What is the MOBILEX Challenge?
The MOBILEX Challenge (MOBILité en environnement complEXe - mobility in a complex environment) aims to meet the challenge of autonomous vehicle navigation in unstructured, complex environments. Launched by Defense Innovation Agency (AID), in partnership with French National Research Agency (ANR), the Transportation Innovation Agency and the Center National d'Études Spatiales (CNES), this project is dedicated to the research and development of technological innovations. Deployed on a 3-year periodThis new challenge allows 7 pre-selected teams to take on challenges of increasing difficulty.
The MOBILEX Challenge aims to provide new technologies to enable robots to operate in hostile environments (natural and artificial obstacles, slopes, poor GNSS coverage, etc.) and under environmental constraints (heavy rain, smoke, strong winds, etc.). It aims to reduce the physical and mental strain on operators by enabling them to autonomously manage the trajectory of ground vehicles. Ultimately, the technologies developed will be useful in more operational projects such as advanced navigation capabilities, whether on-board vehicles or remotely via assisted teleoperation, applicable in both civilian and military fields, on Earth or in space.
Of the seven teams selected to tackle these challenges, Magellium Artal Group, in association with CEA-List, a laboratory specializing in intelligent digital systems, is participating through the REFLEX team for REactive Field Locomotion EXploration.
A robotic demonstrator is provided for each team. Controlled via an interface, the operator can steer the robot, monitor its status and manage its functionalities in real time. Teams have to come up with their own solutions and innovations to make the robot autonomous.
Teams face one challenge a year, and each year must present the results of the challenge and the technologies developed to a panel of judges.
Défi 1 : L’approche innovante de l’équipe REFLEX inspirée de la robotique spatiale
L’utilisation de l’expertise historique en robotique spatiale
Challenge 1 consists of the initial design and implementation of the MOBILEX brick, in order to set up a reliable system that integrates open-source algorithmic bricks (and/or existing bricks within the consortium) and enables the main functionalities of the challenge to be ensured.
To meet this first challenge, our REFLEX team draws on Magellium Artal Group's long-standing expertise in space robotics, and more specifically in camera-based computer vision algorithms. Our team's approach is to transfer these technologies from space to terrestrial applications. Indeed, the problems and constraints encountered in this challenge are similar to those encountered for decades by planetary exploration missions such as PathFinder, Spirit, Curiosity and Perseverance, or the European Exomars project, to which Magellium Artal Group contributed the autonomous navigation software.
Stereo vision is the main perception modality for planetary exploration rovers. REFLEX's main advantage lies in the use of passive sensors, which offer numerous advantages over active LiDAR-type sensors.
While LiDARs generally have a greater range and better long-distance accuracy than stereo rigs, they behave like infrared beacons, making them ill-suited to certain operational military contexts requiring a specific level of discretion. What's more, they are not yet widely used in space robotics, notably because of their high power consumption and relative fragility.
Conversely, the use of stereo benches, consisting of two cameras, enables the robot to calculate, in a similar way to human vision, a high-frequency 3D representation of its immediate environment. As these sensors are passive, they have the advantage of being energy-efficient, not very fragile, relatively inexpensive and discreet. These characteristics make them particularly interesting for space and military applications.
In addition to the cameras, the REFLEX team has integrated various sensors: an inertial measurement system (to measure acceleration), a gyroscope (to measure orientation) and a Dual-GNSS RTK system (for precise absolute position and yaw angle).
This equipment, mounted on the robotic platform, enables the robot to understand its environment, locate precisely and navigate safely and efficiently.
Development of specific algorithms to meet the needs of the MOBILEX Challenge
The equipment added to the robot will enable it to generate a large amount of data about its environment. It will then process this data in order to model its environment, calculate the most optimal trajectory, and track this trajectory autonomously and in real time.
Several categories of algorithms have been designed by the REFLEX team to assist the robot:
- Perception to independently create a representation of its environment ;
- Localisation for precise position and orientation at all times;
- Navigation to enable the robot to move autonomously and safely in its environment;
- Supervision for error handling, calculation and data backup.
CEA-list contributes its expertise in the development of specific algorithms for trajectory calculation and tracking. The predictive controller has been developed by CEA and the trajectory tracking algorithm is currently the subject of a thesis.
Reading this data is made possible by installing an HMI based on our LittleFox solution, the video player used by French special forces. Through this HMI, our team's goal is to provide operators with a clear, user-friendly, and comprehensive interface for controlling and supervising the robotic platform. Generating a video stream in STANAG4609 format, this NATO-standardized format allows a metadata stream to be associated with a video stream and provides information such as the robot's position, orientation, and speed. Finally, the robot is controlled using a joystick to select the speed of movement.
However, even if the aim is to make the robot autonomous, the operator retains control of it during autonomous navigation. A secure stop is available on the HMI, enabling the platform to be stopped remotely.
Défi 2 : Une brique de navigation qui gagne en maturité et en fonctionnalités
Le 19 décembre 2025 à Paris, les organisateurs du Challenge MOBILEX ont présenté le bilan du second défi réalisé le 16 octobre 2025 pour l’équipe REFLEX sur le terrain de la DGA TT (Direction générale des armées – Techniques terrestres) à Bourges. Ce défi imposait un cahier des charges particulièrement dense incluant non seulement l’amélioration du code du défi 1 mais aussi le développement de nouvelles fonctionnalités :
- L’étoffement des capacités de perception (détection des obstacles négatifs, transparents, réfléchissants, détection des pentes et dévers)
- La robustesse aux conditions météorologiques dégradées (pluie, fumée)
- La résilience aux pertes de capteurs (simulation de pannes)
Lors de ce deuxième défi, l’équipe REFLEX a pu présenter ses travaux et dernières innovations. Le système de perception basé vision a permis de détecter correctement les obstacles négatifs (fossé, trous) pour lesquels les solutions reposant uniquement sur le LiDAR ont montré leur limite. Combinée au contrôle de la plateforme développé par notre partenaire CEA-List, cette détection, a permis au robot de franchir facilement les premières épreuves du défi 2 et de démontrer la pertinence des choix pris par l’équipe.
Lors du premier défi, la brique REFLEX avait déjà démontré sa robustesse aux conditions météorologiques dégradées en réalisant des scénarios de navigation autonome sous forte pluie sans impact notable sur la perception. Cela se confirme de nouveau avec le deuxième défi, où un scénario de pluie artificielle a été mis en place et l’équipe REFLEX fait partie des rares équipes à avoir traversé la zone affectée.
Un important effort de développement a été réalisé sur l’interface de contrôle afin de faciliter la préparation de la mission et de soulager la charge mentale de l’opérateur. Les informations remontées par l’IHM sont claires et permettent de comprendre les décisions du robot ; l’opérateur peut ainsi facilement intervenir pour débloquer la plateforme en cas de nécessité.
Malgré une instabilité du signal GNSS rencontrée pendant les épreuves, la brique de localisation de l’équipe REFLEX, basée sur une fusion multi-capteurs, a su prendre le relais. Cela a permis d’assurer une précision suffisante pour poursuivre les démonstrations et sécuriser le comportement du robot tout au long des scénarios.
L’équipe REFLEX prépare désormais la suite du challenge en améliorant les acquis et en répondant au défi 3 où il faudra entre autres gérer la navigation dans les hautes herbes, la détection de flaques d’eau, et la présence d’obstacles dans des pentes.
Autonomous mobility, a major innovation for many sectors
The MOBILEX Challenge is launched to revolutionise autonomous mobility technologies. This initiative aims to set tomorrow's standards with significant, long-term impacts in a wide range of sectors, from military to civil and scientific.
In the future, the technologies developed during the MOBILEX Challenge could be integrated into military operations to enhance safety and mission effectiveness in hostile terrain. Similarly, in the civilian sector, they could enable safer, more effective intervention during natural disasters, in disaster zones where ground conditions are often uncertain. Last but not least, these technologies could also be applied in the space sector, with applications such as lunar base construction, astronaut assistance or scientific exploration(cf. CNES Spaceship FR project).
The autonomy gained by the robot promises to optimize resources during interventions and guarantee uninterrupted operations, without fatigue or loss of vigilance.
The development of state-of-the-art solutions and the optimization of algorithms and sensors to maximise path completion speed will therefore be addressed in the following challenges.
Follow us on social networks to keep up to date with the progress of this project!
To download the communication kit:
