Human-Aware Robot Task Planning Based on a Hierarchical Task Model
Human-Aware Robot Task Planning Based on a Hierarchical Task Model
复制标题
DOI：
10.1109/lra.2021.3056370
复制
发表时间：
2021-04
影响因子：
5.2
通讯作者：
Yujiao Cheng;Liting Sun;M. Tomizuka
中科院分区：
文献类型：
--
作者：
Yujiao Cheng;Liting Sun;M. Tomizuka
当机器人与人类合作完成任务时，它们需要在考虑人类行动的同时规划自己的行动。然而，由于任务的复杂性和人类合作者的随机性，这是相当具有挑战性的机器人有效地与人类合作。为了解决这一挑战，在这封信中，我们首先提出了一个算法，自动构建一个分层的任务模型，从单代理演示。分层任务模型明确地捕获了任务在所有抽象级别上的顺序和并行关系。然后，我们提出了一个基于优化的规划器，它利用并行关系并优先考虑与人类动作并行的动作。通过这种方式，可以避免潜在的空间接口，可以减少任务完成时间，并可以提高人的满意度。我们进行了模拟的机器人手臂与人类合作的几个合作任务。与几个基线的比较结果证明，我们提出的规划器是更好的效率，安全性和人的满意度。
When robots work with humans for collaborative task, they need to plan their actions while taking humans’ actions into account. However, due to the complexity of the tasks and stochastic nature of human collaborators, it is quite challenging for the robot to efficiently collaborate with the humans. To address this challenge, in this letter, we first propose an algorithm to automatically construct a hierarchical task model from single-agent demonstrations. The hierarchical task model explicitly captures the sequential and parallel relationships of the task at all levels of abstraction. We then propose an optimization-based planner, which exploits the parallel relationships and prioritizes actions that are parallel to the humans’ actions. In such a way, potential spatial interfaces can be avoided, task completion time can be reduced, and human's satisfaction level can be improved. We conducted simulations of a robot arm collaborating with a human for several collaborative tasks. The comparison results with several baselines proved that our proposed planner is better in terms of efficiency, safety and human satisfaction.