how to install collision avoidance library

[Thanos254]

Since I would like to get collision avoidance on my RTF drone, Method 1 of the build and install step on the collision and avoidance library github page would be the step I would like to take. While trying to follow the instructions, I cant find the meta-coav folder and I cant find the steps to add the yocto layer on the provided link. I would be very grateful for help, I dont have a whole lot of experience. | Since I would like to get collision avoidance on my RTF drone, Method 1 of the build and install step on the collision and avoidance library github page would be the step I would like to take. While trying to follow the instructions, I cant find the meta-coav folder and I cant find the steps to add the yocto layer on the provided link. I would be very grateful for help, I dont have a whole lot of experience.

[avinash-palleti]

Hi @octavian111,

The links in the wiki seems to be broken, Please follow below steps to get collision avoidance library into yocto image.

1. collision-avoidance-library recipe is available at https://github.com/intel-aero/meta-intel-aero/tree/master/recipes-support/collision-avoidance-library
2. To add this as part of image, add IMAGE_INSTALL += "coav-control" line in https://github.com/intel-aero/meta-intel-aero/blob/master/recipes-core/images/intel-aero-image.bb
3. Rebuild the yocto image as per these instructions

[Thanos254]

Does it also work to install on Ubuntu with ROS image?

[MathiasDePaepe]

@Thanos254 I'm also using the Intel Aero RTF drone and I have the same issue. Did you find an answer for your question? I'm a newbie and I have no idea how I can install the library.

Greets
Mathias

[anselmolsm]

@Thanos254 @MathiasDePaepe on Ubuntu you need to build it, see https://github.com/intel/collision-avoidance-library#method-2---compile-and-install-yourself

[jonathan84clark]

I am new as well. I built it (coav-control), installed it and placed it in my Aero root directory. The drone still flies towards a wall and makes no attempt to avoid it. I am using Ubuntu.
I have tried the following:

1. Built and installed exactly as the instructions indicated (drone flies but does not respond to obstacles)
2. Created a startup ssh script that starts coav-control. (added code to start my script in rc.local). (/usr/local/bin/coav-control -d DI_OBSTACLE -a QC_STOP -s ST_REALSENSE) It runs but never indicates it is even trying to avoid an obstacle.

I am piping the output of coav-control to a file and I am seeing nothing interesting. The software just ignores obstacles. I could fly the drone right into a wall and it wouldn't do anything.

Has anyone ever got this working on an actual drone (Intel RTF drone)? Or does it just work in simulations?

[rchiossi]

Do you get any output after the "Waiting for vehicle..." message?

[jonathan84clark]

Yes the last message I get is
[MavQuadCopter] mavlink_vehicle instantiated

Then nothing.

[rchiossi]

It seems to me that no data is arriving through the socket. Are you using PX4 or ardupilot? Can you use other program that rely on mavlink (e.g. QGroundControl). If so, are you using more than one at once?

[jonathan84clark]

QGroundControl connects fine. I can fly the aircraft with my remote. It is running PX4.

[jonathan84clark]

Any other suggestions? Is this library supposed to work on Ubuntu for Aero?
Should I use a different library?
Should I put Yocto back on the drone?
Should I start putting debug printf statements in the code and debug it? (since it appears to be broken)
Any help would be great.

[szebedy]

Hi @jonathan84clark. I think we are facing the exact same issue (Also running Ubuntu on an Aero RTF with PX4). Please let me know if you were able to solve the issue, and I will also keep you updated in case it starts working for me!

[jonathan84clark]

Thanks for the info. I am glad I am not the only one struggling with this. I wish we could get more support from Intel. I may need to go in a different direction and find an obstacle avoidance product/software that has more support.

[szebedy]

@jonathan84clark I think this library is just a framework for testing and benchmarking collision avoidance strategies (as it is stated on the main page) and it won't prevent the drone from flying into a wall unless you integrate it in your offboard code.