Development of obstacle detection and avoidance system based on integration of different based-sensor for small-sized unmanned aerial vehicle using cues from expansion of feature points and direction of flow field vectors
Achieving a reliable obstacle detection and avoidance system that can provide an effective safe avoidance path for small unmanned aerial vehicle (UAV) is very challenging due to its physical size and weight constraints. Prior works tend to employ the vision based-sensor as the main detection sensor...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English English |
| Published: |
2020
|
| Subjects: | |
| Online Access: | http://eprints.uthm.edu.my/10802/1/24p%20MUHAMMAD%20FAIZ%20RAMLI.pdf http://eprints.uthm.edu.my/10802/2/MUHAMMAD%20FAIZ%20RAMLI%20COPYRIGHT%20DECLARATION.pdf http://eprints.uthm.edu.my/10802/3/MUHAMMAD%20FAIZ%20RAMLI%20WATERMARK.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Achieving a reliable obstacle detection and avoidance system that can provide an effective safe avoidance path for small unmanned aerial vehicle (UAV) is very challenging due to its physical size and weight constraints. Prior works tend to employ the vision based-sensor as the main detection sensor but resulting to high dependency on texture appearance while not having a distance sensing capabilities. Besides, vision-based sensor detection system suffers from creating a trusted safe avoidance path due to inability to detect the free region. The previous system only focused on the detection of the frontal obstacle without observing the environment as a whole which is strictly not resemble the real environment. On the other hand, most of the wide spectrum range sensors are heavy and expensive hence not suitable for small UAV. In this thesis, integration of different based-sensor was proposed for a small UAV obstacle detection and avoidance system. Cues from expansion of the features points are used to extract the depth information of the environment and classify the region in the predictable obstacle appearance situation. In the unpredictable obstacle appearance situation, the detection of the obstacle is done by analysing the flow field vectors in the image frames sequence. The proposed system was evaluated by conducting the experiments in a real environment for both of the observed situations, which consisted of different configuration of the obstacles. The results show that the proposed system able to create the safe avoidance path regardless of the texture appearance (e.g. poor texture or textureless) and size of the obstacle. It also able to handle multiple obstacles with the distance of the introduced side obstacle was up to 270 cm from the UAV platform. In addition, the success rate for the sudden introduced obstacle experiments is high which is 70 % and above. It is also found that the safe avoidance path by the proposed system will depend on the situation and position of the obstacle in the environment. Finally, the obstacle appearance in the image views plays a critical role in deciding the direction of the safe avoidance path. |
|---|