Image Region Salience for Improving Appearance-Based Place Recognition using a Supervised Classifier System

Many state of the art vision-based Simultaneous Localisation And Mapping (SLAM) and place recognition systems compute the salience of visual features in their environment. As computing salience can be problematic in radically changing environments new low resolution feature-less systems have been introduced, such as SeqSLAM, all of which consider the whole image. In this paper, we implement a supervised classifier system (UCS) to learn the salience of image regions for place recognition by feature-less systems. SeqSLAM only slightly benefits from the results of training, on the challenging real world Eynsham dataset, as it already appears to filter less useful regions of a panoramic image. However, when recognition is limited to specific image regions performance improves by more than an order of magnitude by utilising the learnt image region saliency. We then investigate whether the region salience generated from the Eynsham dataset generalizes to another car-based dataset using a perspective camera. The results suggest the general applicability of an image region salience mask for optimizing route-based navigation applications.

Towards brain-based 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 during a navigation task, 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.

Vertical infrastructure inspection using a quadcopter and shared autonomy control

This paper presents a shared autonomy control scheme for a quadcopter that is suited for inspection of vertical infrastructure — tall man-made structures such as streetlights, electricity poles or the exterior surfaces of buildings. Current approaches to inspection of such structures is slow, expensive, and potentially hazardous. Low-cost aerial platforms with an ability to hover now have sufficient payload and endurance for this kind of task, but require significant human skill to fly. We develop a control architecture that enables synergy between the ground-based operator and the aerial inspection robot. An unskilled operator is assisted by onboard sensing and partial autonomy to safely fly the robot in close proximity to the structure. The operator uses their domain knowledge and problem solving skills to guide the robot in difficult to reach locations to inspect and assess the condition of the infrastructure. The operator commands the robot in a local task coordinate frame with limited degrees of freedom (DOF). For instance: up/down, left/right, toward/away with respect to the infrastructure. We therefore avoid problems of global mapping and navigation while providing an intuitive interface to the operator. We describe algorithms for pole detection, robot velocity estimation with respect to the pole, and position estimation in 3D space as well as the control algorithms and overall system architecture. We present initial results of shared autonomy of a quadrotor with respect to a vertical pole and robot performance is evaluated by comparing with motion capture data.

Link discovery for Chinese/English cross-language web information retrieval

Nowadays people heavily rely on the Internet for information and knowledge. Wikipedia is an online multilingual encyclopaedia that contains a very large number of detailed articles covering most written languages. It is often considered to be a treasury of human knowledge. It includes extensive hypertext links between documents of the same language for easy navigation. However, the pages in different languages are rarely cross-linked except for direct equivalent pages on the same subject in different languages. This could pose serious difficulties to users seeking information or knowledge from different lingual sources, or where there is no equivalent page in one language or another. In this thesis, a new information retrieval task—cross-lingual link discovery (CLLD) is proposed to tackle the problem of the lack of cross-lingual anchored links in a knowledge base such as Wikipedia. In contrast to traditional information retrieval tasks, cross language link discovery algorithms actively recommend a set of meaningful anchors in a source document and establish links to documents in an alternative language. In other words, cross-lingual link discovery is a way of automatically finding hypertext links between documents in different languages, which is particularly helpful for knowledge discovery in different language domains. This study is specifically focused on Chinese / English link discovery (C/ELD). Chinese / English link discovery is a special case of cross-lingual link discovery task. It involves tasks including natural language processing (NLP), cross-lingual information retrieval (CLIR) and cross-lingual link discovery. To justify the effectiveness of CLLD, a standard evaluation framework is also proposed. The evaluation framework includes topics, document collections, a gold standard dataset, evaluation metrics, and toolkits for run pooling, link assessment and system evaluation. With the evaluation framework, performance of CLLD approaches and systems can be quantified. This thesis contributes to the research on natural language processing and cross-lingual information retrieval in CLLD: 1) a new simple, but effective Chinese segmentation method, n-gram mutual information, is presented for determining the boundaries of Chinese text; 2) a voting mechanism of name entity translation is demonstrated for achieving a high precision of English / Chinese machine translation; 3) a link mining approach that mines the existing link structure for anchor probabilities achieves encouraging results in suggesting cross-lingual Chinese / English links in Wikipedia. This approach was examined in the experiments for better, automatic generation of cross-lingual links that were carried out as part of the study. The overall major contribution of this thesis is the provision of a standard evaluation framework for cross-lingual link discovery research. It is important in CLLD evaluation to have this framework which helps in benchmarking the performance of various CLLD systems and in identifying good CLLD realisation approaches. The evaluation methods and the evaluation framework described in this thesis have been utilised to quantify the system performance in the NTCIR-9 Crosslink task which is the first information retrieval track of this kind.

Regression and hypothesis tests for multivariate GNSS state time series

A satellite based observation system can continuously or repeatedly generate a user state vector time series that may contain useful information. One typical example is the collection of International GNSS Services (IGS) station daily and weekly combined solutions. Another example is the epoch-by-epoch kinematic position time series of a receiver derived by a GPS real time kinematic (RTK) technique. Although some multivariate analysis techniques have been adopted to assess the noise characteristics of multivariate state time series, statistic testings are limited to univariate time series. After review of frequently used hypotheses test statistics in univariate analysis of GNSS state time series, the paper presents a number of T-squared multivariate analysis statistics for use in the analysis of multivariate GNSS state time series. These T-squared test statistics have taken the correlation between coordinate components into account, which is neglected in univariate analysis. Numerical analysis was conducted with the multi-year time series of an IGS station to schematically demonstrate the results from the multivariate hypothesis testing in comparison with the univariate hypothesis testing results. The results have demonstrated that, in general, the testing for multivariate mean shifts and outliers tends to reject less data samples than the testing for univariate mean shifts and outliers under the same confidence level. It is noted that neither univariate nor multivariate data analysis methods are intended to replace physical analysis. Instead, these should be treated as complementary statistical methods for a prior or posteriori investigations. Physical analysis is necessary subsequently to refine and interpret the results.

Investigating effective wayfinding in airports : a Bayesian network approach

Effective Wayfinding is the successful interplay of human and environmental factors resulting in a person successfully moving from their current position to a desired location in a timely manner. To date this process has not been modelled to reflect this interplay. This paper proposes a complex modelling system approach of wayfinding by using Bayesian Networks to model this process, and applies the model to airports. The model suggests that human factors have a greater impact on effective wayfinding in airports than environmental factors. The greatest influences on human factors are found to be the level of spatial anxiety experienced by travellers and their cognitive and spatial skills. The model also predicted that the navigation pathway that a traveller must traverse has a larger impact on the effectiveness of an airport’s environment in promoting effective wayfinding than the terminal design.

Examining the effects of an eco-driving system on driver distraction

Driver distraction has recently been defined by Regan as "the diversion of attention away from activities critical for safe driving toward a competing activity, which may result in insufficient or no attention to activities critical for safe driving (Regan, Hallett & Gordon, 2011, p.1780)". One source of distraction is in-vehicle devices, even though they might provide other benefits, e.g. navigation systems. Currently, eco-driving systems have been growing rapidly in popularity. These systems send messages to drivers so that driving performance can be improved in terms of fuel efficiency. However, there remain unanswered questions about whether eco-driving systems endanger drivers by distracting them. In this research, the CARRS-Q advanced driving simulator was used in order to provide safety for participants and meanwhile simulate real world driving. The distraction effects of tasks involving three different in-vehicle systems were investigated: changing a CD, entering a five digit number as a part of navigation task and responding to an eco-driving task. Driving in these scenarios was compared with driving in the absence of these distractions, and while drivers engaged in critical manoeuvres. In order to account for practice effects, the same scenarios were duplicated on a second day. The three in-vehicle systems were not the exact facsimiles of any particular existing system, but were designed to have similar characteristics to those of system available. In general, the results show that drivers’ mental workloads are significantly higher in navigation and CD changing scenarios in comparison to the two other scenarios, which implies that these two tasks impose more visual/manual and cognitive demands. However, eco-driving mental workload is still high enough to be called marginally significant (p ~ .05) across manoeuvres. Similarly, event detection tasks show that drivers miss significantly more events in the navigation and CD changing scenarios in comparison to both the baseline and eco-driving scenario across manoeuvres. Analysis of the practice effect shows that drivers’ baseline scenario and navigation scenario exhibit significantly less demand on the second day. However, the number of missed events across manoeuvres confirmed that drivers can detect significantly more events on the second day for all scenarios. Distraction was also examined separately for five groups of manoeuvres (straight, lane changing, overtaking, braking for intersections and braking for roundabouts), in two locations for each condition. Repeated measures mixed ANOVA results show that reading an eco-driving message can potentially impair driving performance. When comparing the three in–vehicle distractions tested, attending to an eco-driving message is similar in effect to the CD changing task. The navigation task degraded driver performance much more than these other sources of distraction. In lane changing manoeuvres, drivers’ missed response counts degraded when they engaged in reading eco-driving messages at the first location. However, drivers’ event detection abilities deteriorated less at the second lane changing location. In baseline manoeuvres (driving straight), participants’ mean minimum speed degraded more in the CD changing scenario. Drivers’ lateral position shifted more in both CD changing and navigation tasks in comparison with both eco-driving and baseline scenarios, so they were more visually distracting. Participants were better at event detection in baseline manoeuvres in comparison with other manoeuvres. When approaching an intersection, the navigation task caused more events to be missed by participants, whereas eco-driving messages seemed to make drivers less distracted. The eco-driving message scenario was significantly less distracting than the navigation system scenario (fewer missed responses) when participants commenced braking for roundabouts. To sum up, in spite of the finding that two other in-vehicle tasks are more distracting than the eco-driving task, the results indicate that even reading a simple message while driving could potentially lead to missing an important event, especially when executing critical manoeuvres. This suggests that in-vehicle eco-driving systems have the potential to contribute to increased crash risk through distraction. However, there is some evidence of a practice effect which suggests that future research should focus on performance with habitual rather than novel tasks. It is recommended that eco-driving messages be delivered to drivers off-line when possible.

Analysis of regionally enhanced GPS orbit and clock solutions and contribution to improvement of real-time precise point positioning

This work experimentally examines the performance benefits of a regional CORS network to the GPS orbit and clock solutions for supporting real-time Precise Point Positioning (PPP). The regionally enhanced GPS precise orbit solutions are derived from a global evenly distributed CORS network added with a densely distributed network in Australia and New Zealand. A series of computational schemes for different network configurations are adopted in the GAMIT-GLOBK and PANDA data processing. The precise GPS orbit results show that the regionally enhanced solutions achieve the overall orbit improvements with respect to the solutions derived from the global network only. Additionally, the orbital differences over GPS satellite arcs that are visible by any of the five Australia-wide CORS stations show a higher percentage of overall improvements compared to the satellite arcs that are not visible from these stations. The regional GPS clock and Uncalibrated Phase Delay (UPD) products are derived using the PANDA real time processing module from Australian CORS networks of 35 and 79 stations respectively. Analysis of PANDA kinematic PPP and kinematic PPP-AR solutions show certain overall improvements in the positioning performance from a denser network configuration after solution convergence. However, the clock and UPD enhancement on kinematic PPP solutions is marginal. It is suggested that other factors, such as effects of ionosphere, incorrectly fixed ambiguities, may be the more dominating, deserving further research attentions.

Long summer days : grounded learning of words for the uneven cycles of real world events

Time and space are fundamental to human language and embodied cognition. In our early work we investigated how Lingodroids, robots with the ability to build their own maps, could evolve their own geopersonal spatial language. In subsequent studies we extended the framework developed for learning spatial concepts and words to learning temporal intervals. This paper considers a new aspect of time, the naming of concepts like morning, afternoon, dawn, and dusk, which are events that are part of day-night cycles, but are not defined by specific time points on a clock. Grounding of such terms refers to events and features of the diurnal cycle, such as light levels. We studied event-based time in which robots experienced day-night cycles that varied with the seasons throughout a year. Then we used meet-at tasks to demonstrate that the words learned were grounded, where the times to meet were morning and afternoon, rather than specific clock times. The studies show how words and concepts for a novel aspect of cyclic time can be grounded through experience with events rather than by times as measured by clocks or calendars

Faster Monte Carlo simulation of radiotherapy geometries

Using Monte Carlo simulation for radiotherapy dose calculation can provide more accurate results when compared to the analytical methods usually found in modern treatment planning systems, especially in regions with a high degree of inhomogeneity. These more accurate results acquired using Monte Carlo simulation however, often require orders of magnitude more calculation time so as to attain high precision, thereby reducing its utility within the clinical environment. This work aims to improve the utility of Monte Carlo simulation within the clinical environment by developing techniques which enable faster Monte Carlo simulation of radiotherapy geometries. This is achieved principally through the use new high performance computing environments and simpler alternative, yet equivalent representations of complex geometries. Firstly the use of cloud computing technology and it application to radiotherapy dose calculation is demonstrated. As with other super-computer like environments, the time to complete a simulation decreases as 1=n with increasing n cloud based computers performing the calculation in parallel. Unlike traditional super computer infrastructure however, there is no initial outlay of cost, only modest ongoing usage fees; the simulations described in the following are performed using this cloud computing technology. The definition of geometry within the chosen Monte Carlo simulation environment - Geometry & Tracking 4 (GEANT4) in this case - is also addressed in this work. At the simulation implementation level, a new computer aided design interface is presented for use with GEANT4 enabling direct coupling between manufactured parts and their equivalent in the simulation environment, which is of particular importance when defining linear accelerator treatment head geometry. Further, a new technique for navigating tessellated or meshed geometries is described, allowing for up to 3 orders of magnitude performance improvement with the use of tetrahedral meshes in place of complex triangular surface meshes. The technique has application in the definition of both mechanical parts in a geometry as well as patient geometry. Static patient CT datasets like those found in typical radiotherapy treatment plans are often very large and present a significant performance penalty on a Monte Carlo simulation. By extracting the regions of interest in a radiotherapy treatment plan, and representing them in a mesh based form similar to those used in computer aided design, the above mentioned optimisation techniques can be used so as to reduce the time required to navigation the patient geometry in the simulation environment. Results presented in this work show that these equivalent yet much simplified patient geometry representations enable significant performance improvements over simulations that consider raw CT datasets alone. Furthermore, this mesh based representation allows for direct manipulation of the geometry enabling motion augmentation for time dependant dose calculation for example. Finally, an experimental dosimetry technique is described which allows the validation of time dependant Monte Carlo simulation, like the ones made possible by the afore mentioned patient geometry definition. A bespoke organic plastic scintillator dose rate meter is embedded in a gel dosimeter thereby enabling simultaneous 3D dose distribution and dose rate measurement. This work demonstrates the effectiveness of applying alternative and equivalent geometry definitions to complex geometries for the purposes of Monte Carlo simulation performance improvement. Additionally, these alternative geometry definitions allow for manipulations to be performed on otherwise static and rigid geometry.

Long exposure localization in darkness using consumer cameras

In this paper we demonstrate passive vision-based localization in environments more than two orders of magnitude darker than the current benchmark using a 100 webcam and a 500 camera. Our approach uses the camera’s maximum exposure duration and sensor gain to achieve appropriately exposed images even in unlit night-time environments, albeit with extreme levels of motion blur. Using the SeqSLAM algorithm, we first evaluate the effect of variable motion blur caused by simulated exposures of 132 ms to 10000 ms duration on localization performance. We then use actual long exposure camera datasets to demonstrate day-night localization in two different environments. Finally we perform a statistical analysis that compares the baseline performance of matching unprocessed greyscale images to using patch normalization and local neighbourhood normalization – the two key SeqSLAM components. Our results and analysis show for the first time why the SeqSLAM algorithm is effective, and demonstrate the potential for cheap camera-based localization systems that function across extreme perceptual change.

Eye on the Street? Sensory Experience in Public Places

Research on the sensual experience of place is not a mainstream topic in the architectural debate; it is more common in other disciplines like landscape architecture or interior design. The curriculum sometime offers opportunities of cross-pollination between disciplines; students in architecture courses might be exposed to different theories of space more typical to other fields. This paper explore the teaching/research nexus within QUT Master of Architecture research stream; the focus of the discussion is students’ experimentation with people’s experience and navigation of the public space. Theories of placemaking in relation to urban design are first introduced; then the teaching/research nexus is discussed; finally students’ experience in approaching phenomenological research within the Master of Architecture are presented.

[Book Review] Urban Maps : Instruments of Narrative and Interpretation in the City by Richard Brook and Nick Dunn, Ashgate, Farham, United Kingdom, 2011, 250pp. ISBN 978-0-754676-5-77.

Urban maps discusses new ways and tools to read and navigate the contemporary city. Each chapter investigates a possible approach to unravel the complexity of contemporary urban forms. Each tool is first defined, introducing its philosophical background, and is then discussed with case studies, showing its relevance for the navigation of the built environment. Urbanism classics such as the work of Lynch, Jacobs, Venuti and Scott-Brown, Lefebrve and Walter Benjamin are fundamental in setting the framework of the volume. In the introduction cities and mapping are first discussed, the former are illustrated as ‘a composite of invisible networks devoid of landmarks and overrun by nodes’ (p. 3), and ‘a series of unbounded spaces where mass production and mass consumption reproduce a standardised quasi-global culture’ (p. 6).

Fixed failure rate ambiguity validation methods for GPS and compass

Ambiguity resolution plays a crucial role in real time kinematic GNSS positioning which gives centimetre precision positioning results if all the ambiguities in each epoch are correctly fixed to integers. However, the incorrectly fixed ambiguities can result in large positioning offset up to several meters without notice. Hence, ambiguity validation is essential to control the ambiguity resolution quality. Currently, the most popular ambiguity validation is ratio test. The criterion of ratio test is often empirically determined. Empirically determined criterion can be dangerous, because a fixed criterion cannot fit all scenarios and does not directly control the ambiguity resolution risk. In practice, depending on the underlying model strength, the ratio test criterion can be too conservative for some model and becomes too risky for others. A more rational test method is to determine the criterion according to the underlying model and user requirement. Miss-detected incorrect integers will lead to a hazardous result, which should be strictly controlled. In ambiguity resolution miss-detected rate is often known as failure rate. In this paper, a fixed failure rate ratio test method is presented and applied in analysis of GPS and Compass positioning scenarios. A fixed failure rate approach is derived from the integer aperture estimation theory, which is theoretically rigorous. The criteria table for ratio test is computed based on extensive data simulations in the approach. The real-time users can determine the ratio test criterion by looking up the criteria table. This method has been applied in medium distance GPS ambiguity resolution but multi-constellation and high dimensional scenarios haven't been discussed so far. In this paper, a general ambiguity validation model is derived based on hypothesis test theory, and fixed failure rate approach is introduced, especially the relationship between ratio test threshold and failure rate is examined. In the last, Factors that influence fixed failure rate approach ratio test threshold is discussed according to extensive data simulation. The result shows that fixed failure rate approach is a more reasonable ambiguity validation method with proper stochastic model.

Achieving high reliability for ambiguity resolutions with multiple GNSS constellations

Global Navigation Satellite Systems (GNSS)-based observation systems can provide high precision positioning and navigation solutions in real time, in the order of subcentimetre if we make use of carrier phase measurements in the differential mode and deal with all the bias and noise terms well. However, these carrier phase measurements are ambiguous due to unknown, integer numbers of cycles. One key challenge in the differential carrier phase mode is to fix the integer ambiguities correctly. On the other hand, in the safety of life or liability-critical applications, such as for vehicle safety positioning and aviation, not only is high accuracy required, but also the reliability requirement is important. This PhD research studies to achieve high reliability for ambiguity resolution (AR) in a multi-GNSS environment. GNSS ambiguity estimation and validation problems are the focus of the research effort. Particularly, we study the case of multiple constellations that include initial to full operations of foreseeable Galileo, GLONASS and Compass and QZSS navigation systems from next few years to the end of the decade. Since real observation data is only available from GPS and GLONASS systems, the simulation method named Virtual Galileo Constellation (VGC) is applied to generate observational data from another constellation in the data analysis. In addition, both full ambiguity resolution (FAR) and partial ambiguity resolution (PAR) algorithms are used in processing single and dual constellation data. Firstly, a brief overview of related work on AR methods and reliability theory is given. Next, a modified inverse integer Cholesky decorrelation method and its performance on AR are presented. Subsequently, a new measure of decorrelation performance called orthogonality defect is introduced and compared with other measures. Furthermore, a new AR scheme considering the ambiguity validation requirement in the control of the search space size is proposed to improve the search efficiency. With respect to the reliability of AR, we also discuss the computation of the ambiguity success rate (ASR) and confirm that the success rate computed with the integer bootstrapping method is quite a sharp approximation to the actual integer least-squares (ILS) method success rate. The advantages of multi-GNSS constellations are examined in terms of the PAR technique involving the predefined ASR. Finally, a novel satellite selection algorithm for reliable ambiguity resolution called SARA is developed. In summary, the study demonstrats that when the ASR is close to one, the reliability of AR can be guaranteed and the ambiguity validation is effective. The work then focuses on new strategies to improve the ASR, including a partial ambiguity resolution procedure with a predefined success rate and a novel satellite selection strategy with a high success rate. The proposed strategies bring significant benefits of multi-GNSS signals to real-time high precision and high reliability positioning services.