Description
Autonomous robotic weeding systems represent a promising alternative, as they can identify and eliminate weeds mechanically while preserving crops and minimizing soil disturbance. However, achieving reliable operation under real-world agricultural conditions remains challenging due to variations in illumination, different plant growth stages, uneven terrain, occlusions, dust, and the strong visual similarity between crops and weeds.
This PhD project aims to develop a theoretical and practical computer vision framework for an autonomous robotic weeding system, covering three closely interconnected functions: robot navigation, weed recognition and localization, and precise mechanical weed removal.
The research will focus on robust perception in challenging outdoor environments, representation learning, multimodal sensor fusion, real-time image and point-cloud processing, and model compression and optimization for on-board deployment. The project will involve the design, implementation, and evaluation of computer vision and machine learning algorithms on real robotic platforms and agricultural
Weeds directly compete with crops for water, nutrients, and light, significantly affecting agricultural yields when they are not properly controlled. The management of unwanted plants has therefore always been a major challenge in agricultural production. Reducing the use of chemical herbicides is now a key priority for sustainable agriculture.
Required qualifications and skills
We are looking for a motivated M2 (or equivalent) student in Computer Science, Artificial Intelligence, Robotics, or Computer Vision, with:
- Strong knowledge of computer vision and deep learning.
- Good programming skills in Python and/or C++; experience with PyTorch is a plus.
- An interest in robotics, autonomous systems, and real-world AI applications.
- Curiosity, autonomy, and motivation for research. Experience with ROS/ROS2 or embedded AI is a plus.
Application
To apply, please send your CV and a brief statement of motivation to : adel.hafiane@univ-orleans.fr; raphael.canals@univ-orleans.fr
