Promptable Behaviors: Personalizing Multi-Objective Rewards from Human Preferences
This work addresses the problem of personalizing robotic agents for diverse human preferences in complex environments, representing an incremental advancement in multi-objective reinforcement learning for embodied AI.
The paper tackles the challenge of customizing robotic behaviors to align with diverse human preferences in embodied AI by introducing Promptable Behaviors, a framework that uses multi-objective reinforcement learning to train a single policy adaptable to various preferences, and demonstrates its effectiveness in personalized object-goal and flee navigation tasks in ProcTHOR and RoboTHOR environments.
Customizing robotic behaviors to be aligned with diverse human preferences is an underexplored challenge in the field of embodied AI. In this paper, we present Promptable Behaviors, a novel framework that facilitates efficient personalization of robotic agents to diverse human preferences in complex environments. We use multi-objective reinforcement learning to train a single policy adaptable to a broad spectrum of preferences. We introduce three distinct methods to infer human preferences by leveraging different types of interactions: (1) human demonstrations, (2) preference feedback on trajectory comparisons, and (3) language instructions. We evaluate the proposed method in personalized object-goal navigation and flee navigation tasks in ProcTHOR and RoboTHOR, demonstrating the ability to prompt agent behaviors to satisfy human preferences in various scenarios. Project page: https://promptable-behaviors.github.io