by Tracy R. Walsh
Robots with a fixed gaze don’t seem that convincing:
Sean Andrist at the University of Wisconsin-Madison and his team have a number of other tricks to make a humanoid robot seem “alive.” One is to introduce small random movements into a robot’s head rotation motor, so that instead of appearing stationary, the robot’s head twitches slightly now and then. A face-tracking camera can ensure a robot always looks at the person it is interacting with, but instead of staring straight at their face, the team have programmed in a tendency for the robot to avert its gaze from time to time. The idea is to mimic the human habit of glancing fleetingly to one side when thinking of an answer to a question.
The team asked 30 students to assess conversations with Nao robots programmed to act like librarians or job interviewers, some of which had been set up for gaze aversion. They found that people thought robots that glanced around seemed more purposeful and thoughtful.
Previous Dish on robot-human interaction and the uncanny valley here, here and here.
The Dish 