Considering a class of linear multi-agent systems, we study the problem of consensus-seeking in the presence of an exogenous signal, possibly representing a disturbance. We formulate the robust consensus-seeking problem as a nonzero-sum differential game.

The purpose of this article is to take stock of the progress made in remote sensing thanks to the recent development of deep learning techniques. This assessment is made by means of a systematic presentation of the various activities carried out at ONERA in remote sensing imagery using deep learning methods. It covers a large part of the observation problems: filtering, object detection, land-use classification, change detection, and biomass estimation.

This paper presents two use cases where deep learning is able to help scientists by removing the burden of manual review of large volumes of physical data. Such examples highlight why deep learning could become a transverse tool across many scientific fields.

Computer vision techniques have made considerable progress in recent years. This advance now makes possible the practical use of computer vision in civil drones or aircraft, replacing human pilots. The question that naturally arises is then to provide a way to certify those types of systems at a given level of safety.

In this article, we present planning and scheduling techniques that we developed for optimizing the operations of space telescopes. The latter are satellites whose mission is to observe celestial objects such as planets, exoplanets, stars, or galaxies. After a survey of some existing mission planning tools, we present three case studies that we tackled using a constraint-based optimization and operations research approach, with for each case study the lessons that we learned.

This paper studies two-agent path planning algorithms in graphs, where the two agents are assigned independent initial and goal states but are incentivized to share some parts of their travel glued together by scaling down the duet cost function when they move in formation. Applications range from ride sharing to formation flights.

Airborne platforms such as Remote Piloted Aircraft Systems (RPAS) operate in highly critical contexts. The next generation of RPAS will be endowed with multifunction sensors (i.e., each sensor offers a large panel of functions to the platform's manager during the mission). As a platform, RPAS carry out a wide collection of complex tasks, thanks to the interleaving of the various services of sensors. The sensors are in charge of collecting data from the environment.

Until recently, the processing of a rapidly changing dataflow used to be very costly in terms of computation duration. Thus, the extraction of semantic information, requiring complex correlations of events in a temporal pattern, was not possible in real time. Computer performance has sufficiently improved to now allow such processing to take place, with, obviously, a very broad range of interesting applications.

This paper addresses the problem of combining human observations and sensor data for entity tracking and identification in dynamic environments. The complexity of the track-and-detect task for realistic applications requires dynamic fusion of sensor data and observations, and a semantic mediation approach is adopted. Moving targets are detected and classified based on sensor data.

Artificial Intelligence (AI) is a field of computer science which, through its ability to efficiently process data, to build decisions and to automate some processes, is likely to play a major societal and economic role. The fields of Aeronautics, Space and Defense (ASD) are no exception to this trend.