Internship Offer - Multimodal Data Analysis Using Drone-Based Multispectral And Lidar Systems For Smart Agriculture H/F Direction générale de l'Institut Mines-Télécom

Located on two main campuses dedicated to research and education in Douai and Lille, IMT Nord Europe offers research facilities of almost 20,000m² in the following areas:

- Digital science,
- Energy and Environment,
- Materials and Processes.

For more details, visit the School's website : www.imt-nord-europe.fr

Scientific Context:

Precision agriculture aims to optimize the use of inputs by integrating a wide range of sensing and data analysis technologies. Drones equipped with multispectral and LiDAR sensors offer complementary views of crop structure and physiology, while ground and weather data provide essential environmental context. Combining these heterogeneous sources enables a deeper understanding of the agroecosystem and supports AI-driven decision-making for precision irrigation.

This internship is part of the industrial chair POMME d'API, which brings together IMT Nord Europe, Polytech Lille, DOWNS, and Osiris Agriculture. Built around two PhD projects, the chair mobilizes the expertise of each laboratory and company to develop cutting-edge technologies for data acquisition, processing, and fusion aimed at improving the conditioning and irrigation of potato crops and making production more sustainable.

Osiris Agriculture is a human-scale company and an ambitious technological venture. With two pre-industrial versions deployed in 2024, its autonomous irrigation robot OSCAR is capable of delivering the precise daily dose of inputs (water and fertilizers) needed for optimal plant development.

This internship contributes directly to that effort by combining aerial (drone-based) and ground-level sensing to better understand and model crop-soil-weather interactions under real field conditions, particularly in potato test plots.

Objectives:

Develop a research-oriented workflow for multisource data acquisition, processing, analysis, and multimodal modeling. The main objectives include:

· Design and execute data acquisition campaigns using drones (multispectral, LiDAR) coordinated with ground measurements and weather data streams.

· Develop calibration and synchronization protocols to ensure consistency across sensors and sources (intrinsic/extrinsic calibration, precise timestamping, GPS/IMU alignment).

· Implement preprocessing and cleaning pipelines (radiometric, geometric, and temporal corrections) for analysis-ready datasets.

· Conduct exploratory and statistical data analysis; extract biophysical indicators per modality (e.g., vegetation indices, canopy height models).

· Build and evaluate a multimodal model integrating aerial, ground, and environmental data to characterize field conditions and support precision irrigation decisions (example model to explore: XGBoost).

Research Contributions:

The internship will contribute to ongoing research on multisensor calibration, data fusion, and multimodal learning for smart farming. The work will investigate alignment strategies and fusion schemes (early/late fusion) that combine spectral, structural (3D), and environmental variables to interpret crop status and ecosystem dynamics; results may support future publications within the PhD framework.

Expected Outcomes

· A validated multisource data acquisition protocol integrating aerial, ground, and weather data.

· A harmonized multimodal dataset (multispectral imagery, LiDAR point clouds, ground measurements, and weather records) with complete metadata.

· A research prototype of a multimodal fusion or predictive model for field ecosystem analysis, with baseline performance and ablation studies.

· A technical and scientific report documenting methodology, experiments, and results, with recommendations for PhD continuation.