Creating a Uniform Framework for Task and Motion Planning: A Case for Incremental Heuristic Search? [Overview Paper]

In this short overview paper, we describe our vision for combining task and motion planning and present a historical perspective to show which parts of it have already become reality. Robots do not have to plan only once but repeatedly. Replanning from scratch is often very time consuming. Incremental heuristic search addresses this issue by reusing information from previous searches to find solutions to series of similar search tasks often much faster than is possible by solving each search task from scratch. Incremental heuristic search has mostly been used for path planning in the past but we argue that it applies to many layers of robot architectures, in particular task and motion planning, which might allow one to design very homogeneous robot architectures with clean interfaces between the layers.