in other polysaccharides present in wheat, oats and barley. Despite overall low quantities, sorghum germplasm exhibited a remarkable range in polysaccharide amount and structure. Total (1,3;1,4)-β-glucan ranged from 0.06 to 0.43 (w/w) whilst internal cellotriose:cellotetraose ratios ranged from 1.8 to 2.9:1. Arabinoxylan amounts fell between 1.5 and 3.6 (w/w) and the arabinose:xylose ratio, denoting arabinoxylan structure, ranged from 0.95 to 1.35. The distribution of these and other cell wall polysaccharides varied across grain tissues as assessed by electron microscopy. When ten genotypes were tested across five environmental sites, genotype (G) was the dominant source of variation for both (1,3;1,4)-β-glucan and arabinoxylan content (69–74), with environment (E) responsible for 5–14. There was a small G × E effect for both polysaccharides. This study defines the amount and spatial distribution of polysaccharides and reveals a significant genetic influence on cell wall composition in sorghum grain.

Experimental and numerical investigation of jet injection in a wall bounded supersonic flow

Numerical and experimental studies of a supersonic jet (Helium) inclined at 45 degrees to a oncoming Mach 2 flow have been carried out. The numerical study has been used to arrive at a geometry that could reduce an oncoming Mach 5.75 flow to Mach 2 flow and in determining the jet parameters. Experiments are carried out in the IISc. hypersonic shock tunnel HST2 at similar conditions obtained from numerical studies. Flow visualization studies carried out using Schlieren technique clearly show the presence of the bow shock in front of the jet exposed to supersonic cross flow. The jet Mach number is experimentally found to be approximate to 3. Visual observations show that the jet has penetrated up to 60% of the total height of the chamber.

Efficient eucalypt cell wall deconstruction and conversion for sustainable lignocellulosic biofuels

In order to meet the world’s growing energy demand and reduce the impact of greenhouse gas emissions resulting from fossil fuel combustion, renewable plant-based feedstocks for biofuel production must be considered. The first-generation biofuels, derived from starches of edible feedstocks, such as corn, create competition between food and fuel resources, both for the crop itself and the land on which it is grown. As such, biofuel synthesized from non-edible plant biomass (lignocellulose) generated on marginal agricultural land will help to alleviate this competition. Eucalypts, the broadly defined taxa encompassing over 900 species of Eucalyptus, Corymbia, and Angophora are the most widely planted hardwood tree in the world, harvested mainly for timber, pulp and paper, and biomaterial products. More recently, due to their exceptional growth rate and amenability to grow under a wide range of environmental conditions, eucalypts are a leading option for the development of a sustainable lignocellulosic biofuels. However, efficient conversion of woody biomass into fermentable monomeric sugars is largely dependent on pretreatment of the cell wall, whose formation and complexity lend itself toward natural recalcitrance against its efficient deconstruction. A greater understanding of this complexity within the context of various pretreatments will allow the design of new and effective deconstruction processes for bioenergy production. In this review, we present the various pretreatment options for eucalypts, including research into understanding structure and formation of the eucalypt cell wall.

Inventory of K.K.S.I. chest, receipted by Alexander Zuntz, with note by Isaac H. Levy

Digital image

Inventory of real estate on Wall Street

Digital image