How do you localize a gas source as fast and accurate as possible?
Gas leaks can pose serious threats to human safety and cause severe environmental damage. Robotic Gas Source Localization (GSL) leverages mobile robots to efficiently localize gas sources. However, this task remains inherently challenging due to the complex interplay between gas diffusion and wind flows. GSL methods often struggle to generalize to unstructured and three-dimensional environments.
This project aims to utilize multicopters for GSL missions and enhance their robustness using the following main ingredients. First, replace classical algorithms or selected components of them by machine learning techniques. Second, employ advanced wind sensing and modeling to better understand gas transport in the air. Third, extend the algorithms to multi-robot systems. All efforts aim to make GSL more robust in complex three-dimensional environments that include diverse obstacles, wind conditions, and gas release rates.
Collaborators
Research Period and Sponsors
This project started in September of 2024. SNSF is sponsoring this project for a duration of 4 years.
Related Student Projects and Internships
- DISAL-SP200: Jad Benabdelkader, Gas Source Localization using Neural Networks
- DISAL-SP199: Thomas Cirillo, Optimizing Gas Sensor Placement on Micro Drones via CFD and Wind Tunnel Experiments
- DISAL-SP198: Badil Mujovi, Drag Model Identification and Wind Prediction with “Off-The-Shelf” Drones