Skyviper V2450gps and Image Recognition

ArduCopter SkyViper
1

Hi there, Ken here.
I’m a Multi-noob. Noob to Linux, Noob to Python, Noob to Drones, Noob to ArduPilot, and Noob to other stuff I don’t even know I need to know yet!!!..but I am patient and persistent and will do what I have to to do what I want. I do have a background with Windows applicaiton software development and I started my career as an electronics technician.
I am extremely appreciative that SkyRocket Toys provided the SkyViper platform so I could explore the ArduPilot project…thank you. I also would like to express my sincere admiration and appreciation to all the individuals that contribute their time and efforts developing the ArduPilot project. My thanks to all!!! I can’t believe this is free!!! Absolutely AWESOME.
After trying to fly manually (which I suck at 'cause I can’t see well) I found the Mission Planner program. I was able to create missions and send the SV around my property…however it lacked accuracy using just GPS coordinates.
I am trying to accurately position my SkyViper V2450GPS at various way points. Using Mission Planner and GPS coordinates gets me close, but not exact. I thought I would try using a companion computer and image recognition to locate and hover over a marker…like an ‘X’ pattern. My plan is to use the Mission GPS data to get close to a waypoint and then kick in the image recognition to accurately position the drone over the ‘X’.
My initial questions are broad and have to do with the preferred way to get the image/video from the Skyviper and send commands to accurately position the drone over the ‘X’ pattern.
I did a lot of reading on how it can be done but nothing specific for the STM32 or Sonix board. I mainly found posts on how to interface with a Pixhawk flight controller and companion computer. I did a lot of thinking about how to make that work with a SkyViper but I still don’t have enough knowledge to know what can and can’t be done with the SkyViper hardware to get it to work.
I plan on using a RaspberryPI3 as a companion computer.

I would appreciate any assistance to get me going on the right track…
For instance, to get images from the SkyViper
Do I enable a com port on the Sonix board and hardwire a serial connection between it and a companion computer?
The Sonix Board provides a web interface…do I use MavLink commands and send them to its UDP port 14550 to get images?
Do I use OpenCV?
Do I use APSync?
Do I use APWeb?

Generic SkyViperV2450GPS interface questions
What does Pixhawk compatible mean? Is there an API that I can use with the SkyViper to ‘interface’ with it’s hardware?

Thank you for any assistance.

Best regards,
Ken

1 Like
2

I can’t provide any assistance at all, but I’m very interested to see if this is possible and what insights others might have.

1 Like
3

I wish i could help but I’m just going to follow this topic

4

So, to get the photos off the Sonix, you could use this

The Ardupilot (and APSync etc) don’t transmit the photos (AFAIK). But the Sonix board with APWeb does. And this document in the link does show how you can get the live video into an off-board computer.

However, I think you’re going to find that you’re not going to have very good performance. First off, you’re limited to the FPS of the onboard camera (live feed is only 15FPS) and then you’re going to be limited by the storage time + transfer rate of the video to your CC. Then your CC needs to process the video, determine a X/Y/Z error, then translate that into roll/pitch/yaw commands for the copter all at about probably close to 50hz or more to have stable performance…

I just don’t think that is going to happen. Even if you had a supercomputer doing the video processing, the FPS of the camera is 15FPS, and that’s it. It only hits 30FPS when storing to the SD card, and even that can sometimes be spotty.

Oh, and then there is the image resolution and the “jello”

anyways, that doc above would get you started. Just didn’t want to drop a link on you without providing some technical gotchas that you’re going to face.

good luck!