Zero-shot generalization using cascaded system-representations

TL;DR

This paper introduces CASNET, a novel framework that learns a single control policy for classes of analogous systems, enabling zero-shot generalization in robotics without additional training.

Contribution

CASNET leverages cascade neural network encoders and decoders to create general-purpose system representations, facilitating zero-shot transfer across similar control systems.

Findings

CASNET achieves zero-shot optimal performance on unseen robot models.

The framework generalizes across different robot morphologies.

Empirical results show performance is limited by the learning algorithm, not the framework.

Abstract

Deep reinforcement learning has been applied to solve a variety of control problems in isolation. However, the learned latent representations cannot be optimally reused for other analogous tasks and/or control systems without additional training or tuning. In this regard, we propose a novel framework that can be used to learn a single control policy for a whole class of analogous control systems. The framework is abbreviated as CASNET and it leverages the similarities in the designs of analogous control-systems to learn general-purpose abstract system-representations. The framework uses a cascade of recurrent neural networks-based encoders to create these representations which are then fed to a conventional policy network as input. A similar cascade of decoders decodes the output of the policy network to generate system-specific output. We illustrate the effectiveness of this framework on arguably the most significant use-case of DRL: Robotics. In this paper, we use CASNET to learn generalizable control policies for two separate classes of robots: planer-manipulators and crawling robots, using 15+ and 55+ morphologically analogous simulated robots respectively. These robot models encompass the most common design variations used in the real world. Our empirical results using state of the art on and off policy learning algorithms show that on average, CASNET agent achieves zero shot optimal performance (performance equivalent to expert agents trained for individual robot models) on unseen robot models. These results illustrate that the performance of the learned policy is bound the learning algorithm rather than the framework itself. The proposed framework serves a major step towards universal controllers.