Learning under Misspecified Objective Spaces
This addresses the issue of robots misinterpreting human corrections in interactive learning, which is crucial for safe and effective human-robot collaboration, though it is an incremental improvement over existing methods.
The paper tackles the problem of robots learning objective functions from human corrections when the human's true objective is not within the robot's hypothesis space, which can lead to unintended learning; it proposes a method that reasons about the relevance of corrections in real time and demonstrates in a user study with a 7DoF robot manipulator that this alleviates unintended learning.
Learning robot objective functions from human input has become increasingly important, but state-of-the-art techniques assume that the human's desired objective lies within the robot's hypothesis space. When this is not true, even methods that keep track of uncertainty over the objective fail because they reason about which hypothesis might be correct, and not whether any of the hypotheses are correct. We focus specifically on learning from physical human corrections during the robot's task execution, where not having a rich enough hypothesis space leads to the robot updating its objective in ways that the person did not actually intend. We observe that such corrections appear irrelevant to the robot, because they are not the best way of achieving any of the candidate objectives. Instead of naively trusting and learning from every human interaction, we propose robots learn conservatively by reasoning in real time about how relevant the human's correction is for the robot's hypothesis space. We test our inference method in an experiment with human interaction data, and demonstrate that this alleviates unintended learning in an in-person user study with a 7DoF robot manipulator.