Jerome Leudet

Jerome has a strong background in AI. After a master’s in AI and VR, he developed AI for the gaming industry for more than a decade and worked in industrial simulation. With the rise of AI, he had the opportunity to begin a PhD and study how synthetic data could be used for training neural networks. He founded AILiveSim on a mission to create rich, interactive worlds to train and test algorithms.

---

Presentation

Data set creation for autonomous machines in an underground environment

Your autonomous development is intrinsically correlated to the data you use for teaching and training. You will need high-volume and/or relevant data for your specific applications. Counting only on real annotated data is limiting, time-consuming and expensive, even more so in complex environments such as ever-changing construction sites or underground tunnels. To complement an existing data set, generating data via simulation with very realistic models and easy access to more configurable parameters becomes a game-changer and can help to shorten development time and cost related to data set acquisition.