3nd week: Troubles keep appearing, but more applications have been done.

On this third week I continued with the follow_road, but several problems keep appearing. First of all I wanted to thanks Pedro Arias for trying to help me resolve all the problems.

Now the drone takes off correctly, but it didn’t respond to the ros topic’s messages I sent, so it didnt actually move. After some config changes, the drone moved, but always tried to go back to the takeoff point.

The problem here was that the drone was not in “OFFBOARD” mode. After running a simple python script to make the drone keep the OFFBOARD mode (it goes back to mission mode if u dont request OFFBOARD mode every 5s), it stayed in this mode and followed the topic velocities, but it tries to return to the takeoff point resulting in some movements like this.

After all the changes and research, we couldn’t find what the problem was, so it remains unresolved. :C

But not everything in this week was as bad as this. I also developed some new applications.

One of them is “flipped”, a VisualCircuit block I created my self to rotate and image 90º. My camera has only 15fps (frames per second) so all the frequencies I used are 15, but it works with higher freqs.

Also I used a VisualCircuit block called “object_detector” that uses yolo and darknet to find different objects. Here I found a couple of errors that I fixed and I will also create a new issue involving these errors. Here is a video that shows the “object_detector” block working.