黑料福利网

Topic 

Safe Robot learning in the real world

Abstract

Humans are able to safely interact with the world and quickly learn new skills in novel situations. Despite the success of machine learning in several domains, we still do not see any example of such behaviors in the humanoid robotics field. I argue that the existing frameworks lack two fundamental capabilities: 1. they cannot offer a safe and goal-directed exploration strategy in novel situations, 2. they are extremely sample-inefficient which makes them impractical for learning in the real world. In this talk, I present the work of my lab in addressing these two issues leveraging tools from optimization, optimal control and supervised learning toward enabling safe and sample-efficient learning of humanoid loco-manipulation skills in the real world.

About the speaker

Majid Khadiv is an assistant professor in the school of Computation, Information and Technology (CIT) at TUM. He leads the chair of AI Planning in Dynamic Environments and is also a member of the Munich Institute of Robotics and Machine Intelligence (MIRMI). Prior to joining TUM, he was a research scientist at the Empirical Inference Department at the Max Planck Institute for Intelligent systems. Before that he was a postdoctoral researcher in the Machines in Motion, a joint laboratory between New York University and Max Planck Institute.

Since the start of his PhD in 2012, he has been performing research on motion planning, control and learning for legged robots ranging from quadrupeds, lower-limb exoskeleton up to humanoid robots.

On March 10 Professor Khadiv will give a lecture titled: Beyond tele-operation; generating large-scale data for training humanoid VLAs.

Your host

Alessandro Saccon, Associate Professor at the department of Mechanical Engineering.

is required but free of charge.