Learning to Navigate by Growing Deep Networks

TL;DR

This paper introduces RA-DAE, a self-supervised, adaptive deep learning framework that incrementally builds and refines a neural network for robot navigation using reinforcement learning, without prior environment knowledge.

Contribution

It presents a novel self-supervised adaptive deep architecture that dynamically adjusts its structure during navigation learning, improving performance over fixed-structure models.

Findings

RA-DAE outperforms non-adaptive deep learning models in navigation tasks.

The framework effectively balances past and present information through network adaptation.

Experiments confirm robustness in both simulation and real-world environments.

Abstract

Adaptability is central to autonomy. Intuitively, for high-dimensional learning problems such as navigating based on vision, internal models with higher complexity allow to accurately encode the information available. However, most learning methods rely on models with a fixed structure and complexity. In this paper, we present a self-supervised framework for robots to learn to navigate, without any prior knowledge of the environment, by incrementally building the structure of a deep network as new data becomes available. Our framework captures images from a monocular camera and self labels the images to continuously train and predict actions from a computationally efficient adaptive deep architecture based on Autoencoders (AE), in a self-supervised fashion. The deep architecture, named Reinforced Adaptive Denoising Autoencoders (RA-DAE), uses reinforcement learning to dynamically change the network structure by adding or removing neurons. Experiments were conducted in simulation and real-world indoor and outdoor environments to assess the potential of self-supervised navigation. RA-DAE demonstrates better performance than equivalent non-adaptive deep learning alternatives and can continue to expand its knowledge, trading-off past and present information.