- Abdulla Hil Kafi
- Raihana Shams Islam Antara
- Maisun Ibn Monowar
Remote sensing is directly dependent of satellite images as updated data are critical for geological and meteorological data. Since Bangladesh has no satellite of its own, GIS researchers are heavily defendant upon buying or acquiring permission to use image captured by some other satellite. An immediate alternative is to use a UAV to map a target area. However, flying a UAV requires skills that requires practice. Therefore, an autonomous platform is required which can fly and capture images on its own upon receiving target coordinates.
“Porjobekkhok” is a modular designed quadcopter which serves as an alternate platform for remote sensing. Equipped with open source hardware and software, it promises Geological researchers to cut down time and cost associated with their land survey. Its modular design allows geological researchers to add their own software for faster and more efficient processing. In favorable condition, It is sufficiently programmed to fly without any pilot. Modular design of “Porjobekkhok” allows the user to rebuild entire quadcopter using locally available parts and tools, which makes any repairs and rebuild super easy.
While designing “Porjobekkhok”, our UAV model, special emphasis were given on using locally available parts. For on board computing, we used ArduPilot, an open source robust flight computer giving us the maximum flexibility. The body is made for PVC plastic. PVC pipes enabled us to create something light and cheap for a quadcopter without exceeding our budget.
Figure 1: Main body of “Porjobekkhok”.
Creating a reliable core software is as essential as building a good quadcopter frame. For flight control and navigation we used “Mission Planner”, which is open source and compatible with ArduPilot. Our main challenge was to create a software which could successfully acquire and process images. We could successfully create three demo applications.
Based on the color of crop we can estimate the condition of crop on field.
- Original Image b. Separation Of crop from other object
c. Portion of the image found healthy d. Numerical estimation of crop health
Figure 2: Crop health estimation.
We used CCD camera to capture our images. In order to over the limitation of small resolution camera, we wrote a software to create super resolution image from video.
Figure 3: Super resolution image from video.
Bangladesh is a riverine country. To forecast river bed corrosion we capture and process of a target site at a certain interval. This would enable us to predict river bed corrosion with some degree of confidence.
- Test image b. Processed image
Figure 4: Riverline detection
In order to make our project more usable, we created a LabView interface to run image processing. This enables third party developers to add their own algorithms.
Figure 5: LabView interface.
We can also acquire flight telemetry data in real time. This enables us not to look at the images only, but also exact attitude of the camera while the image is being captured.
Figure 6: Altitude data from GPS with respect to time.
Figure 7: 3D plot of our test flight at East Shewrapara (GPS data plotting on Google earth)
Abdulla Hil Kafi
Raihana Shams Islam Antara
Maisun Ibn Monowar
Citation Link: http://hdl.handle.net/10361/4213