Search

Centimetre level precision mapping is essential for many applications such as land-use, infrastructure inspection, cultural heritage preservation, and construction site monitoring. However, the acquisition and its preparation (in particular the setting of a ground control point network (GCPs)) are still expensive or even impossible in cluttered or dangerous areas. The recent development of UAVs together with the miniaturization of the sensors is a promising evolution for reducing costs and expand opportunities. The sensors embedded on the drone: GNSS antenna, IMU, camera and (optional) LIDAR are light and often low-cost. The low quality of their raw measurements must be counterbalanced by their rigorous modeling in order to obtain accurate final results: if we cannot expect the sensors to be error-free, one must model these in order to correct them. This is achieved by in-situ calibration or on a dedicated calibration field, together with a rigorous fusion of the raw data acquired by the different sensors with the so-called bundle-adjustment method. This thesis proposes several models to describe the behavior of the sensors, in order to hybridize them rigorously in the bundle-adjustment. Consistent datasets have been acquired on the field specifically to assess the relevance of both the sensor models and their hybridizing in complex photogrammetric processing. The contribution of this thesis could be divided into two mains categories. On one hand, this thesis suggests tools and recommendation to improve directly the procedures achieved by end-users using current UAV-mapping commercial solutions (in particular for the GCPs placement, for the choice of the camera calibration and model and for the flight-plan). On the other hand, this thesis put forward exotic methods (methods considered as exotic at the time of the writing of the thesis) such as Photo-LIDAR hybridizing and collaborative mapping achieved by a terrestrial-aerial tandem (a terrestrial vehicle holding a LIDAR, GNSS, imaging and inertial sensors followed by a drone conceived to proceed to airborne photogrammetry) or an aerial-aerial tandem (two drones flying in formation to proceed to airborne photogrammetry). The contribution of this thesis will permit to reduce costs, to improve the quality of mapping products and to enlarge the possibilities of mapping: in particular, map cluttered or inaccessible zones which are nowadays considered as difficult or even impossible to map.

Traffic parameter forecasting is critical to effective traffic management but is a challenging task due to the stochasticity of traffic flow characteristics, especially in urban road networks. Traffic networks can be affected by external factors, such as weather, events, accidents, and road construction works. The impact of these factors can affect traffic flow parameters by influencing travel time, density, occupancy, and operating speed. Although deep neural networks (DNNs) have recently shown promising signs in traffic prediction using big data, there still exists the issue of maximizing the use of the model capabilities by using big data sources. This paper proposes an improved urban traffic speed prediction approach involving input-level data fusion and deep learning. Motivated by deep learning prediction methods, we propose a Long Short-Term Memory Neural Network (LSTM-NN) for traffic speed prediction that combines traffic and weather datasets on an urban road network in Greater Manchester, United Kingdom. The experimental results substantiate the value of the approach when compared to the use of traffic-only data sources for traffic speed prediction.

International audience; Near infrared spectroscopy and thermogravimetry have been coupled with chemometric exploratory methods in order to investigate ancient (pre-Roman/Roman) human bones from two different necropolises in Central-South Italy (Cavo degli Zucchi and Elea Velia). These findings have been investigated by principal component analysis and they have also been compared with ancient human bones from two Sudanese necropolises (Saggai and Geili). Samples coming from African and European necropolises, mainly differ in two aspects: the burial procedures and their historical period. The ritual applied in the European region involved cremation, while the one applied in the African necropolises did not. Bones from Italian sites (Cavo degli Zucchi and Elea Velia) are Pre-Roman/Roman while the others (from middle Nile) come from the Prehistoric, Meroitic, and Christian Sudanese age. Near infrared spectroscopy and thermogravimetric measures have been analysed either individually or by a mid-level data-fusion approach. Principal component analysis of the near infrared spectroscopy data allowed differentiation between burnt and unburnt samples, while from the scores plots extracted from the principal component analysis model based on the entire derived thermograms, it was possible to recognize the different clusters related to the various dating of samples. The data-fusion analysis led to considerations similar to those obtained from the model based on thermogravimetry data. Finally, instead of inspecting the entire thermogravimetry curves, principal component analysis was carried out on carbonates, total collagen and water losses only. In this case, the data-fusion approach has led to extremely interesting results; in fact, this model clearly shows that samples group in separate clusters in agreement with their age and the different burial rituals.

This study investigates how two existing pan-tropical above-ground biomass (AGB) maps (Saatchi 2011, Baccini 2012) can be combined to derive forest ecosystem specific carbon estimates. Several data-fusion models which combine these AGB maps according to their local correlations with independent datasets such as the spectral bands of SPOT VEGETATION imagery are analyzed. Indeed these spectral bands convey information about vegetation type and structure which can be related to biomass values. Our study area is the island of Borneo. The data-fusion models are evaluated against a reference AGB map available for two forest concessions in Sabah. The highest accuracy was achieved by a model which combines the AGB maps according to the mean of the local correlation coefficients calculated over different kernel sizes. Combining the resulting AGB map with a new Borneo land cover map (whose overall accuracy has been estimated at 86.5%) leads to average AGB estimates of 279.8 t/ha and 233.1 t/ha for forests and degraded forests respectively. Lowland dipterocarp and mangrove forests have the highest and lowest AGB values (305.8 t/ha and 136.5 t/ha respectively). The AGB of all natural forests amounts to 10.8 Gt mainly stemming from lowland dipterocarp (66.4%), upper dipterocarp (10.9%) and peat swamp forests (10.2%). Degraded forests account for another 2.1 Gt of AGB. One main advantage of our approach is that, once the best fitting data-fusion model is selected, no further AGB reference dataset is required for implementing the data-fusion process. Furthermore, the local harmonization of AGB datasets leads to more spatially precise maps. This approach can easily be extended to other areas in Southeast Asia which are dominated by lowland dipterocarp forest, and can be repeated when newer or more accurate AGB maps become available.

On-board electrical power demands in modern aircraft are substantially increasing. Scaling-up the size of the current starter/generators to provide the additional power requirements inevitably increases their mass. Instead, aircraft electrical designers are considering to increase the rotational speed of these machines. This imposes severe loading demands in the current starter/generator bearings. Hybrid bearings offer the most potential to deal with these demands. However, not much is known about their wear behavior in this new application. Our research is assessing the degradation of hybrid ceramic bearings in the newest generation of starter/generators for condition monitoring and health management. This paper discusses bearing degradation as foundation for the definition of relevant health condition assessment and decision making approach, and the integration of a monitoring system into a prototype test bearing to be used for data generation and algorithm validation. Initial results indicate this approach can effectively diagnose bearings faults. Validation was conducted by using repository data.

To be reliable in outdoor operations, perception systems should use a combination of sensing modalities, such as laser, radar, visual camera, or IR camera, since they respond differently to distinct environmental conditions. In this thesis, a novel multisensory data fusion framework designed to appropriately combine data acquired by multiple exteroceptive sensing modalities in field robotics scenarios is presented. Compared with the traditional data fusion methods, the proposed framework will account for the fact that sensors perceive the environment in different ways. To this end, it is proposed to automatically detect commonalities and discrepancies between the data sources before performing data fusion or combination. The consistency test evaluates dependencies between data from multiple-sensing-modalities or representations and chooses intelligently the data that most likely leads to synergy. The data fusion methods described in this thesis create high fidelity representations taking advantage of the full potential of each sensing modality leading to resilient perception system. This thesis introduces a data fusion with a parametric consistency test framework for multiple sensing modalities using Gaussian process data fusion. This approach avoids the data quality degradation inherent to fusing multiple types of sensor data by using a parametric consistency test. In addition, a non-parametric consistency test for multiple-sensing-modalities using Gaussian process data fusion is presented. This approach avoids local geometric threshold parameters and can be more discriminatory because it takes into account the global model. An experimental analysis was evaluated for simulations and for real-world implementation using laser and radar sensor in different environmental conditions. More specifically the performance of 3D surface reconstruction in the context of field robotics was tested in a variety of scenarios. It was demonstrated that by identifying discrepancies and exploiting data commonalities between the data of each sensing modality the framework creates a resilient perception system to adverse environmental conditions.

Foodomics is a newly developed discipline that has become more and more important in the last years where focus on food and the understanding of food systems has increased significantly. In this review, the flow of a typical foodomics study will be followed with a focus on the core components, where chemometric expertise is more deeply involved. These are: how to acquire sound data, how to exploit an experimental design, how to use classification in a proper way, how to look at more analytical platforms at the same time and, not the least, how to understand the limitations when interpreting the developed models. For each of these phases, the most common data issues will be highlighted and some of the most recent chemometric methods that are able to help solving them, will be presented.

During must fermentation by Saccharomyces cerevisiae strains thousands of volatile aroma compounds are formed. The objective of the present work was to adapt computational approaches to analyze pheno-metabolomic diversity of a S. cerevisiae strain collection with different origins. Phenotypic and genetic characterization together with individual must fermentations were performed, and metabolites relevant to aromatic profiles were determined. Experimental results were projected onto a common coordinates system, revealing 17 statistical-relevant multi-dimensional modules, combining sets of most-correlated features of noteworthy biological importance. The present method allowed, as a breakthrough, to combine genetic, phenotypic and metabolomic data, which has not been possible so far due to difficulties in comparing different types of data. Therefore, the proposed computational approach revealed as successful to shed light into the holistic characterization of S. cerevisiae pheno-metabolome in must fermentative conditions. This will allow the identification of combined relevant features with application in selection of good winemaking strains., Ines Mendes was recipient of a fellowship from the Portuguese Science Foundation, FCT (SFRH/BD/74798/2010). This work was supported by FCT I.P. through the strategic funding UID/BIA/04050/2013, and the project PTDC/AGR-ALI/121062/2010.

International audience; This paper exposes the Navigation and Control technology embedded in a recently commercialized micro Unmanned Air Vehicle (UAV), the AR.Drone, which cost and performance are unprecedented among any commercial product for mass markets. The system relies on state-of-the-art indoor navigation systems combining low-cost inertial sensors, computer vision techniques, sonar, and accounting for aerodynamics models.

Microarray technology has had a significant impact in the field of systems biology involving the investigation into the biological systems that regulate human life. Identifying genes of significant interest within any given disease on an individual basis is no doubt time consuming and inefficient when considering the complexity of the human genome. Thus, the genetic profiling of the entire human genome in a single experiment has resulted in microarray technology becoming a widely used experimental tool. However, without the use of tools for several aspects of microarray data analysis the technology is limited. To date, no such tool has been developed that allows the integration of numerous microarray results from different research laboratories as well as the design of customised gene chips in a cost-effective manner. In light of this, we have designed the first integrated and automated software called Chip Integration, Design and Annotation (CIDA) for the cross comparison, design and functional annotation of microarray gene chips. The software provides molecular biologists with the control to cross compare the biological signatures generated from multiple microarray studies, design custom microarray gene chips based on their research requirements and lastly characterise microarray data in the context of immunogenomics. Through the relative comparison of related microarray experiments we have identified 258 genes with common gene expression profiles that are not only upregulated in anergic T cells, but also in cells over-expressing the transcription factor Egr2, that has been identified to play a role in T cell anergy. Using the gene chip design aspect of CIDA we have designed and subsequently fabricate immuno-tolerance gene chips consisting of 1758 genes for further research.The software and database schema is freely available at ftp://ftp.brunel.ac.uk/cspgssk/CIDA/. Additional material is available online at http://www.brunel.ac.uk/about/acad/health/healthres/researchgroups/mi/publications/supplementary/cida

The use of damage indicators based on changes of the systems’ dynamic properties before and after the occurrence of damage, generally represents a good tool in detecting the presence, the position and also the severity of the damage itself. However, signal noises, difficulties to excite systems for identifying the significant modes of vibrations, particularly the ones more sensible to the damage scenarios and also the presence of multiple damage locations can particularly affect the efficiency of these indicators. In recent years, new algorithms and innovative techniques devoted to improve the efficiency of the classical damage indicators have been carried out in many studies. In this context, the paper presents an approach for damage identification of linear systems developed by combining the use of classical damage indicators based on the modal strain energy (MSE) with a multi-stage data-fusion procedure. In particular, modal strain energy change ratios (MSECR) are evaluated by accounting different sets of modes of vibration. Then, the obtained MSECRs are converted in local decisions and involved in a multi-stage data-fusion process which provides indices able to detect the damage location and its extent. The approach provides a significant improvement of the efficiency of the damage indicators particularly in the presence of noises and multiple damages as shown by the numerical applications reported in the paper.

Nowadays the necessity to reveal the hidden information from complex data sets is increasing due to the development of high-throughput instrumentation. The possibility to jointly analyze data sets arising from different sources (e.g. different analytical determinations/platforms) allows capturing the latent information that would not be extracted by the individual analysis of each block of data. Several approaches are proposed in the literature and are generally referred to as data fusion approaches. In this work a mid level data fusion is proposed for the characterization of three varieties ( Salamino di Santa Croce , Grasparossa di Castelvetro , Sorbara ) of Lambrusco wine, a typical PDO wine of the district of Modena (Italy). Wine samples of the three different varieties were analyzed by means of 1 H-NMR spectroscopy, Emission-Excitation Fluorescence Spectroscopy and HPLC-DAD of the phenolic compounds. Since the analytical outputs are characterized by different dimensionalities (matrix and tensor), several multivariate analyses were applied (PCA, PARAFAC, MCR-ALS) in order to extract and merge, in a hierarchical way, the information present in each data set. The results showed that this approach was able to well characterize Lambrusco samples giving also the possibility to understand the correlation between the sources of information arising from the three analytical techniques.

This work presents a sensor system develop for the robot ALDURO (Antropomorphically Legged and Wheeled Duisburg Robot), in order to allow it to detect and avoid obstacles when moving in unstructured terrains. The robot is a large-scale hydraulically driven 4-legged walking-machine, developed at the Duisburg-Essen University, with 16 degrees of freedom at each leg and will be steered by an operator sitting in a cab on the robot body. The Cartesian operator instructions are processed by a control computer, which converts them into appropriate autonomous leg movements, what makes necessary that the robot automatically recognizes the obstacles (rock, trunks, holes, etc.) on its way, locates and avoids them. A system based on ultra-sound sensors was developed to carry this task on, but there are intrinsic problems with such sensors, concerning to their poor angular precision. To overcome that, a fuzzy model of the used ultra-sound sensor, based on the characteristics of the real one, was developed to include the uncertainties about the measures. A posterior fuzzy inference builds from the measured data a map of the robot’s surroundings, to be used as input to the navigation system. This whole sensor system was implemented at a test stand, where a real size leg of the robot is fully functional. The sensors are assembled in an I2C net, which uses a micro-controller as interface to the main controller (a personal computer). That enables to relieve the main controller of some data processing, which is carried by the microcontroller on. The sensor system was tested together with the fuzzy data inference, and different arrangements to the sensors and settings of the inference system were tried, in order to achieve a satisfactory result.

At present the authors are developing the system FASA, an airborne combination of a Fourier Transform Spectrometer and an imaging system. The aim is to provide a system that is usable to investigate and monitor emissions from natural disasters such as wild fires and from volcanoes. Besides temperatures and (burned) areas FASA will also provide concentration profiles of the gaseous combustion products. These data are needed to improve the knowledge of the effects of such emissions on the global ecosystem. The paper presents a description of the instrumentation, the data evaluation procedure and shows first results of retrieval calculations based on simulated spectra.

Die Arbeit hat die Entwicklung von Klassifikationsstrategien zur automatischen Erfassung der Landnutzung aus hochauflösenden Fernerkundungsdaten zum Thema. Im einzelnen wurden IKONOS-Satellitenbilder, multitemporale Luftbilder und Data-Fusion Ansätze untersucht, um großmaßstäbliche Landnutzungsinformationen zu erhalten. Kernstück der Arbeit war die Entwicklung eines strukturierten Auswertekonzepts für ein objektorientiertes Klassifikationsprojekt. Das entwickelte Auswertekonzept wurde erfolgreich an drei Fallstudien überprüft. Dabei wurden verschiedene Aspekte des objektorientierten Ansatzes in einem Untersuchungsgebiet in Süddeutschland. This thesis is about the development of objectoriented classification strategies for automatic landuse assessment from high resolution remote sensing data. IKONOS satellite images, multitemporal aerial images and data-fusion approaches were analysed, in order to collect large-scale landuse data. Main item was the development of a structured procedure for an objectoriented classification project. This procedure was tested successfully in three case studies. In doing so, different aspects of the objectoriented approach were analysed in a study site in Southern Germany.

International audience; A prerequisite to the design of future Advanced Driver Assistance Systems for cars is a sensing system providing all the information required for high-level driving assistance tasks. Carsense is a European project whose purpose is to develop such a new sensing system. It will combine different sensors (laser, radar and video) and will rely on the fusion of the information coming from these sensors in order to achieve better accuracy, robustness and an increase of the information content. This paper demonstrates the interest of using probabilistic reasoning techniques to address this challenging multi-sensor data fusion problem. The approach used is called Bayesian Programming. It is a general approach based on an implementation of the Bayesian theory. It was introduced first to design robot control programs but its scope of application is much broader and it can be used whenever one has to deal with problems involving uncertain or incomplete knowledge.