Advanced Deep Learning: Object Picking based on CAD model and 2D Camera (train AI models with synthetic data)

From CAD to AI robots: learn to train robots to detect and pick objects with precision! The objective of this course is to learn how to train an object pose estimation neural network for robot picking applications. This includes:

1. Theoretical introduction on AI training from CAD file
2. Hands-on experiment: rendering of synthetic data, application of network on images, communication with robot
3. Discussion of alternative pose estimation and picking strategies

Agenda

• 09:00 - Welcome coffee
• 09:30-12:00 - Theoretical introduction on AI training from CAD file
• 12:00-13:00 - Lunch
• 13:00-15:00 - Hands-on experiment: rendering of synthetic data, application of network on images, communication with robot
• 15:00-17:00 - Discussion of alternative pose estimation and picking strategies

About Reboot Skills

The course is developed under the framework of the EU-funded REBOOT SKILLS project, an initiative designed to enhance the digital expertise of professionals in the European manufacturing sector, addressing the digital skills gap exposed by rapid digitization and the COVID-19 pandemic. Supported by the Digital Europe program, it collaborates with academia, research, and industry to offer accessible training in four key areas: Artificial Intelligence, Internet of Things, Robotics, and Cybersecurity.