Synthesizing Manipulation Sequences for Under-Specified Tasks using Unrolled Markov Random Fields

TL;DR

This paper introduces a dynamic planning method for robotic manipulation in unpredictable human environments, using unrolled Markov Random Fields trained on example sequences to generate flexible, high-level task plans.

Contribution

It presents a novel MRF-based scoring function and training approach for robust, adaptable manipulation sequence synthesis in unstructured environments.

Findings

Successfully plans manipulation sequences in diverse scenarios

Generalizes well to unseen environments

Demonstrates robustness in unpredictable settings

Abstract

Many tasks in human environments require performing a sequence of navigation and manipulation steps involving objects. In unstructured human environments, the location and configuration of the objects involved often change in unpredictable ways. This requires a high-level planning strategy that is robust and flexible in an uncertain environment. We propose a novel dynamic planning strategy, which can be trained from a set of example sequences. High level tasks are expressed as a sequence of primitive actions or controllers (with appropriate parameters). Our score function, based on Markov Random Field (MRF), captures the relations between environment, controllers, and their arguments. By expressing the environment using sets of attributes, the approach generalizes well to unseen scenarios. We train the parameters of our MRF using a maximum margin learning method. We provide a detailed empirical validation of our overall framework demonstrating successful plan strategies for a variety of tasks.