Robots learn to take initiative and help without being told
Robots That Take Initiative: A Framework for Building and Evaluating Proactive Robots
RoboticsArtificial Intelligence
Summary
Robots often wait to be told what to do, which limits how helpful they can be. This paper introduces a way to make robots more proactive, letting them figure out what needs doing on their own. The authors show that testing these proactive robots by just simulating people can be misleading, so they created a better test where the robot’s actions influence how the human behaves. They also developed a new method called GAP that learns from watching people and can predict what they want, helping more effectively than past approaches.
What this means in practice
- •For robotics engineers: Design robots that can independently identify and perform tasks users need without explicit commands.
- •For smart home device makers: Create home assistant robots that proactively support occupants by predicting and acting on their goals.$Commercial implications: Enables sale of more intuitive and helpful home robots that improve user convenience by anticipating needs.
Authors
Maithili Patel, Sonia Chernova
Abstract
Effective robot assistance beyond narrow roles and repetitive tasks requires robots to be proactive - to decide what needs to be done rather than waiting to be told. While proactivity is increasingly explored, it lacks a unified formulation, and work in the domain is typically evaluated offline against static human models that cannot capture the effect of a robot's actions on the environment and the user's own behavior. We introduce a unified formalism for proactive robot assistance, organize it into three levels, and provide a framework to address the highest level of unprompted proactive assistance. We then show that offline evaluation overstates performance in this setting, and contribute a closed-loop evaluation with a human model that adapts to the robot. Finally, we present a method, GAP, that instantiates our framework, learning from passive observation to anticipate user goals and act. Under closed-loop evaluation, prior state-of-the-art methods collapse, in some cases adding more work than they save, while GAP remains robust and substantially outperforms them.