2D Grid Map Generation for Deep-Learning-based Navigation Approaches

TL;DR

This paper introduces an algorithm for generating 2D dungeon-like maps with detailed attributes, along with a dataset of 10,000 maps, to facilitate training and testing deep learning-based robotic navigation methods.

Contribution

The paper presents a novel map generation algorithm and provides a comprehensive dataset to support deep learning research in robotic navigation.

Findings

Generated maps include validation of path existence and optimal paths.

The dataset contains 10,000 maps with extensive attribute information.

Maps are suitable for training and testing deep learning navigation algorithms.

Abstract

In the last decade, autonomous navigation for roboticshas been leveraged by deep learning and other approachesbased on machine learning. These approaches have demon-strated significant advantages in robotics performance. Butthey have the disadvantage that they require a lot of data toinfer knowledge. In this paper, we present an algorithm forbuilding 2D maps with attributes that make them useful fortraining and testing machine-learning-based approaches.The maps are based on dungeons environments where sev-eral random rooms are built and then those rooms are con-nected. In addition, we provide a dataset with 10,000 mapsproduced by the proposed algorithm and a description withextensive information for algorithm evaluation. Such infor-mation includes validation of path existence, the best path,distances, among other attributes. We believe that thesemaps and their related information can be very useful forrobotics enthusiasts and researchers who want to test deeplearning approaches. The dataset is available athttps://github.com/gbriel21/map2D_dataSet.git