Continuous appearance-based localisation and mapping

This thesis presents a novel approach to mobile robot navigation using visual information towards the goal of long-term autonomy. A novel concept of a continuous appearance-based trajectory is proposed in order to solve the limitations of previous robot navigation systems, and two new algorithms for mobile robots, CAT-SLAM and CAT-Graph, are presented and evaluated. These algorithms yield performance exceeding state-of-the-art methods on public benchmark datasets and large-scale real-world environments, and will help enable widespread use of mobile robots in everyday applications.

An evaluation of different features and learning models for anomalous event detection

The huge amount of CCTV footage available makes it very burdensome to process these videos manually through human operators. This has made automated processing of video footage through computer vision technologies necessary. During the past several years, there has been a large effort to detect abnormal activities through computer vision techniques. Typically, the problem is formulated as a novelty detection task where the system is trained on normal data and is required to detect events which do not fit the learned ‘normal’ model. There is no precise and exact definition for an abnormal activity; it is dependent on the context of the scene. Hence there is a requirement for different feature sets to detect different kinds of abnormal activities. In this work we evaluate the performance of different state of the art features to detect the presence of the abnormal objects in the scene. These include optical flow vectors to detect motion related anomalies, textures of optical flow and image textures to detect the presence of abnormal objects. These extracted features in different combinations are modeled using different state of the art models such as Gaussian mixture model(GMM) and Semi- 2D Hidden Markov model(HMM) to analyse the performances. Further we apply perspective normalization to the extracted features to compensate for perspective distortion due to the distance between the camera and objects of consideration. The proposed approach is evaluated using the publicly available UCSD datasets and we demonstrate improved performance compared to other state of the art methods.

Co-creating knowledge online : approaches for community artists

Forming peer alliances to share and build knowledge is an important aspect of community arts practice, and these co-creation processes are increasingly being mediated by the internet. This paper offers guidance for practitioners who are interested in better utilising the internet to connect, share, and make new knowledge. It argues that new approaches are required to foster the organising activities that underpin online co-creation, building from the premise that people have become increasingly networked as individuals rather than in groups (Rainie and Wellman 2012: 6), and that these new ways of connecting enable new modes of peer-to-peer production and exchange. This position advocates that practitioners move beyond situating the internet as a platform for dissemination and a tool for co-creating media, to embrace its knowledge collaboration potential. Drawing on a design experiment I developed to promote online knowledge co-creation, this paper suggests three development phases – developing connections, developing ideas, and developing agility – to ground six methods. They are: switching and routing, engaging in small trades of ideas with networked individuals; organising, co-ordinating networked individuals and their data; beta-release, offering ‘beta’ artifacts as knowledge trades; beta-testing, trialing and modifying other peoples ‘beta’ ideas; adapting, responding to technological disruption; and, reconfiguring, embracing opportunities offered by technological disruption. These approaches position knowledge co-creation as another capability of the community artist, along with co-creating art and media.

A clear view of costs ‘of and incidental to’ : Mio Art Pty Ltd v Macequest (No.2) Pty Ltd [2013] QSC 271

In Mio Art Pty Ltd v Macequest (No.2) Pty Ltd [2013] QSC 271 Jackson J provided considered analysis of several aspects of costs law. His Honour regarded various orders which are commonly sought or made as reflecting practice that is inappropriate or unnecessary under the Uniform Civil Procedure Rules 1999 (Qld) (UCPR).

Copyright and computer generated materials : is it time to reboot the discussion about authorship?

Computer generated materials are ubiquitous and we encounter them on a daily basis, even though most people are unaware that this is the case. Blockbuster movies, television weather reports and telephone directories all include material that is produced by utilising computer technologies. Copyright protection for materials generated by a programmed computer was considered by the Federal Court and Full Court of the Federal Court in Telstra Corporation Limited v Phone Directories Company Pty Ltd. The court held that the White and Yellow pages telephone directories produced by Telstra and its subsidiary, Sensis, were not protected by copyright because they were computer-generated works which lacked the requisite human authorship. The Copyright Act 1968 (Cth) does not contain specific provisions on the subsistence of copyright in computer-generated materials. Although the issue of copyright protection for computer-generated materials has been examined in Australia on two separate occasions by independently-constituted Copyright Law Review Committees over a period of 10 years (1988 to 1998), the Committees’ recommendations for legislative clarification by the enactment of specific amendments to the Copyright Act have not yet been implemented and the legal position remains unclear. In the light of the decision of the Full Federal Court in Telstra v Phone Directories it is timely to consider whether specific provisions should be enacted to clarify the position of computer-generated works under copyright law and, in particular, whether the requirement of human authorship for original works protected under Part III of the Copyright Act should now be reconceptualised to align with the realities of how copyright materials are created in the digital era.

Optimal camera planning under versatile user constraints in multi-camera image processing systems

The selection of optimal camera configurations (camera locations, orientations, etc.) for multi-camera networks remains an unsolved problem. Previous approaches largely focus on proposing various objective functions to achieve different tasks. Most of them, however, do not generalize well to large scale networks. To tackle this, we propose a statistical framework of the problem as well as propose a trans-dimensional simulated annealing algorithm to effectively deal with it. We compare our approach with a state-of-the-art method based on binary integer programming (BIP) and show that our approach offers similar performance on small scale problems. However, we also demonstrate the capability of our approach in dealing with large scale problems and show that our approach produces better results than two alternative heuristics designed to deal with the scalability issue of BIP. Last, we show the versatility of our approach using a number of specific scenarios.

Brain-inspired sensor fusion for navigating robots

Current state of the art robot mapping and navigation systems produce impressive performance under a narrow range of robot platform, sensor and environmental conditions, in contrast to animals such as rats that produce “good enough” maps that enable them to function under an incredible range of situations. In this paper we present a rat-inspired featureless sensor-fusion system that assesses the usefulness of multiple sensor modalities based on their utility and coherence for place recognition, without knowledge as to the type of sensor. We demonstrate the system on a Pioneer robot in indoor and outdoor environments with abrupt lighting changes. Through dynamic weighting of the sensors, the system is able to perform correct place recognition and mapping where the static sensor weighting approach fails.

Semi-binary based video features for activity representation

Efficient and effective feature detection and representation is an important consideration when processing videos, and a large number of applications such as motion analysis, 3D scene understanding, tracking etc. depend on this. Amongst several feature description methods, local features are becoming increasingly popular for representing videos because of their simplicity and efficiency. While they achieve state-of-the-art performance with low computational complexity, their performance is still too limited for real world applications. Furthermore, rapid increases in the uptake of mobile devices has increased the demand for algorithms that can run with reduced memory and computational requirements. In this paper we propose a semi binary based feature detectordescriptor based on the BRISK detector, which can detect and represent videos with significantly reduced computational requirements, while achieving comparable performance to the state of the art spatio-temporal feature descriptors. First, the BRISK feature detector is applied on a frame by frame basis to detect interest points, then the detected key points are compared against consecutive frames for significant motion. Key points with significant motion are encoded with the BRISK descriptor in the spatial domain and Motion Boundary Histogram in the temporal domain. This descriptor is not only lightweight but also has lower memory requirements because of the binary nature of the BRISK descriptor, allowing the possibility of applications using hand held devices.We evaluate the combination of detectordescriptor performance in the context of action classification with a standard, popular bag-of-features with SVM framework. Experiments are carried out on two popular datasets with varying complexity and we demonstrate comparable performance with other descriptors with reduced computational complexity.

Development of a computer based 1D-2D dynamic flood model. Case study of the Pagsangaan river basin in Leyte, the Philippines

Floods are among the most devastating events that affect primarily tropical, archipelagic countries such as the Philippines. With the current predictions of climate change set to include rising sea levels, intensification of typhoon strength and a general increase in the mean annual precipitation throughout the Philippines, it has become paramount to prepare for the future so that the increased risk of floods on the country does not translate into more economic and human loss. Field work and data gathering was done within the framework of an internship at the former German Technical Cooperation (GTZ) in cooperation with the Local Government Unit of Ormoc City, Leyte, The Philippines, in order to develop a dynamic computer based flood model for the basin of the Pagsangaan River. To this end, different geo-spatial analysis tools such as PCRaster and ArcGIS, hydrological analysis packages and basic engineering techniques were assessed and implemented. The aim was to develop a dynamic flood model and use the development process to determine the required data, availability and impact on the results as case study for flood early warning systems in the Philippines. The hope is that such projects can help to reduce flood risk by including the results of worst case scenario analyses and current climate change predictions into city planning for municipal development, monitoring strategies and early warning systems. The project was developed using a 1D-2D coupled model in SOBEK (Deltares Hydrological modelling software package) and was also used as a case study to analyze and understand the influence of different factors such as land use, schematization, time step size and tidal variation on the flood characteristics. Several sources of relevant satellite data were compared, such as Digital Elevation Models (DEMs) from ASTER and SRTM data, as well as satellite rainfall data from the GIOVANNI server (NASA) and field gauge data. Different methods were used in the attempt to partially calibrate and validate the model to finally simulate and study two Climate Change scenarios based on scenario A1B predictions. It was observed that large areas currently considered not prone to floods will become low flood risk (0.1-1 m water depth). Furthermore, larger sections of the floodplains upstream of the Lilo- an’s Bridge will become moderate flood risk areas (1 - 2 m water depth). The flood hazard maps created for the development of the present project will be presented to the LGU and the model will be used to create a larger set of possible flood prone areas related to rainfall intensity by GTZ’s Local Disaster Risk Management Department and to study possible improvements to the current early warning system and monitoring of the basin section belonging to Ormoc City; recommendations about further enhancement of the geo-hydro-meteorological data to improve the model’s accuracy mainly on areas of interest will also be presented at the LGU.

Small signal stability analysis of power systems with high penetration of wind power

The integration of large amount of wind power into a power system imposes a new challenge for the secure and economic operation of the system. It is necessary to investigate the impacts of wind power generation on the dynamic behavior of the power system concerned. This paper investigates the impacts of large amount of wind power on small signal stability and the corresponding control strategies to mitigate the negative effects. The concepts of different types of wind turbine generators (WTGs) and the principles of the grid-connected structures of wind power generation systems are first briefly introduced. Then, the state-of-the-art of the studies on the impacts of WTGs on small signal stability as well as potential problems to be studied are clarified. Finally, the control strategies on WTGs to enhance power system damping characteristics are presented.

Training-free probability models for whole-image based place recognition

Whole-image descriptors such as GIST have been used successfully for persistent place recognition when combined with temporal filtering or sequential filtering techniques. However, whole-image descriptor localization systems often apply a heuristic rather than a probabilistic approach to place recognition, requiring substantial environmental-specific tuning prior to deployment. In this paper we present a novel online solution that uses statistical approaches to calculate place recognition likelihoods for whole-image descriptors, without requiring either environmental tuning or pre-training. Using a real world benchmark dataset, we show that this method creates distributions appropriate to a specific environment in an online manner. Our method performs comparably to FAB-MAP in raw place recognition performance, and integrates into a state of the art probabilistic mapping system to provide superior performance to whole-image methods that are not based on true probability distributions. The method provides a principled means for combining the powerful change-invariant properties of whole-image descriptors with probabilistic back-end mapping systems without the need for prior training or system tuning.

Towards condition-invariant, top-down visual place recognition

In this paper we present a novel place recognition algorithm inspired by recent discoveries in human visual neuroscience. The algorithm combines intolerant but fast low resolution whole image matching with highly tolerant, sub-image patch matching processes. The approach does not require prior training and works on single images (although we use a cohort normalization score to exploit temporal frame information), alleviating the need for either a velocity signal or image sequence, differentiating it from current state of the art methods. We demonstrate the algorithm on the challenging Alderley sunny day – rainy night dataset, which has only been previously solved by integrating over 320 frame long image sequences. The system is able to achieve 21.24% recall at 100% precision, matching drastically different day and night-time images of places while successfully rejecting match hypotheses between highly aliased images of different places. The results provide a new benchmark for single image, condition-invariant place recognition.

It's just an old apron : aprons, artefacts and art - representations of personal and private 'her-stories', as public histories

This paper examines art and artefact in the representation and recollection of deeply personal WWII women’s experiences as POW’s under the Japanese. This kind of treatment of internees in the Tjideng Women and Children’s internment camp (and others) in Batavia under the Japanese in WWII, stands in stark and brutal contrast to the idyllic life lived by many families up to that time in what was then known as the Dutch East Indies (Indonesia). The deprivation and brutality of the Japanese incarceration of these women and children evoked responses - not military, but certainly militant, if muted. Representations of those responses – as both art and artefact - may be found in the most unlikely places and unexpected forms - and are still being unearthed to this day. However close we might personally be to these artists and artisans, can we, as observers from a distance, ever truly comprehend through spoken or written words alone, the day-today realities of those extraordinary times?

The art of co-creative media : an Australian survey

This article describes how - in the processes of responding to participatory storytelling practices - community, public service, and to a lesser extent, commercial media institutions are themselves negotiated and changed. Although there are significant variations in the conditions, durability, extent, motivations and quality of these developments and their impacts, they nonetheless increase the possibilities and pathways of participatory media culture. This description first frames digital storytelling as a ‘co-creative’ media practice. It then discusses the role of community arts and cultural development (CACD) practitioners and networks as co-creative media intermediaries, and then considers their influence in Australian broadcast and Internet media. It looks at how participatory storytelling methods are evolving in the Australian context and explores some of the implications for cultural inclusion arising from a shared interest in ‘co-creative’ media methods and approaches.

Improving PLDA speaker verification with limited development data

This paper analyses the probabilistic linear discriminant analysis (PLDA) speaker verification approach with limited development data. This paper investigates the use of the median as the central tendency of a speaker’s i-vector representation, and the effectiveness of weighted discriminative techniques on the performance of state-of-the-art length-normalised Gaussian PLDA (GPLDA) speaker verification systems. The analysis within shows that the median (using a median fisher discriminator (MFD)) provides a better representation of a speaker when the number of representative i-vectors available during development is reduced, and that further, usage of the pair-wise weighting approach in weighted LDA and weighted MFD provides further improvement in limited development conditions. Best performance is obtained using a weighted MFD approach, which shows over 10% improvement in EER over the baseline GPLDA system on mismatched and interview-interview conditions.