Teresa Conceicao
Enhancing AI-based Robotics with Simulation Workflows
#1about 3 minutes
Understanding the limitations of pre-programmed robots
Pre-programmed robots often fail when environmental conditions change, highlighting the need for more adaptive AI-driven autonomy.
#2about 3 minutes
Core requirements for developing AI-powered robots
AI-based robots require massive amounts of diverse data, extensive parallel training, and rigorous testing, which are challenging to achieve in the real world.
#3about 3 minutes
An overview of NVIDIA Omniverse and Isaac Sim
NVIDIA Omniverse provides a platform for creating physically accurate virtual worlds, while Isaac Sim offers specialized tools for robotics simulation.
#4about 3 minutes
Getting started with your first robot in Isaac Sim
Learn the initial development steps by running a "Hello World" example, importing a robot model via the URDF importer, and inspecting its properties.
#5about 2 minutes
Building and collaborating on simulation environments
Leverage Omniverse's live-sync capabilities to collaborate on creating rich simulation environments by connecting to industry tools like Revit and Rhino.
#6about 4 minutes
Programming robot behavior with Python, OmniGraph, and ROS
Explore various methods for controlling robots, including direct Python APIs, the OmniGraph visual programming interface, and integration with the Robot Operating System (ROS).
#7about 3 minutes
Using synthetic data generation for AI training
Overcome the challenges of real-world data collection by using simulation to generate perfectly labeled, diverse synthetic data for training perception models.
#8about 5 minutes
Closing the sim-to-real gap with domain randomization
Mitigate the appearance and content gaps between simulation and reality by using domain randomization to create more robust and generalizable AI models.
#9about 3 minutes
Real-world examples of simulation-trained robots
See how partners like Fraunhofer, Festo, and ETH Zurich use Isaac Sim to develop, train, and test advanced robots for logistics and collaboration.
#10about 2 minutes
Resources for getting started and final Q&A
Find resources like documentation, developer forums, and conference talks to learn more, followed by a Q&A on procedural versus handmade data.
Related jobs
Jobs that call for the skills explored in this talk.
Featured Partners
Related Videos
24:48How Robots Learn to be Robots
Alexander Schwarz
35:16How AI Models Get Smarter
Ankit Patel
29:34Robots are coming into the wild! Full-Stack Robotics Engineers, be ready!
Falk-Moritz Schaefer
38:12Developing an AI.SDK
Daniel Graff & Andreas Wittmann
27:26Staying Safe in the AI Future
Cassie Kozyrkov
42:18On the straight and narrow path - How to get cars to drive themselves using reinforcement learning and trajectory optimization
Francis Powlesland & Elena Kotljarova
22:07WWC24 - Ankit Patel - Unlocking the Future Breakthrough Application Performance and Capabilities with NVIDIA
Ankit Patel
19:24Finding the unknown unknowns: intelligent data collection for autonomous driving development
Liang Yu
From learning to earning
Jobs that call for the skills explored in this talk.
AI Engineer (Santiago de Compostela)
NeuralTrust
Municipality of Santiago de Compostela, Spain
Python
PyTorch
TensorFlow
Machine Learning