Bitcraze Forums

Hi guys, I'm working on my house on a crazyflie that hover by slam not by flowdeck, and it map the a specific room, and afterwards if I tell him to go to specific point 1,0,0 and in my room there was an obstacle stake and it's circular .. so I want that my crazyflie do a map to that room by a camera that I configured it on it also by using SLAM and it works fine, now my problem is how can I avoid those obstacles in my room by frames/mapping camera? for instance lets assume there was a stake in room, and this stake is an obstacle to my crazyflie's route that I told him to go to specific point ..how he can avoid this by using camera/data from mapping the room?
I'm attaching an example to be more clear, my crazyflie is map a room by using camera & SLAM, then afterwards I want to tell my crazyflie in that room that he mapped if he sees an obstacle like stake in the route that he's doing .. then avoid it and go where I told him to go.. like if I told him to go to 1,0,0 while my crazyflie inside the room, then he will go to that point and if there's avoidance he can avoid it by using camera/the MAP of the Room that I captured first ..


Any help how can I tell my crazyflie after I have the map of the room, if there's obstacle while he's doing specific route inside that room (that I have mapped ) then avoid those obstacle and go where you should go(to specific coordination that I tell him to go to .. ) ?!


thanks alot

Hi!

Really cool that you managed of doing SLAM with your Crazyflie. Now that you have the map and you are able to get the location of the CF in that map, then you go to the next step of higher level navigation: Motion planning. So based on the map, or an 'occupancy grid', you can plan the next locations of your robot in the empty spaces of the map.

Motion planning is an separate field entirely, so in order to familiarize yourself, it would be good to start looking at introductory online classes like this one of Udacity (https://www.youtube.com/watch?v=KHAu5A_flcQ) or look at the wikipage about it: https://en.wikipedia.org/wiki/Motion_planning. This is an important aspect in robotics, so do take time to understand the techniques and then you will be able to figure out with method applies to your situation.

Hi I didn't think that I understand you very well, but I'm looking for already built algorithm that's recognizing and identifying specific think like circle from image/video .. could I find an open source algorithm that's already built for this purpose? I mean for instance like slam I didn't build it .. it was already built .. so I'm looking for any already built algorithm that's recognizing specific thinks from image/video .. where can I find something like that already built?

ANY HELP PLEASE?

This is pretty far from the Crazyflie scope so I will not be able to help with that. If I had to implement such algorithm I would search for OpenCV and image processing in ROS. What you are trying to achieve has likely been done already so you should be able to find examples on the internet.