Planetary rovers are robots that need to perform autonomous navigation, because of the long delay communication and no human assistance. Furthermore, they need to perform the optimal estimation of its position in order to have a good performance on its navigation system. The need for good performance filters for estimating the actual position of mobile robots of this kind is needed, due to the fact that sensors are noisy and that information is of vital importance for a planetary rover’s mission. Besides, good accurate sensors for the matter, are not easy to find for space application. Extended Kalman Filter (EKF) and Unscented Kalman Filter (UKF) were implemented to analyze a data set of a 4-wheel robot, and later used for comparison on accuracy in the estimation of its pose. The analysis will give the possibility to know the right combination of sensors, recognize some issues during the trajectory. Furthermore, this study has been made with aims to give the reader knowledge of state of the art in planetary rovers, their constraints and consideration while developing them. The robot used for the research has been developed for an international competition of field robot automation. The main goal is to navigate autonomously through flowerpots performing different tasks as flowerpot collection, distance traveled and robustness on localization and navigation algorithms. / <p>Validerat; 20120822 (anonymous)</p>

Les méthodes à l’état de l’art de la reconnaissance de l’écriture sont fondées sur des réseaux de neurones récurrents (RNN) à cellules LSTM ayant des performances remarquables. Dans cette thèse, nous proposons deux nouveaux principes la vérification lexicale et la génération de cohorte afin d’attaquer les problèmes de la reconnaissance de l’écriture : i) le problème des grands lexiques et des décodages dirigés par le lexique ii) la problématique de combinaison de modèles optiques pour une meilleure reconnaissance iii) la nécessité de constituer de très grands ensembles de données étiquetées dans un contexte d’apprentissage profond. La vérification lexicale est une alternative aux décodages dirigés par le lexique peu étudiée à cause des faibles performances des modèles optiques historiques (HMM). Nous montrons dans cette thèse qu’elle constitue une alternative intéressante aux approches dirigées par le lexique lorsqu’elles s’appuient sur des modèles optiques très performants comme les RNN LSTM. La génération de cohorte permet de générer facilement et rapidement un grand nombre de réseaux récurrents complémentaires en un seul apprentissage. De ces deux techniques nous construisons et proposons un nouveau schéma de cascade pour la reconnaissance de mots isolés, une nouvelle combinaison au niveau ligne LV-ROVER et une nouvelle stratégie d’auto-apprentissage de RNN LSTM pour la reconnaissance de mots isolés. La cascade proposée permet de combiner avec la vérification lexicale des milliers de réseaux et atteint des résultats à l’état de l’art pour les bases Rimes et IAM. LV-ROVER a une complexité réduite par rapport à l’algorithme original ROVER et permet de combiner des centaines de réseaux sans modèle de langage tout en dépassant l’état de l’art pour la reconnaissance de lignes sur le jeu de donnéesRimes. Notre stratégie d’auto-apprentissage permet d’apprendre à partir d’un seul réseau BLSTM et sans paramètres grâce à la cohorte et la vérification lexicale, elle montre d’excellents résultats sur les bases Rimes et IAM. / State-of-the-art methods for handwriting recognition are based on LSTM recurrent neural networks (RNN) which achieve high performance recognition. In this thesis, we propose the lexicon verification and the cohort generation as two new building blocs to tackle the problem of handwriting recognition which are : i) the large vocabulary problem and the use of lexicon driven methods ii) the combination of multiple optical models iii) the need for large labeled dataset for training RNN. The lexicon verification is an alternative to the lexicon driven decoding process and can deal with lexicons of 3 millions words. The cohort generation is a method to get easily and quickly a large number of complementary recurrent neural networks extracted from a single training. From these two new techniques we build and propose a new cascade scheme for isolated word recognition, a new line level combination LV-ROVER and a new self-training strategy to train LSTM RNN for isolated handwritten words recognition. The proposed cascade combines thousands of LSTM RNN with lexicon verification and achieves state-of-the art word recognition performance on the Rimes and IAM datasets. The Lexicon Verified ROVER : LV-ROVER, has a reduce complexity compare to the original ROVER algorithm and combine hundreds of recognizers without language models while achieving state of the art for handwritten line text on the RIMES dataset. Our self-training strategy use both labeled and unlabeled data with the unlabeled data being self-labeled by its own lexicon verified predictions. The strategy enables self-training with a single BLSTM and show excellent results on the Rimes and Iam datasets.

Preponderant visual interface use for conveying information from machine to human admits failures due to overwhelming the visual channel. This thesis investigates the suitability of auditory feedback and certain related design choices in settings involving background speech. Communicating a tele-operated vehicle’s tilt angle was the focal application. A simulator experiment with pitch feedback on one system variable, tilt angle, and its safety threshold was conducted. Manipulated in a within-subject design were: (1) presence vs. absence of speech, (2) discrete tilt alarm vs. discrete alarm and tilt sonification (continuous feedback), (3) tilt sonification vs. tilt and threshold sonification, and (4) linear vs. quadratic transfer function of variable to pitch. Designs with both variable and reference sonification were found to significantly reduce the time drivers spent exceeding the safety limit compared to the designs with no sonification, though this effect was not detected within the set of conditions with background speech audio.

Preponderant visual interface use for conveying information from machine to human admits failures due to overwhelming the visual channel. This thesis investigates the suitability of auditory feedback and certain related design choices in settings involving background speech. Communicating a tele-operated vehicle’s tilt angle was the focal application. A simulator experiment with pitch feedback on one system variable, tilt angle, and its safety threshold was conducted. Manipulated in a within-subject design were: (1) presence vs. absence of speech, (2) discrete tilt alarm vs. discrete alarm and tilt sonification (continuous feedback), (3) tilt sonification vs. tilt and threshold sonification, and (4) linear vs. quadratic transfer function of variable to pitch. Designs with both variable and reference sonification were found to significantly reduce the time drivers spent exceeding the safety limit compared to the designs with no sonification, though this effect was not detected within the set of conditions with background speech audio.

abstract: The search for life on Mars is a major NASA priority. A Mars Sample Return (MSR) mission, Mars 2020, will be NASA's next step towards this goal, carrying an instrument suite that can identify samples containing potential biosignatures. Those samples will be later returned to Earth for detailed analysis. This dissertation is intended to inform strategies for fossil biosignature detection in Mars analog samples targeted for their high biosignature preservation potential (BPP) using in situ rover-based instruments. In chapter 2, I assessed the diagenesis and BPP of one relevant analog habitable Martian environment: a playa evaporite sequence within the Verde Formation, Arizona. Coupling outcrop-scale observations with laboratory analyses, results revealed four diagenetic pathways, each with distinct impacts on BPP. When MSR occurs, the sample mass returned will be restricted, highlighting the importance of developing instruments that can select the most promising samples for MSR. Raman spectroscopy is one favored technique for this purpose. Three Raman instruments will be sent onboard two upcoming Mars rover missions for the first time. In chapters 3-4, I investigated the challenges of Raman to identify samples for MSR. I examined two Raman systems, each optimized in a different way to mitigate a major problem commonly suffered by Raman instruments: background fluorescence. In Chapter 3, I focused on visible laser excitation wavelength (532 nm) gated (or time-resolved Raman, TRR) spectroscopy. Results showed occasional improvement over conventional Raman for mitigating fluorescence in samples. It was hypothesized that results were wavelength-dependent and that greater fluorescence reduction was possible with UV laser excitation. In Chapter 4, I tested this hypothesis with a time-resolved UV (266 nm) gated Raman and UV fluorescence spectroscopy capability. I acquired Raman and fluorescence data sets on samples and showed that the UV system enabled identifications of minerals and biosignatures in samples with high confidence. The results obtained in this dissertation may inform approaches for MSR by: (1) refining models for biosignature preservation in habitable Mars environments; (2) improving sample selection and caching strategies, which may increase the success of Earth-based biogenicity studies; and (3) informing the development of Raman instruments for upcoming rover-based missions. / Dissertation/Thesis / Doctoral Dissertation Geological Sciences 2016

Martian clouds and dust play an important part of the radiative transfer and energy balance budget. To assist in fully understanding the impact of clouds and dust, the complete diurnal cycle needs to be characterized. One of the best methods to track diurnal variations on Mars is by measuring optical depth. The spatial and temporal trends of optical depth give insight into the dust and water cycles of the Martian atmosphere. Until now, spacecraft could only obtain optical depth during the day. In this thesis, nighttime images from the Mars Exploration Rover Spirit are used to calculate nighttime optical depth using photometric methods to capture star flux. Bright stars in well-known constellations are used in this analysis. The observed flux was compared to the expected flux to give nighttime optical depth values. The observed nighttime optical depth was consistently similar to the daytime optical depth values on both an individual image and sol-averaged basis. Recommendations are made going forward to use the Mars Science Laboratory Curiosity for conducting an optimal nighttime optical depth campaign to fully characterize the diurnal dust and water cycles of Mars. The Curiosity rover is well suited for nighttime imaging and can potentially provide valuable insight into the nighttime dust and cloud trends.

<p>The present-day sedimentary rock record on Mars provides insights into the early surface and subsurface geologic processes. Understanding the sediment characteristics in different environments can help to constrain the climate regimes, potential for habitability, and provide a record of ancient surface processes. The research presented in this dissertation uses complementary remote sensing techniques and datasets from rovers at the surface, satellites in orbit, and at terrestrial analogs that are relevant to current Mars exploration to better characterize alteration through water-rock alteration at multiple scales.</p><p>The martian field site for this work is Mt. Sharp, a 5-kilometer-high mountain in Gale crater that is predominantly composed of fluviolacustrine strata overlain by aeolian strata. At the rover-scale, the effects of large clay-mineral rich deposits were characterized using landscape- and hand lens-scale visible images from the Mastcam and MAHLI instruments, and multispectral visible/near-infrared images from Mastcam (445-1013 nm). Detailed analysis of the observed textures and spectral properties showed that the clay-rich deposits preserve the early surface environment, based on their lack of diagenetic features. While the regions immediately surrounding the clay-rich deposit experienced prolonged exposure to water, leading to enhanced alteration zones, and destroying characteristics from the early environment but providing insight into later water-rock processes.</p><p>At the orbital-scale, three visually distinct, dark-toned, and erosion-resistant layers were mapped and characterized using visible to short wave infrared hyperspectral (700-2650 nm) and image data. Two of these units have been identified as either aeolian or lacustrine through in situ rover investigations and the third unit will not be explored in situ so its origin can only be constrained through orbital analyses. We conducted a comparison of the morphological and spectral properties of the two known units to constrain whether their respective environments can be differentiated from orbit and apply this knowledge to the unknown third unit. The composition of all three units is similar, dominated by mafic minerals, suggesting a similar sediment source. The morphology is distinct between the lacustrine and aeolian units, with the unknown unit having similar morphology as the lacustrine unit, suggesting similar environments. We propose that the lacustrine unit in this study likely represent short-timescale transitions between wet and dry environments, where mafic sands are exposed to water prior to burial and lithification. While in the aeolian unit, most water-rock interactions occur upon lithification and later diagenesis. This has climatic implications in terms of the presence of surface water as these units were deposited as part of the original Mt. Sharp strata (i.e., the lacustrine unit) while some mantling existing topography (i.e., the aeolian and unknown units), representing similar processes but at a much later time.</p><p>The terrestrial analog field site for this dissertation was conducted in Iceland which represents a cold and wet/icy climate. We characterized sediments produced through glaciovolcanism and how they are sorted and altered through transport from source to sink along to characterize unique identifiers of glaciovolcanism that can be determined with Mars-relevant techniques. Decorrelation stretched visible images and lab visible/near-infrared reflectance and thermal-infrared emission data sets (400-2500 nm and 1200-400 cm-1, respectively) show that it is possible to differentiate sediments from glaciovolcanic and subaerial volcanic systems. In some glaciovolcanic systems, a high glass abundance (50-90 %) is observed in sediment grains due to the erosion of hyaloclastite and hyalotuff, deposits that form in water- and ice-magma interactions. These glass grains did not readily breakdown physically or chemically during transport, suggesting that they could still be observed on the martian surface today and be used to identify possible glaciovolcanic deposits.</p><p>The research described in this thesis improves the understanding of different geologic environments using remote sensing techniques and their climatic implications. This will help to better constrain early environments on Mars and identify areas where water may have been present through the rock record, as observed from the surface and from orbit.</p>