Interpretable Responsibility Sharing as a Heuristic for Task and Motion Planning

TL;DR

A novel heuristic for Task and Motion Planning (TAMP) named Interpretable Responsibility Sharing (IRS), which enhances planning efficiency in domestic robots by leveraging human-constructed environments and inherent biases and thereby improving the interpretability and effectiveness of robotic planning.

Abstract

This article introduces a novel heuristic for Task and Motion Planning (TAMP) named Interpretable Responsibility Sharing (IRS), which enhances planning efficiency in domestic robots by leveraging human-constructed environments and inherent biases. Utilizing auxiliary objects (e.g., trays and pitchers), which are commonly found in household settings, IRS systematically incorporates these elements to simplify and optimize task execution. The heuristic is rooted in the novel concept of Responsibility Sharing (RS), where auxiliary objects share the task's responsibility with the embodied agent, dividing complex tasks into manageable sub-problems. This division not only reflects human usage patterns but also aids robots in navigating and manipulating within human 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, thereby improving the interpretability and effectiveness of robotic planning. Experiments conducted across various household tasks demonstrate that IRS significantly outperforms traditional methods by reducing the effort required in task execution and enhancing the overall decision-making process. This approach not only aligns with human intuitive methods but also offers a scalable solution adaptable to diverse domestic environments. Code is available at https://github.com/asyncs/IRS.

Figures (6)

• Figure 1: This study emphasizes the common human factor in both environment construction and task specifications present in household tasks (e.g., serving, cleaning, and caring), and how it influences Task and Motion Planning (TAMP). We highlight the role of human bias, represented through the use of auxiliary objects such as trays in kitchens, which are often created by humans for convenience. Our research proposes a systematic method to utilize these biases in an interpretable manner, aiming to enhance the efficiency and effectiveness of TAMP in human-designed environments.
• Figure 2: We propose a novel heuristic, Interpretable Responsibility Sharing (IRS), for Task and Motion Planning (TAMP). We begin by curating a dataset consisting of first-order logic definitions of initial states and goals, with labels derived from counterfactual scenarios indicating whether the use of auxiliary objects would benefit the agent. The Optimized Rule Synthesis (ORS) component then generates a set of rules that define the appropriate conditions for the agent to use these objects. If the initial state and goal do not violate these rules, the agent adopts responsibility sharing by dividing the original problem into sub-problems dependent on the auxiliary objects. IRS enhances the effectiveness and interpretability of existing TAMP formulations by incorporating human bias through the ORS-guided decision mechanism.
• Figure 3: Counterfactual Plan Generation (CPG) for Dataset Construction
• Figure 4: Optimized Rule Synthesis (ORS) for Generating Responsibility Sharing (RS) Conditions
• Figure 5: Experimental settings used in simulation and human experiments. In the simulation, a north sign is placed to identify the tables mentioned in the task descriptions.