Hierarchical and Partially Observable Goal-driven Policy Learning with Goals Relational Graph

TL;DR

This paper introduces a hierarchical reinforcement learning method with a Goals Relational Graph that improves goal generalization in partially observable environments like visual navigation.

Contribution

It proposes a novel two-layer hierarchical RL framework with a Goals Relational Graph using a Dirichlet-categorical process for better goal relation modeling and generalization.

Findings

Outperforms baselines in unseen environments

Effective in goal generalization to new tasks

Applicable to both grid-world and robotic search tasks

Abstract

We present a novel two-layer hierarchical reinforcement learning approach equipped with a Goals Relational Graph (GRG) for tackling the partially observable goal-driven task, such as goal-driven visual navigation. Our GRG captures the underlying relations of all goals in the goal space through a Dirichlet-categorical process that facilitates: 1) the high-level network raising a sub-goal towards achieving a designated final goal; 2) the low-level network towards an optimal policy; and 3) the overall system generalizing unseen environments and goals. We evaluate our approach with two settings of partially observable goal-driven tasks -- a grid-world domain and a robotic object search task. Our experimental results show that our approach exhibits superior generalization performance on both unseen environments and new goals.