Cascade Attribute Network: Decomposing Reinforcement Learning Control Policies using Hierarchical Neural Networks

TL;DR

The paper introduces the Cascade Attribute Network (CAN), a hierarchical neural network that decomposes complex reinforcement learning control policies into attribute modules, enabling zero-shot policy assembly for robot control tasks.

Contribution

It presents a novel hierarchical neural network architecture that decomposes control policies into attribute modules, facilitating reusable and adaptable reinforcement learning policies.

Findings

CAN effectively decomposes control policies into attribute modules.

Zero-shot assembly of control policies achieves ideal performance.

Validated on multiple robot control scenarios with various attributes.

Abstract

Reinforcement learning methods have been developed to achieve great success in training control policies in various automation tasks. However, a main challenge of the wider application of reinforcement learning in practical automation is that the training process is hard and the pretrained policy networks are hardly reusable in other similar cases. To address this problem, we propose the cascade attribute network (CAN), which utilizes its hierarchical structure to decompose a complicated control policy in terms of the requirement constraints, which we call attributes, encoded in the control tasks. We validated the effectiveness of our proposed method on two robot control scenarios with various add-on attributes. For some control tasks with more than one add-on attribute attribute, by directly assembling the attribute modules in cascade, the CAN can provide ideal control policies in a zero-shot manner.