Robotics in natural environments: Four projects are currently running to support our activities around robotics in natural environments. ANR MUSCAA is interested in predicting how the terrain appearance is related to mobility performance. In PEPR NINSAR, we work on semantic 3D mapping in the context of precision agriculture, a topic that will be continued in Horizon Europe project MSCA DN AIGreenbots. In parallel, ANR R³AMA and CARNOT SAWASP projects explore the deployment of robust agent trained in simulation with Reinforcement Learning, with a focus on unmanned surface vessels.
BugWright2: BUGWRIGHT2 is a H2020 European project, started in January 2020, completed in March 2024, coordinated by Prof. Cédric Pradalier.
The objective of BUGWRIGHT2 was to bridge the gap between the current and desired capabilities of ship inspection and service robots by developing and demonstrating an adaptable autonomous robotic solution for servicing ship outer hulls. By combining the survey capabilities of autonomous Micro Air Vehicles (MAV) and small Autonomous Underwater Vehicles (AUV), with teams of magnetic-wheeled crawlers operating directly on the surface of the structure, the objective was to acquire a global overview of the structure while performing a detailed multi-robot visual and acoustic inspection of the structure.
WoodSeer: The WoodSeer project has been funded by the French ANR. Its purpose is to evaluate the use of machine learning to predict the interior distribution of defects inside roundwood from the external geometry of the wood. In addition to the machine learning tools, the DREAM team designed a specialised tree scanner producing high-resolution model of trees as illustrated in the picture below.
