Deep Neural Network for Real-Time Autonomous Indoor Navigation

TL;DR

This paper presents a real-time indoor navigation system for micro aerial vehicles using a deep learning-based controller that enables a quadcopter to find a specific target with a single camera, overcoming GPS limitations.

Contribution

It introduces a practical deep learning approach employing a ConvNet to control a quadcopter for indoor navigation and target detection using only visual data.

Findings

Successful real-time indoor navigation in diverse environments

ConvNet effectively mimics expert pilot decisions

Visualization techniques enhance understanding of the learned model

Abstract

Autonomous indoor navigation of Micro Aerial Vehicles (MAVs) possesses many challenges. One main reason is that GPS has limited precision in indoor environments. The additional fact that MAVs are not able to carry heavy weight or power consuming sensors, such as range finders, makes indoor autonomous navigation a challenging task. In this paper, we propose a practical system in which a quadcopter autonomously navigates indoors and finds a specific target, i.e., a book bag, by using a single camera. A deep learning model, Convolutional Neural Network (ConvNet), is used to learn a controller strategy that mimics an expert pilot's choice of action. We show our system's performance through real-time experiments in diverse indoor locations. To understand more about our trained network, we use several visualization techniques.