A Path Planning Method with Perception Optimization Based on Sky Scanning for UAVs.

Songhe Yuan,Kaoru Ota,Jianghai Zhao,Mianxiong Dong

doi:10.3390/s22030891

Abstract

Unmanned aerial vehicles (UAVs) are frequently adopted in disaster management. The vision they provide is extremely valuable for rescuers. However, they face severe problems in their stability in actual disaster scenarios, as the images captured by the on-board sensors cannot consistently give enough information for deep learning models to make accurate decisions. In many cases, UAVs have to capture multiple images from different views to output final recognition results. In this paper, we desire to formulate the fly path task for UAVs, considering the actual perception needs. A convolutional neural networks (CNNs) model is proposed to detect and localize the objects, such as the buildings, as well as an optimization method to find the optimal flying path to accurately recognize as many objects as possible with a minimum time cost. The simulation results demonstrate that the proposed method is effective and efficient, and can address the actual scene understanding and path planning problems for UAVs in the real world well.

Highlights

A serious problem with Unmanned aerial vehicles (UAVs) is that their battery is limited and cannot be used for long-term or large-scale survey tasks
H, the playground is detected as a building and it is meaningless for a drone to search that area. We find this error can be decreased by improving the confidence threshold set
Taking the object confidence as a computing factor, we proposed an optimization method to find the optimal flying path to realize effective and efficient detection with a minimum time cost

Summary

Introduction

How to develop a feasible flight path for efficiently and accurately target detection has become a very important issue This task mainly includes two aspects, namely path planning [1] and scene understanding [2]. Most of the previous methods consider this task as two totally-separated problems They may not be well applied to real-world detection tasks in disaster management. Deep learning methods for object detection have been applied in many areas including automatic driving, medical application, urban research, and so on [3,4,5] These kinds of technology can be utilized on scenes understanding for path planning.

Methods

Discussion

Conclusion