CAT-GRAPH+ : towards odometry-driven place consolidation in changing environments

Changing environments present a number of challenges to mobile robots, one of the most significant being mapping and localisation. This problem is particularly significant in vision-based systems where illumination and weather changes can cause feature-based techniques to fail. In many applications only sections of an environment undergo extreme perceptual change. Some range-based sensor mapping approaches exploit this property by combining occasional place recognition with the assumption that odometry is accurate over short periods of time. In this paper, we develop this idea in the visual domain, by using occasional vision-driven loop closures to infer loop closures in nearby locations where visual recognition is difficult due to extreme change. We demonstrate successful map creation in an environment in which change is significant but constrained to one area, where both the vanilla CAT-Graph and a Sum of Absolute Differences matcher fails, use the described techniques to link dissimilar images from matching locations, and test the robustness of the system against false inferences.

Fourier Lucas-Kanade Algorithm

In this paper we propose a framework for both gradient descent image and object alignment in the Fourier domain. Our method centers upon the classical Lucas & Kanade (LK) algorithm where we represent the source and template/model in the complex 2D Fourier domain rather than in the spatial 2D domain. We refer to our approach as the Fourier LK (FLK) algorithm. The FLK formulation is advantageous when one pre-processes the source image and template/model with a bank of filters (e.g. oriented edges, Gabor, etc.) as: (i) it can handle substantial illumination variations, (ii) the inefficient pre-processing filter bank step can be subsumed within the FLK algorithm as a sparse diagonal weighting matrix, (iii) unlike traditional LK the computational cost is invariant to the number of filters and as a result far more efficient, and (iv) this approach can be extended to the inverse compositional form of the LK algorithm where nearly all steps (including Fourier transform and filter bank pre-processing) can be pre-computed leading to an extremely efficient and robust approach to gradient descent image matching. Further, these computational savings translate to non-rigid object alignment tasks that are considered extensions of the LK algorithm such as those found in Active Appearance Models (AAMs).

Influence of electrolyte cations on electron transport and electron transfer in dye-sensitized solar cells

The influence of different electrolyte cations ((Li+, Na+, Mg2+, tetrabutyl ammonium (TBA+)) on the TiO2 conduction band energy (Ec) the effective electron lifetime (τn), and the effective electron diffusion coefficient (Dn) in dye-sensitized solar cells (DSCs) was studied quantitatively. The separation between Ec and the redox Fermi level, EF,redox, was found to decrease as the charge/radius ratio of the cations increased. Ec in the Mg2+ electrolyte was found to be 170 meV lower than that in the Na+ electrolyte and 400 meV lower than that in the TBA+ electrolyte. Comparison of Dn and τn in the different electrolytes was carried out by using the trapped electron concentration as a measure of the energy difference between Ec and the quasi-Fermi level, nEF, under different illumination levels. Plots of Dn as a function of the trapped electron density, nt, were found to be relatively insensitive to the electrolyte cation, indicating that the density and energetic distribution of electron traps in TiO2 are similar in all of the electrolytes studied. By contrast, plots of τn versus nt for the different cations showed that the rate of electron back reaction is more than an order of magnitude faster in the TBA+ electrolyte compared with the Na+ and Li+ electrolytes. The electron diffusion lengths in the different electrolytes followed the sequence of Na+ > Li+ > Mg2+ > TBA+. The trends observed in the AM 1.5 current–voltage characteristics of the DSCs are rationalized on the basis of the conduction band shifts and changes in electron lifetime.

Temporal summation of intrinsically photosensitive retinal ganglion cell (ipRGC) contributions to the human pupil.

Purpose: IpRGCs mediate non-image forming functions including photoentrainment and the pupil light reflex (PLR). Temporal summation increases visual sensitivity and decreases temporal resolution for image forming vision, but the summation properties of nonimage forming vision are unknown. We investigated the temporal summation of inner (ipRGC) and outer (rod/cone) retinal inputs to the PLR. Method: The consensual PLR of the left eye was measured in six participants with normal vision using a Maxwellian view infrared pupillometer. Temporal summation was investigated using a double-pulse protocol (100 ms stimulus pairs; 0–1024 ms inter-stimulus interval, ISI) presented to the dilated fellow right eye (Tropicamide 1%). Stimulus lights (blue λmax = 460 nm; red λmax = 638 nm) biased activity to inneror outer retinal inputs to non-image forming vision. Temporal summation was measured suprathreshold (15.2 log photons.cm−2.s−1 at the cornea) and subthreshold (11.4 log photons.cm−2.s−1 at the cornea). Results: RM-ANOVAs showed the suprathreshold and subthreshold 6 second post illumination pupil response (PIPR: expressed as percentage baseline diameter) did not significantly vary for red or blue stimuli (p > .05). The PIPR for a subthreshold red 16 ms double-pulse control condition did not significantly differ with ISI (p > .05). The maximum constriction amplitude for red and blue 100 ms double- pulse stimuli did not significantly vary with ISI (p > .05). Conclusion: The non-significant changes in suprathreshold PIPR and subthreshold maximum pupil constriction indicate that inner retinal ipRGC inputs and outer retinal photoreceptor inputs to the PLR do not show temporal summation. The results suggest a fundamental difference between the temporal summation characteristics of image forming and non-image forming vision.

Every One, Every Day

Creative Statement: “There are those who see Planet Earth as a gigantic living being, one that feeds and nurtures humanity and myriad other species – an entity that must be cared for. Then there are those who see it as a rock full of riches to be pilfered heedlessly in a short-term quest for over-abundance. This ‘cradle to grave’ mentality, it would seem, is taking its toll (unless you’re a virulent disbeliever in climate change). Why not, ask artists Priscilla Bracks and Gavin Sade, take a different approach? To this end they have set out on a near impossible task; to visualise the staggering quantity of carbon produced by Australia every year. Their eerie, glowing plastic cube resembles something straight out of Dr Who or The X Files. And, like the best science fiction, it has technical realities at its heart. Every One, Every Day tangibly illustrates our greenhouse gas output – its 27m3 volume is approximately the amount of green-house gas emitted per capita, daily. Every One, Every Dayis lit by an array of LED’s displaying light patterns representing energy use generated by data from the Australian Energy Market. Every One, Every Day was formed from recycled, polyethylene – used milk bottles – ‘lent’ to the artists by a Visy recycling facility. At the end of the Vivid Festival this plastic will be returned to Visy, where it will re-enter the stream of ‘technical nutrients.’ Could we make another world? One that emulates the continuing cycles of nature? One that uses our ‘technical nutrients’ such as plastic and steel in continual cycles, just like a deciduous tree dropping leaves to compost itself and keep it’s roots warm and moist?” (Ashleigh Crawford. Melbourne – April, 2013) Artistic Research Statement: The research focus of this work is on exploring how to represent complex statistics and data at a human scale, and how produce a work where a large percentage of the materials could be recycled. The surface of Every One, Every Day is clad in tiles made from polyethylene, from primarily recycled milk bottles, ‘lent’ to the artists by the Visy recycling facility in Sydney. The tiles will be returned to Visy for recycling. As such the work can be viewed as an intervention in the industrial ecology of polyethylene, and in the process demonstrates how to sustain cycles of technical materials – by taking the output of a recycling facility back to a manufacturer to produce usable materials. In terms of data visualisation, Every One, Every Day takes the form of a cube with a volume of 27 cubic meters. The annual per capita emissions figures for Australia are cited as ranging between 18 to 25 tons. Assuming the lower figure, 18tons per capital annually, the 27 cubic meters represents approximately one day per capita of CO2 emissions – where CO2 is a gas at 15C and 1 atmosphere of pressure. The work also explores real time data visualisation by using an array of 600 controllable LEDs inside the cube. Illumination patterns are derived from a real time data from the Australian Energy Market, using the dispatch interval price and demand graph for New South Wales. The two variables of demand and price are mapped to properties of the illumination - hue, brightness, movement, frequency etc. The research underpinning the project spanned industrial ecology to data visualization and public art practices. The result is that Every One, Every Day is one of the first public artworks that successfully bring together materials, physical form, and real time data representation in a unified whole.

Blocking layer optimisation of poly(3-hexylthiopene)based solid state dye sensitized solar cells

The optimisation study of the fabrication of a compact TiO2 blocking layer (via Spray Pyrolysis Deposition) for poly (3-hexylthiopene) (P3HT) for Solid State Dye Sensitized Solar Cells (SDSCs) is reported. We used a novel spray TiO2 precursor solution composition obtained by adding acetylacetone to a conventional formulation (Diisopropoxytitanium bis (acetylacetonate) in ethanol). By Scanning Electron Microscopy a TiO2 layer with compact morphology and thickness of around 100 nmis shown. Through a Tafel plot analysis an enhancement of the device diode-like behaviour induced by the acetylacetone blocking layer respect to the conventional one is observed. Significantly, the device fabricatedwith the acetylacetone blocking layer shows an overall increment of the cell performance with respect to the cellwith the conventional one (DJsc/Jsc = +13.8%, DFF/FF = +39.7%, DPCE/PCE = +55.6%). A conversion efficiency optimumis found for 15 successive spray cycles where the diode-like behaviour of the acetylacetone blocking layer is more effective. Over three batches of cells (fabricated with P3HT and dye D35) an average conversion efficiency value of 3.9% (under a class A sun simulator with 1 sun A.M. 1.5 illumination conditions) was measured. From the best cell we fabricated a conversion efficiency value of 4.5% was extracted. This represents a significant increment with respect to previously reported values for P3HT/dye D35 based SDSCs.

Detection of dugongs from unmanned aerial vehicles

Monitoring and estimation of marine populations is of paramount importance for the conservation and management of sea species. Regular surveys are used to this purpose followed often by a manual counting process. This paper proposes an algorithm for automatic detection of dugongs from imagery taken in aerial surveys. Our algorithm exploits the fact that dugongs are rare in most images, therefore we determine regions of interest partially based on color rarity. This simple observation makes the system robust to changes in illumination. We also show that by applying the extended-maxima transform on red-ratio images, submerged dugongs with very fuzzy edges can be detected. Performance figures obtained here are promising in terms of degree of confidence in the detection of marine species, but more importantly our approach represents a significant step in automating this type of surveys.

Visual front-end wars : Viola-Jones face detector vs Fourier Lucas-Kanade

The performance of visual speech recognition (VSR) systems are significantly influenced by the accuracy of the visual front-end. The current state-of-the-art VSR systems use off-the-shelf face detectors such as Viola- Jones (VJ) which has limited reliability for changes in illumination and head poses. For a VSR system to perform well under these conditions, an accurate visual front end is required. This is an important problem to be solved in many practical implementations of audio visual speech recognition systems, for example in automotive environments for an efficient human-vehicle computer interface. In this paper, we re-examine the current state-of-the-art VSR by comparing off-the-shelf face detectors with the recently developed Fourier Lucas-Kanade (FLK) image alignment technique. A variety of image alignment and visual speech recognition experiments are performed on a clean dataset as well as with a challenging automotive audio-visual speech dataset. Our results indicate that the FLK image alignment technique can significantly outperform off-the shelf face detectors, but requires frequent fine-tuning.

Optimal catalyst thickness in titanium dioxide fixed film reactors: Mathematical modelling and experimental validation

The use of immobilised TiO2 for the purification of polluted water streams introduces the necessity to evaluate the effect of mechanisms such as the transport of pollutants from the bulk of the liquid to the catalyst surface and the transport phenomena inside the porous film. Experimental results of the effects of film thickness on the observed reaction rate for both liquid-side and support-side illumination are here compared with the predictions of a one-dimensional mathematical model of the porous photocatalytic slab. Good agreement was observed between the experimentally obtained photodegradation of phenol and its by-products, and the corresponding model predictions. The results have confirmed that an optimal catalyst thickness exists and, for the films employed here, is 5 μm. Furthermore, the modelling results have highlighted the fact that porosity, together with the intrinsic reaction kinetics are the parameters controlling the photocatalytic activity of the film. The former by influencing transport phenomena and light absorption characteristics, the latter by naturally dictating the rate of reaction.

Modelling of ellipsoidal nanowires : control and application

The properties of ellipsoidal nanowires are yet to be examined. They have likely applications in sensing, solar cells, microelectronics and cloaking devices. Little is known of the qualities that ellipse nanowires exhibit as we vary the aspect ratio with different dielectric materials and how varying these attributes affects plasmon coupling and propagation. It is known that the distance a plasmon can travel is further if it is supported by a thicker circular nanowire, while thinner nanowires are expected to be able to increase QD coupling. Ellipsoidal nanowires may be a good compromise due to their ability to have both thin and thick dimensions. Furthermore it has been shown that the plasmon resonances along the main axis of an ellipsoidal particle is governed by the relative aspect ratio of the ellipsoid, which may lead to further control of the plasmon. Research was done by the use of COMSOL Multiphysics by looking at the fundamental plasmon mode supported by an ellipsoidal nanowire and then studying this mode for various geometrical parameters, materials and illumination wavelength. Accordingly it was found that ellipsoidal nanowires exhibit a minimum for the wavenumber and a maximum for the propagation distance at roughly the same dimensions - Highlighting that there is an aspect ratio for which there is poor coupling but low loss. Here we investigate these and related attributes.

Odometry-driven inference to link multiple exemplars of a location

A major challenge for robot localization and mapping systems is maintaining reliable operation in a changing environment. Vision-based systems in particular are susceptible to changes in illumination and weather, and the same location at another time of day may appear radically different to a system using a feature-based visual localization system. One approach for mapping changing environments is to create and maintain maps that contain multiple representations of each physical location in a topological framework or manifold. However, this requires the system to be able to correctly link two or more appearance representations to the same spatial location, even though the representations may appear quite dissimilar. This paper proposes a method of linking visual representations from the same location without requiring a visual match, thereby allowing vision-based localization systems to create multiple appearance representations of physical locations. The most likely position on the robot path is determined using particle filter methods based on dead reckoning data and recent visual loop closures. In order to avoid erroneous loop closures, the odometry-based inferences are only accepted when the inferred path's end point is confirmed as correct by the visual matching system. Algorithm performance is demonstrated using an indoor robot dataset and a large outdoor camera dataset.

Local inter-session variability modelling for object classification

Object classification is plagued by the issue of session variation. Session variation describes any variation that makes one instance of an object look different to another, for instance due to pose or illumination variation. Recent work in the challenging task of face verification has shown that session variability modelling provides a mechanism to overcome some of these limitations. However, for computer vision purposes, it has only been applied in the limited setting of face verification. In this paper we propose a local region based intersession variability (ISV) modelling approach, and apply it to challenging real-world data. We propose a region based session variability modelling approach so that local session variations can be modelled, termed Local ISV. We then demonstrate the efficacy of this technique on a challenging real-world fish image database which includes images taken underwater, providing significant real-world session variations. This Local ISV approach provides a relative performance improvement of, on average, 23% on the challenging MOBIO, Multi-PIE and SCface face databases. It also provides a relative performance improvement of 35% on our challenging fish image dataset.

A bayesian approach for place recognition

This paper presents a robust place recognition algorithm for mobile robots that can be used for planning and navigation tasks. The proposed framework combines nonlinear dimensionality reduction, nonlinear regression under noise, and Bayesian learning to create consistent probabilistic representations of places from images. These generative models are incrementally learnt from very small training sets and used for multi-class place recognition. Recognition can be performed in near real-time and accounts for complexity such as changes in illumination, occlusions, blurring and moving objects. The algorithm was tested with a mobile robot in indoor and outdoor environments with sequences of 1579 and 3820 images, respectively. This framework has several potential applications such as map building, autonomous navigation, search-rescue tasks and context recognition.

Object tracking via non-Euclidean geometry : a Grassmann approach

A robust visual tracking system requires an object appearance model that is able to handle occlusion, pose, and illumination variations in the video stream. This can be difficult to accomplish when the model is trained using only a single image. In this paper, we first propose a tracking approach based on affine subspaces (constructed from several images) which are able to accommodate the abovementioned variations. We use affine subspaces not only to represent the object, but also the candidate areas that the object may occupy. We furthermore propose a novel approach to measure affine subspace-to-subspace distance via the use of non-Euclidean geometry of Grassmann manifolds. The tracking problem is then considered as an inference task in a Markov Chain Monte Carlo framework via particle filtering. Quantitative evaluation on challenging video sequences indicates that the proposed approach obtains considerably better performance than several recent state-of-the-art methods such as Tracking-Learning-Detection and MILtrack.

Matching image sets via adaptive multi convex hull

Traditional nearest points methods use all the samples in an image set to construct a single convex or affine hull model for classification. However, strong artificial features and noisy data may be generated from combinations of training samples when significant intra-class variations and/or noise occur in the image set. Existing multi-model approaches extract local models by clustering each image set individually only once, with fixed clusters used for matching with various image sets. This may not be optimal for discrimination, as undesirable environmental conditions (eg. illumination and pose variations) may result in the two closest clusters representing different characteristics of an object (eg. frontal face being compared to non-frontal face). To address the above problem, we propose a novel approach to enhance nearest points based methods by integrating affine/convex hull classification with an adapted multi-model approach. We first extract multiple local convex hulls from a query image set via maximum margin clustering to diminish the artificial variations and constrain the noise in local convex hulls. We then propose adaptive reference clustering (ARC) to constrain the clustering of each gallery image set by forcing the clusters to have resemblance to the clusters in the query image set. By applying ARC, noisy clusters in the query set can be discarded. Experiments on Honda, MoBo and ETH-80 datasets show that the proposed method outperforms single model approaches and other recent techniques, such as Sparse Approximated Nearest Points, Mutual Subspace Method and Manifold Discriminant Analysis.