Discovering State and Action Abstractions for Generalized Task and Motion Planning
Aidan Curtis, Tom Silver, Joshua B. Tenenbaum, Tomás Lozano-Pérez, Leslie Pack Kaelbling
Abstract
Generalized planning accelerates classical planning by finding an algorithm-like policy that solves multiple instances of a task. A generalized plan can be learned from a few training examples and applied to an entire domain of problems. Generalized planning approaches perform well in discrete AI planning problems that involve large numbers of objects and extended action sequences to achieve the goal. In this paper, we propose an algorithm for learning features, abstractions, and generalized plans for continuous robotic task and motion planning (TAMP) and examine the unique difficulties that arise when forced to consider geometric and physical constraints as a part of the generalized plan. Additionally, we show that these simple generalized plans learned from only a handful of examples can be used to improve the search efficiency of tamp solvers.
中文速览
广义规划（generalized planning）本可以帮机器人"一次学习、处处适用"，但以往方法只能处理离散符号世界，无法应对机器人操作中不可回避的几何碰撞、连续位姿等物理约束。GenTAMP 提出了一套完整框架：先从少量示例轨迹中自动提取包含连续特征的状态抽象，再通过 SAT 求解筛选出有效特征子集，最后把这些特征构造成一个全观测非确定性（FOND）规划问题，求解后得到一个可跨问题实例复用的广义策略；执行时，该策略被用来约束任务与运动规划（task and motion planning, TAMP）求解器的搜索空间，而非直接逐步执行，从而规避了贪心决策导致的几何可行性破坏。实验表明，用少量小规模示例学到的广义计划能大幅加速标准 TAMP 求解器在更大、未见过的问题实例上的搜索效率。这项工作填补了广义规划与连续机器人操作之间的空白，为构建既可扩展到大量物体、又能泛化到新环境的自主机器人系统提供了新路径。
原文 arXiv:2109.11082;中英对照 + 大白话阅读 https://aha.fim.ai/paper/2109.11082v1