This paper addresses Visual Odometry (VO) estimation in challenging underwater scenarios. Robot visual-based navigation faces several additional difficulties in the underwater context, which severely hinder both its robustness and the possibility for persistent autonomy in underwater mobile robots using visual perception capabilities. In this work, some of the most renown VO and Visual Simultaneous Localization and Mapping (v-SLAM) frameworks are tested on underwater complex environments, assessing the extent to which they are able to perform accurately and reliably on robotic operational mission scenarios. The fundamental issue of precision, reliability and robustness to multiple different operational scenarios, coupled with the rise in predominance of Deep Learning architectures in several Computer Vision application domains, has prompted a great a volume of recent research concerning Deep Learning architectures tailored for visual odometry estimation. In this work, the performance and accuracy of Deep Learning methods on the underwater context is also benchmarked and compared to classical methods. Additionally, an extension of current work is proposed, in the form of a visual-inertial sensor fusion network aimed at correcting visual odometry estimate drift. Anchored on a inertial supervision learning scheme, our network managed to improve upon trajectory estimates, producing both metrically better estimates as well as more visually consistent trajectory shape mimicking.

À minha madrinha Celeste, por tudo o que fizeste por mim. Ao Hugo, por seres um exemplo para mim, me mostrares o caminho e incentivares a cada passo do caminho. Aos meus avós, tios, primos e todo o resto da minha família, cada um à sua maneira importante nesta caminhada. Uma palavra especial para quem já não cá está, mas nunca partiu do meu coração. À rapaziada do DMESM, foi um prazer partilhar esta caminhada convosco. 5 anos de pura partilha nesta montanha russa de emoções, não podia pedir mehor companhia nesta etapa. Aos meus amigos Salvador, Kiko, Hugo, Filipe, Diana e Isabel, já sabem que são a família que eu escolhi, e se largos dias tem 100 anos, largas noites tem histórias que ficam para a vida. Obrigado especialmente por terem sido o garante da minha sanidade ao longo deste percurso. E por último, quero agradecer à pessoa única e especial que é a minha namorada Ana. Entrámos já juntos nesta caminhada, e sem ti nada teria sido igual. Obrigado por seres a minha rocha e por sempre me motivares para ser a melhor versão de mim.

No abstract

Achieving persistent and reliable autonomy for mobile robots in challenging field mission scenarios is a long-time quest for the Robotics research community. Deep learning-based LIDAR odometry is attracting increasing research interest as a technological solution for the robot navigation problem and showing great potential for the task.In this work, an examination of the benefits of leveraging learning-based encoding representations of real-world data is provided. In addition, a broad perspective of emergent Deep Learning robust techniques to track motion and estimate scene structure for real-world applications is the focus of a deeper analysis and comprehensive comparison.Furthermore, existing Deep Learning approaches and techniques for point cloud odometry tasks are explored, and the main technological solutions are compared and discussed.Open challenges are also laid out for the reader, hopefully offering guidance to future researchers in their quest to apply deep learning to complex 3D non-matrix data to tackle localization and robot navigation problems.