Interpretable holistic manipulation strategies in household environments for task and motion planning

Abstract

Interpretable Responsibility Sharing (IRS) introduces a novel heuristic for Task and Motion Planning (TAMP), leveraging holistic manipulation strategies to enhance planning efficiency and interpretability in household environments. By systematically incorporating auxiliary objects such as trays and pitchers—common in human-constructed spaces—IRS simplifies and optimizes task execution. The heuristic is based on the concept of Responsibility Sharing (RS), where auxiliary objects share task responsibilities with robotic agents, dividing complex tasks into manageable sub-problems. This division not only mirrors human usage patterns but also aids robots in navigating and manipulating within human-designed spaces more effectively. By integrating Optimized Rule Synthesis (ORS) for decision-making, IRS ensures that the use of auxiliary objects is both strategic and context-aware, enhancing the interpretability and effectiveness of robotic planning. Experiments across diverse household tasks, including serving, pouring, and handover, demonstrate that IRS significantly outperforms traditional methods, reducing effort in task execution and improving decision-making. This approach aligns with human-inspired strategies while offering a scalable framework adaptable to the dynamic complexities of household environments.