Jeannette Bohg: Scaffolding and Imitation Learning - Human Learning Principles Transferred to Robots

Learning contact-rich, robotic manipulation skills is a challenging problem due to the high-dimensionality of the state and action space as well as uncertainty from noisy sensors and inaccurate motor control. In this seminar, Stanford Assistant Professor of Computer Science Jeannette Bohg shows how two principles of human learning can be transferred to robots to combat these factors and achieve more robust manipulation in a variety of tasks.
The first principle is scaffolding. Humans actively exploit contact constraints in the environment. By adopting a similar strategy, robots can also achieve more robust manipulation.
The second principle is learning from demonstrations through imitation. Humans have gradually developed language, mastered complex motor skills, created and utilized sophisticated tools. The act of conceptualization is fundamental to these abilities because it allows humans to mentally represent, summarize and abstract diverse knowledge and skills.
In this talk, Bohg presents work that shows how robots can adapt both principles to ease manipulation skill learning and acquire a variety of manipulation tasks with successful generalization over novel but similar instructions.
Bohg spoke on Oct. 21, 2020, as part of HAI's weekly seminar series. Learn more: hai.stanford.e...