Visual Navigation Machine Learning Competition

No GPS, no problem !

Today drones and other aerial autonomous systems for navigation and control depend heavily on the robustness of GNSS reception for their position estimation. The objective of this competition is to ask users around the world to contribute to TOPO’s research by building and training a model that can be used to estimate the position of the drone only with ground pictures.

To train the model, real and synthetic pictures are provided. The two best competitors will be awarded 8000 CHF and 2000 CHF respectively if they manage to overtake the in-house benchmark!

In order to set the challenge up, TOPO and ENAC-IT4R has developed an scoring pipeline to asset users’ submissions. The Codalab platform has been chosen to host this competition after a substantial market analysis. This open service allows the execution of python scripts to evaluate submissions based on custom metrics. They can run on internal or external worker (the process that run the script) and can therefore be easily adapted to different use cases, from standard machine learning competition to algorithms benchmarking.

More information on the competition website :