Sequential Manipulation Planning on Scene Graph

TL;DR

This paper introduces a novel 3D scene graph representation called contact graph+ (cg+) for efficient sequential robot task planning, enabling complex object rearrangements through graph-based optimization.

Contribution

The work presents a new contact graph-based representation with predicate attributes and a stochastic optimization approach for goal configuration generation, advancing sequential planning methods.

Findings

Robots successfully completed complex object rearrangement tasks in simulations.

The contact graph+ representation effectively abstracts scene layouts for planning.

The method outperforms traditional languages like PDDL in complex task scenarios.

Abstract

We devise a 3D scene graph representation, contact graph+ (cg+), for efficient sequential task planning. Augmented with predicate-like attributes, this contact graph-based representation abstracts scene layouts with succinct geometric information and valid robot-scene interactions. Goal configurations, naturally specified on contact graphs, can be produced by a genetic algorithm with a stochastic optimization method. A task plan is then initialized by computing the Graph Editing Distance (GED) between the initial contact graphs and the goal configurations, which generates graph edit operations corresponding to possible robot actions. We finalize the task plan by imposing constraints to regulate the temporal feasibility of graph edit operations, ensuring valid task and motion correspondences. In a series of simulations and experiments, robots successfully complete complex sequential object rearrangement tasks that are difficult to specify using conventional planning language like Planning Domain Definition Language (PDDL), demonstrating the high feasibility and potential of robot sequential task planning on contact graph.