In the context of autonomous vehicles on highways, one of the first and most important tasks is to localize the vehicle on the road. For this purpose, the vehicle needs to be able to take into account the information from several sensors and fuse them with data coming from road maps. The localization problem on highways can be distilled into three main components. The first one consists of inferring on which road the vehicle is currently traveling. Indeed, Global Navigation Satellite Systems are not precise enough to deduce this information by themselves, and thus a filtering step is needed. The second component consists of estimating the vehicle’s position in its lane. Finally, the third and last one aims at assessing on which lane the vehicle is currently driving. These two last components are mandatory for safe driving as actions such as overtaking a vehicle require precise information about the current localization of the vehicle. In this survey, we introduce a taxonomy of the localization methods for autonomous vehicles in highway scenarios. We present each main component of the localization process, and discuss the advantages and drawbacks of the associated state-of-the-art methods.

Locating the vehicle in its road is a critical part of any autonomous vehicle system and has been subject to different research topics. In most of the works presented in the literature, ego-localization is split into three parts: Road level-localization: the road on which the vehicle travels, Lane level localization: the lane on which the vehicle travels, and Ego lane level localization: the lateral position of the vehicle in the ego-lane. For each part, several researches have been conducted. However, the relationship between the different parts has not been taken into consideration. Through this work, an end-to-end ego-localization framework is introduced with mainly two main novelties. The first one is that we propose a complete solution that tackles every parts of the ego-localization. The second one lies in the information-driven approach used. Indeed, we use prior about the road structure from a digital map in order to reduce the space complexity for the recognition process. Besides, several fusion framework techniques based on Bayesian Network and Hidden Markov Model are elaborated leading to an ego-localization method that is, to a large extent, robust to erroneous sensor data. The robustness of the proposed method is proven on different datasets in varying scenarios.

In the context of autonomous robots, one of the most important tasks is to prevent potential damage to the robot during navigation. For this purpose, it is often assumed that one must deal with known probabilistic obstacles, then compute the probability of collision with each obstacle. However, in complex scenarios or unstructured environments, it might be difficult to detect such obstacles. In these cases, a metric map is used, where each position stores the information of occupancy. The most common type of metric map is the Bayesian occupancy map. However, this type of map is not well suited for computing risk assessments for continuous paths due to its discrete nature. Hence, we introduce a novel type of map called the Lambda Field, which is specially designed for risk assessment. We first propose a way to compute such a map and the expectation of a generic risk over a path. Then, we demonstrate the benefits of our generic formulation with a use case defining the risk as the expected collision force over a path. Using this risk definition and the Lambda Field, we show that our framework is capable of doing classical path planning while having a physical-based metric. Furthermore, the Lambda Field gives a natural way to deal with unstructured environments, such as tall grass. Where standard environment representations would always generate trajectories going around such obstacles, our framework allows the robot to go through the grass while being aware of the risk taken.

In this letter, we propose a way to model the resilience of the Iterative Closest Point (ICP) algorithm in the presence of corrupted measurements. In the context of autonomous vehicles, certifying the safety of the localization process poses a significant challenge. As robots evolve in a complex world, various types of noise can impact the measurements. Conventionally, this noise has been assumed to be distributed according to a zero-mean Gaussian distribution. However, this assumption does not hold in numerous scenarios, including adverse weather conditions, occlusions caused by dynamic obstacles, or long-term changes in the map. In these cases, the measurements are instead affected by large and deterministic faults. This letter introduces a closed-form formula approximating the pose error resulting from an ICP algorithm when subjected to the most detrimental adverse measurements. Using this formula, we develop a metric to certify and pinpoint specific regions within the environment where the robot is more vulnerable to localization failures in the presence of faults in the measurements.

Lambda-Field : une nouvelle méthode pour l'évaluation de risque dans les grilles d'occupation: Dans le contexte des robots autonomes, l'un des sujets les plus controversés est la notion de risque. En effet, aucun robot n'échappe à une telle question, à savoir si ce robot ne causera aucun dommage à lui-même ou aux êtres vivants dans l'environnement qui l'entoure. Les bras robotiques ont un espace de travail fini et relativement petit, où le risque est abordé de telle sorte que le robot s'arrête complètement dès qu'un humain pénètre dans son espace de travail. En robotique mobile, cette notion est plus complexe et reste un problème ouvert. Souvent, la représentation préférée de l'environnement est la représentation sémantique, où chaque obstacle est stocké comme une entité unique. Cependant, dans des scénarios complexes ou des environnements non structurés, la détection de tels obstacles est une tâche fastidieuse et en manquer un peut conduire à des événements désastreux. Dans ces cas, une carte métrique est utilisée, où chaque position stocke l'information d'occupation. Le type le plus courant de carte métrique est la carte d'occupation bayésienne. Cependant, ce type de carte n'est pas bien adapté pour calculer l'évaluation du risque pour les chemins continus en raison de sa nature discrète. Par conséquent, nous introduisons dans cette thèse un nouveau type de carte appelé Lambda-Field, spécialement conçu pour l'évaluation du risque. Les Lambda-Fields sont une contrepartie de la grille d'occupation bayésienne classique. Au lieu de stocker la probabilité d'occupation à chaque position, le Lambda-Field stocke l'intensité de la probabilité qu'une collision se produise à cette position : plus l'intensité est élevée, plus la probabilité de collision est élevée. Grâce à cette nouvelle formulation, les Lambda-Fields sont capables d'évaluer un risque générique sur un chemin. Contrairement à la grille d'occupation bayésienne, l'utilisation de l'intensité au lieu de la probabilité directe de collision permet à la méthode d'évaluation de risque de produire des métriques basées sur la physique qui conservent leurs unités physiques. Tout au long de cette thèse, nous présentons comment construire et utiliser le Lambda-Field dans des environnements statiques et dynamiques. Nous montrons que le Lambda-Field possède également des propriétés cartographiques intéressantes qui induisent des cartes d'environnements non structurés plus précises. En utilisant cette définition du risque et le Lambda-Field, nous montrons que notre méthode est capable de faire de la planification de chemin classique mais aussi de traverser des environnements non structurés où une grille d'occupation bayésienne ne trouverait pas de chemin.