The seminar run from 2pm to 3pm. Please register online
Date: 19th July 2019
Thibaut Tezenas Du Montcel & Jonathan Dumon,
Grenoble Institute of Technology – INPG
Using a very fast trajectory generation method and the possibilities offered by embedded GPUs, we present a novel obstacle avoidance algorithm for quadrotors. Trajectories are generated in an evenly distributed grid which allow to spatially filter the computed costs associated with the trajectories. This spatial filter creates a security margin that is decoupled from the collision margin which finally produces a smoother and safer flight. A review on the modalities to evaluate robots capabilities will then be presented alongside a benchmark dedicated to autonomous flight evaluation.