Batch Recurrent Q-Learning for Backchannel Generation Towards Engaging Agents

Abstract

The ability to generate appropriate verbal and non-verbal backchannels by an agent during humanrobot interaction greatly enhances the interaction experience. Backchannels are particularly important in applications like tutoring and counseling, which require constant attention and engagement of the user. We present here a method for training a robot for backchannel generation during a human-robot interaction within the reinforcement learning (RL) framework, with the goal of maintaining high engagement level. Since online learning by interaction with a human is highly time-consuming and impractical, we take advantage of the recorded human-to-human dataset and approach our problem as a batch reinforcement learning problem. The dataset is utilized as a batch data acquired by some behavior policy. We perform experiments with laughs as a backchannel and train an agent with value-based techniques. In particular, we demonstrate the effectiveness of recurrent layers in the approximate value function for this problem, that boosts the performance in partially observable environments. With off-policy policy evaluation, it is shown that the RL agents are expected to produce more engagement than an agent trained from imitation learning.Keywords human-robot interaction · engagement · partially observable Markov decision process · batch reinforcement learning

Learning Gaze Behaviors for Balancing Participation in Group Human-Robot Interactions

“…Different to [10], we opted for batch RL due to the risk of random action exploration and errors that could compromise interactions with users. In this regard, [30,31] inspired our work. These prior efforts used human interaction data and batch RL to learn non-verbal behaviors that aim to increase engagement in HRI; however, learned policies have not been deployed on a robot to the best of our knowledge.…”