Abstract
DJI Guidance is an entirely new visual sensor navigation system, which has a function of real-time transmission of the depth of field images and other data. It is a robust optical sensing platform which significantly promotes the development of unmanned aerial vehicle (hereinafter referred to as UAV). On this basis, we have further studied and developed Guidance in application to the field of UAVs to also cut a striking figure, by applying the threshold and binaryzation to the single frame of a static image generated from Guidance. Utilizing OpenCV’s contour detection algorithm, we find all the closed contour information and re-plan the feasible region. Also, we specially re-project the depth information to achieve three-dimensional reconstruction of the image. Then combining the re-planed feasible domain with three-dimensional reconstruction, we can perform real-time obstacle avoidance and mapping. The results demonstrate that our improvement can efficiently complete the obstacle avoidance and real-time mapping in the specific circumstances.
Highlights
In recent years, more and more attention has been paid to UAVs as advanced technology, and the research of UAV autonomous obstacle avoidance has become one of the focuses of the study
The area length and width dimensions were 30 cm, and the height of the area was determined from the height histogram of the pixels falling in the area, and the risk of the region was determined according to the region and its coordinate distance. The implement of this method may result in the expansion of obstacle in the depth and the width maximum to 60cm so that the maximum diameter of the feasible region can be reduced by 60cm
We focus on the processing of depth image and binary image, re-plan the feasible region
Summary
More and more attention has been paid to UAVs as advanced technology, and the research of UAV autonomous obstacle avoidance has become one of the focuses of the study. For an obstacle avoidance system, the detection and location of obstacles is the first step and the most critical step to obtain the surrounding environment. The widely used methods of detection and location can be classified into two categories: one is based on UAV's simultaneously localization and mapping (hereinafter referred to as SLAM) [1] to surrounding environment. The accurate location information of UAV and stereo relationship between UAV and obstacles is obtained. The obstacle avoidance of UAV is explored by use of DJI Guidance
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have