BEGIN:VCALENDAR
VERSION:2.0
CALSCALE:GREGORIAN
PRODID:adamgibbons/ics
METHOD:PUBLISH
X-PUBLISHED-TTL:PT1H
BEGIN:VEVENT
UID:dgFtPOspVMm-1tjWVALva
SUMMARY:Just like the simulations: Teaching robots to play chess 
DTSTAMP:20260430T144231Z
DTSTART:20260522T141000Z
DESCRIPTION:Description:\nChess pieces are heavy and our arms are weak. Luc
	kily\, there are strong robots – and with simulation environments and Robo
	t Foundation Models (RFMs)\, teaching them to move chess pieces has never 
	been easier. In this talk\, our Innovation Hacking team teaches you how to
	 train your robot in simulation environments and make it perform in the re
	al world. \n\nThey open with an introduction to RFMs\, what makes them spe
	cial and how they are used. Then they hop to a deep dive of NVIDIA's Gr00t
	\, how it's built\, how it was trained\, how you can use it. After that\, 
	they castle over synthetic data generation: They will show you how you can
	 generate robot training data in NVIDIA Isaac Sim and how to further enhan
	ce the data with generative AI. Finally\, it's time to promote their simul
	ated model to the real world: Our speakers explore how to close the sim2re
	al gap and show how far they can actually push the language and reasoning 
	capability of their trained model. \n\nEn passant\, they capture the conte
	nt with code examples and live demos that show both the impressive capabil
	ities and the hilarious failures of this project. Spoiler alert: Not every
	 game ends with a checkmate. \n--------------------------------\n\nSpeaker
	:\n- Jonas Mayer\n- Tobias Lübbe\n\n--------------------------------\n\nTa
	lk details:\n- Link to the Big Techday website: https://bigtechday.com/en/
	talks#4Bp9gpggSJsDj128KAU9Cp\n
LOCATION:Kohlebunker II
DURATION:PT50M
END:VEVENT
END:VCALENDAR