Airborne Particle Classification in LiDAR Point Clouds Using Deep Learning


In autumn 2018 I started to work on my semester project on “Learning-based Smoke, Dust and Fog Detection” at the autonomous systems lab at ETH. This work resulted in a very nice collaboration with Leo Stanislas from Queensland University of Technology. 

Together, we wrote a paper about airborne particle classification, elaborating and comparing our two approaches. Further, we showed the practical importance of being able to reliably detect (and then remove) airborne particles in LiDAR point clouds. The paper was presented at the Conference on Field and Service Robotics 2019.