Flying drones is an increasingly popular but challenging activity. It is even more challenging for users with special needs. In this paper, we present our collaboration with a drone piloting workshop for people with motor and cognitive disabilities to design CandyFly, an application adaptable to the users' abilities. Several piloting workshops allowed pilots to have fun flying drones with CandyFly and allowed us to explore interactions to help pilots and their caregivers. We conclude with a discussion of the adaptations made along three axes: hardware, software and automation; which can be used for the design of future assistance and leisure systems.

Robustness of deep neural networks is a critical issue in practical applications. In the general case of feed-forward neural networks (including convolutional deep neural network architectures), under random noise attacks, we propose to study the probability that the output of the network deviates from its nominal value by a given threshold. We derive a simple concentration inequality for the propagation of the input uncertainty through the network using the Cramer-Chernoff method and estimates of the local variation of the neural network mapping computed at the training points. We further discuss and exploit the resulting condition on the network to regularize the loss...

The aircraft scheduling problem consists in sequencing aircraft on airport runways and in scheduling their times of operations taking into consideration several operational constraints. It is known to be an NP-hard problem, an ongoing challenge for both researchers and air traffic controllers. The aim of this paper is to present a focused review on the most relevant techniques in the recent literature (since 2010) on the aircraft runway scheduling problem, including exact approaches such as mixed-integer programming and dynamic programming, metaheuristics, and novel approaches based on reinforcement learning. Since the benchmark instances used in the literature are easily...

[...]

Le déclin des capacités cognitives des pilotes lié à l'âge peut mener à l'erreur et altérer ainsi la sécurité des vols. Cette thèse se propose d'étudier différentes solutions d'assistance afin de réduire l'impact du vieillissement sur les performances des pilotes. Des interviews ont fait émerger deux grandes problématiques : la baisse des capacités attentionnelles pouvant être compensée par l'utilisation de différentes modalités d'alarmes, et le déclin cognitif, dont les conséquences peuvent être atténuées par le développement d'interfaces soutenant le pilote dans ses tâches de pilotage. Nous prévoyons de tester ces solutions en vol simulé et/ou en vol réel.

Over the past few years, the need for positioning, and thus the number of positioning services in general, has been in constant growth. This need for positioning has been increasingly focused on constrained environments, such as urban or indoor environments, where GNSS (Global Navigation Satellite System) is known to have significant limitations: multipath as well as the lack of Line-of-Sight (LOS) satellite visibility degrades the GNSS positioning solution and makes it unsuitable for some urban or indoor applications. In order to improve the GNSS positioning performance in constrained environments, many solutions are already available: hybridization with additional...

In recent years, many aircraft manufacturers have proposed inno-vative cockpit concepts based on touchscreens. Although having alarge number of advantages, this type of solution suffers from severelimitations in operational use, in particular, eyes-free interactionis nearly impossible and touchscreens are extremely complex touse during turbulent conditions. We examined the contribution ofphysicality to overcome these weaknesses by introducing a shape-changing touchscreen that offers folds on which the user hand canrest. This surface has been assessed in simulator during pilotingconditions that varied in terms of turbulences and mental workload.Results showed that the...

Motivated by uncertain parameters encountered in Markov decision processes (MDPs) and stochastic games, we study the effect of parameter uncertainty on Bellman operator-based algorithms under a set-based framework. Specifically, we first consider a family of MDPs where the cost parameters are in a given compact set; we then define a Bellman operator acting on a set of value functions to produce a new set of value functions as the output under all possible variations in the cost parameter. We prove the existence of a fixed point of this set-based Bellman operator by showing that it is contractive on a complete metric space, and explore its relationship with the...

[...]