Leila Takayama wins Google Research Award for robot training

Professor Leila Takayama talks to Kevin Weatherwax through BEAM
Professor Leila Takayama talks to Kevin Weatherwax through BEAM
Wednesday, March 6, 2019
James McGirk

A recent Faculty Research Award from Google will help Associate Professor Leila Takayama learn what it takes for a robot to learn from its human peers.

According to Takayama, robots are like new human hires: They need onboarding to do their jobs successfully. While there are often formal guidelines for new [human] employees, she points out that many of the most important rules are implicit (e.g., who can call meetings with whom, how to catch the boss for a quick meeting) and learned through situated “on-the-job” learning. If robots could be designed to elicit feedback and training from their human peers, they’ll be able to integrate better and perform more complex tasks.

Takayama’s award will fund a study titled “Making collaborative robots more teachable: Empirically exploring how to elicit feedback from untrained end-users.” She plans to engineer a series of human-robot interactions in which users encounter a robot that solicits feedback and help from them.

They’ll be testing which dimensions most strongly impact a user’s willingness to provide feedback. These explorations may include framing the robot as a mentee or apprentice; expressing gratitude; giving credit to its trainers; or showing off performance improvement. The experiments will explore how people and robots will perform tasks such as navigating busy workspace corridors.

Leila Takayama began teaching at UC Santa Cruz in 2016. She recently joined the Baskin School of Engineering’s Computational Media department, where other faculty study artificial intelligence and human computer interaction.

Takayana holds a PhD in Communication from Stanford University (with a minor in Psychology), is the founder of Santa Cruz-based Hoku Labs, a human-robot interaction consultancy, and has held research positions at Google X, Willow Garage, Nokia Research, PARC (Palo Alto Research Center), and Stanford’s Communications between Humans and interactive Media (CHiMe) lab.