Abstract
Unmanned Aerial Vehicles (UAVs) are a key ingredient in the industry and in warehouse logistics digital transformation process, providing the ability to perform automatic cyclic counting and real-time inventory, localize hard-to-find items and reach narrow storage areas. The use of UAVs poses new challenges, such as indoor autonomous localization and navigation, collision avoidance and automated UAV fleet management. This paper addresses the development of a vision-based Graph-SLAM approach for UAV indoor localization without predefined warehouse markers positions. A framework is proposed and developed to support different commercial UAV platforms, allowing the estimation in real-time of the UAV position and attitude. Indoor experimental tests were carried out in order to evaluate the performance of the developed method, comparing the results obtained with an approach based on the pre-mapped markers position indoor localization method.

Abstract
The use of autonomous vehicles in maritime operations is a technological challenge. In the particular case of autonomous aerial vehicles (UAVs), their application ranges from inspection and surveillance of offshore power plants, and marine life observation, to search and rescue missions. Manually landing UAVs onboard water vessels can be very challenging due to limited space onboard and wave agitation. This paper proposes an autonomous solution for the task of landing commercial multicopter UAVs with onboard cameras on water vessels, based on the detection of a custom landing platform with computer vision techniques. The autonomous landing behavior was tested in real conditions, using a research vessel at sea, where the UAV was able to detect, locate, and safely land on top of the developed landing platform.

Abstract
We present MONET, a new multimodal dataset captured using a thermal camera mounted on a drone that flew over rural areas, and recorded human and vehicle activities. We captured MONET to study the problem of object localisation and behaviour understanding of targets undergoing large-scale variations and being recorded from different and moving viewpoints. Target activities occur in two different land sites, each with unique scene structures and cluttered backgrounds. MONET consists of approximately 53K images featuring 162K manually annotated bounding boxes. Each image is timestamp-aligned with drone metadata that includes information about attitudes, speed, altitude, and GPS coordinates. MONET is different from previous thermal drone datasets because it features multimodal data, including rural scenes captured with thermal cameras containing both person and vehicle targets, along with trajectory information and metadata. We assessed the difficulty of the dataset in terms of transfer learning between the two sites and evaluated nine object detection algorithms to identify the open challenges associated with this type of data. Project page: https://github.com/fabiopoiesi/monet-dataset.