Autonomous navigation in urban environments faces significant challenges, particularly due to dynamic obstacles and occluded areas, such as those found at intersections. This paper presents a novel occlusion-aware risk assessment and navigation framework designed to address these issues. Unlike traditional methods, which often overlook risks in unobserved regions, the proposed approach leverages prior map data (e.g., OpenStreetMap) and real-time LiDAR sensor inputs to quantify risks from both visible and potential hidden obstacles. By combining static and dynamic information, the system enables robust trajectory planning using a tentacle-based approach, optimizing safety and task efficiency. Risk distributions are represented using particle-based modeling, offering simplicity and parallelizability. Evaluated in the Carla simulator across various scenarios, the framework demonstrates enhanced collision avoidance and optimal navigation, effectively balancing safety and speed in complex urban settings.

In a context of autonomous driving, perception of the surrounding is a crucial task. It characterizes the vehicle’s ability to simultaneously model its surroundings accurately and maintain its position in the environment. In this article, a new framework of mobile robot perception and risk assessment is proposed. Our approach aims to leverage the simultaneous combination of the standard occupancy grid map with a new map that we have called "knowledge map". This proposal was motivated by the fact that risk arises not only from obstacles but also from the lack of knowledge. Using this framework, we are able to assess the risk, mainly of collision, over a given path $\mathscr{P}$ and therefore compute an optimal navigation control of the robot. Thanks to the proposed Bayesian framework the paper also shows how we can combine both local measurements and existing map (eg. OpenStreetMap) and also take account of the robot’s localization errors.

Automotive players have recently shown an increasing interest in high-precision mapping, with the aim of enhancing vehicles safety and autonomy. Nevertheless, the acquisition, processing, and updates of accurate maps remains an economic challenge. Collaborative mapping through vehicles crowdsourcing represents a promising solution to tackle this problem. However, the potential scalability and accuracy provided by such an approach have yet to be studied and assessed. In this paper, we study the use of graph optimization in the scope of collaborative mapping. We build a map of geo-localized landmarks by crowdsourcing observations from multiple vehicles, and applying several successive map updates. We present different strategies to adapt graph optimization to the crowdsourced approach, and compare their performances in terms of map quality and scalability on simulation data. We show the critical requirement, in a long-term context, to ensure consistency of the map updates, and we propose a scalable solution which is able to build an accurate map of geolocalized landmarks.

In this paper, we address the problem of localizing mobile robots based on range-only measurements from low cost Ultra-Wide-Band (UWB) sensors. The proposed solution considers every static or mobile object as beacons with contextual information. A beacon-to-beacon measurement is performed using UWB sensors and the position estimation is computed by the target beacon. This strategy allows to hide the cooperative localization problem behind these measurements. The fusion algorithm is based on a Split Intersection Covariance Filter which allows to correctly handle the correlation between the poses estimations of the beacons. We present the consistency of this solution using a simulation with 3 robots and 4 static beacons and a real experimentation with 1 robot and 3 static beacons.

One of the major current developments in outdoor robotic is providing vehicles with automatic guidance capabilities. Such systems need a localization module to work. However, indoor localization methods are not directly usable in outdoor due to noise and the dynamic aspect of these environments. In this paper, we propose an original active localization system relying on sensors fusion able to supply an accurate position with a high confidence level. The main contributions of this work are: 1) the introduction of the "perceptive triplet" notion that associates landmarks, sensors and detectors to supervise the detections. 2) The use of a supervisor that determines at each time which landmark, with which sensor and detector, should be used to detect this landmark in order to improve the localization. The supervisor constitutes the intelligent part of this localization system. It decides when it's necessary to detect a landmark. 3) The integration of a confidence level over the vehicle's pose estimation that permits to take wrong matching hypothesis into account. Our system was tested in an outdoor environment, where it succeeded in accurately localizing the vehicle