This Research Topic seeks to address vehicle localization and scene understanding algorithms in urban settings. Contrary to generic localization issues, here the focus is to exploit common urban features and combine them in innovative ways to obtain a reliable awareness in urban scenarios (urban is a key aspect). Autonomous vehicles hinge on advanced algorithms for object detection and tracking, self-localization, and vehicle control. Although each of these components is essential to safely plan vehicle actions, all concurrently support the main challenge, i.e., understanding of the surrounding environment.
In urban settings, the perception and the interpretation of objects and entities within a scene is crucial, since a proper scene interpretation could prevent the vehicle from running into potentially treacherous situations, as well as ensuring the safety of Vulnerable Road Users (VRU). On the one hand, navigating automated vehicle navigation in urban scenarios often leads to dealing with Global Navigation and Satellite System (GNSS)-denied or GNSS-limited areas. For this reason, one may benefit from the detection of specific urban features such as intersections, road topologies, multi-lane and traffic-flow detection in case of lack of road markings, buildings, and many others. In addition, the exploitation of cartographic maps as well their augmentation represents a key factor that contributes towards not only a much more reliable localization but, with a broader vision, towards an enhanced scene understanding system. This global awareness also enables a feasible prediction of road user actions, making it possible to preemptively identify dangerous conditions that might lead to non-fatal injuries or deaths.
This Research Topic includes, but is not limited, to the following interests:
• Detection and exploitation of specific urban features such as buildings, curbs, lanes, intersections, and other features that cannot be here anticipated
• Cartographic map exploitation
• Innovative use of different localization systems/cues for localization
• Exploitation of traffic detection and tracking in specific areas for localization
• Datasets containing urban features ground truth for benchmarking purposes
• Identification, tracking, and trajectory prediction of VRUs, Vehicles and other road users
This Research Topic seeks to address vehicle localization and scene understanding algorithms in urban settings. Contrary to generic localization issues, here the focus is to exploit common urban features and combine them in innovative ways to obtain a reliable awareness in urban scenarios (urban is a key aspect). Autonomous vehicles hinge on advanced algorithms for object detection and tracking, self-localization, and vehicle control. Although each of these components is essential to safely plan vehicle actions, all concurrently support the main challenge, i.e., understanding of the surrounding environment.
In urban settings, the perception and the interpretation of objects and entities within a scene is crucial, since a proper scene interpretation could prevent the vehicle from running into potentially treacherous situations, as well as ensuring the safety of Vulnerable Road Users (VRU). On the one hand, navigating automated vehicle navigation in urban scenarios often leads to dealing with Global Navigation and Satellite System (GNSS)-denied or GNSS-limited areas. For this reason, one may benefit from the detection of specific urban features such as intersections, road topologies, multi-lane and traffic-flow detection in case of lack of road markings, buildings, and many others. In addition, the exploitation of cartographic maps as well their augmentation represents a key factor that contributes towards not only a much more reliable localization but, with a broader vision, towards an enhanced scene understanding system. This global awareness also enables a feasible prediction of road user actions, making it possible to preemptively identify dangerous conditions that might lead to non-fatal injuries or deaths.
This Research Topic includes, but is not limited, to the following interests:
• Detection and exploitation of specific urban features such as buildings, curbs, lanes, intersections, and other features that cannot be here anticipated
• Cartographic map exploitation
• Innovative use of different localization systems/cues for localization
• Exploitation of traffic detection and tracking in specific areas for localization
• Datasets containing urban features ground truth for benchmarking purposes
• Identification, tracking, and trajectory prediction of VRUs, Vehicles and other road users
