Human-robot interaction in any form and application domains are susceptible to failures. Failures are of varied definitions and types. The severity of the failure is high for teleoperated tasks. In this paper, we focus on the less explored type of failure called human induced interaction failure using the concept of Intelligent Dis-obedience (ID). ID was first used as a method for training guide dogs, where the dogs disobey the user’s commands that are dangerous to perform. Imparting this behaviour of disobedience to manage human induced interaction failures consists of numerous social, cultural and ethical aspects to consider. This paper discusses a novel framework based on ID to manage human induced interaction failures. Also, we discuss a nuanced approach involved in robot disobedience considering the different social and cultural aspects.
This paper was peer-reviewed, accepted, and presented in the Imperfectly relatable robot workshop organised as part of the Human Robot Interaction Conference in 2023.