Suggestion for designing UAV swarm

asked 2018-11-24 04:28:50 -0500

AbdullahIrfan gravatar image

Before starting, YES I READ THE GUIDELINES "general debugging/programming questions not specific to ROS", I don't have anyone to guide and this one is my last hope, so expecting positive response!

I have been asked to design UAV swarm (two or three quad copters) in the following phases:

  1. Software in the loop testing to verify quad copters stability and communication.
  2. Hardware in the loop testing to justify model.
  3. Rapid prototyping for implementation

The objectives of the project are:

  1. Are completely autonomous
  2. Communicate with each other
  3. When sent to an area, they distribute survey area, perform operations (for beginners lets say they will record video of that specified area) and return to the base
  4. (If possible) Live streaming will also be done.

In order to do this project I have 2 options:

  1. I use ROS kinetic with Gazebo for SITL and HITL verification. Pixhawk 4 micro-controller will be used.
  2. I use NI LabVIEW for SITL and HITL verification. cRio microcontroller will be used with PIX provided by LabVIEW.

Personally I like option 1, I have googled and found some work on it (up to the point where a person is hovering swarm of 5 quad copters). But I am new to ROS (haven't used it before) so I can't come up with technical arguments to prove ROS/Kinetic with gazebo to be better option than NI.

edit retag flag offensive close merge delete


As far as I know 'completely autonomous' Drones working in any area other than a cleared open volume is still an active research area. The collision avoidance challenge is getting there but as far as I know doesn't have a complete solution.

PeteBlackerThe3rd gravatar image PeteBlackerThe3rd  ( 2018-11-24 12:13:58 -0500 )edit

Yes and there are several methods introduced for it but at the moment this isn't my concern, I want to finalize one of the above mentioned two options. Since this is ROS forum, how do professionals see this project, I am interested in that.

AbdullahIrfan gravatar image AbdullahIrfan  ( 2018-11-24 12:25:31 -0500 )edit

To be honest, if you're new to ROS I would not advise you to start with a project as complex as this. It's certainly capable of achieving everything you want, but it will take time to learn it well enough to do this.

PeteBlackerThe3rd gravatar image PeteBlackerThe3rd  ( 2018-11-25 14:02:47 -0500 )edit

I do have enough time (at least 1 year and around 6 months), and I have to learn either of them problem is which stream to chose and why (with strong points). I don't have those points for ROS.

AbdullahIrfan gravatar image AbdullahIrfan  ( 2018-11-25 22:43:29 -0500 )edit

I am only familiar with option 1 so I can tell you it's possible but I can't give a fair comparison of the two. Give the amount of time you have could you not spend a month implementing a single drone experiment on both systems and use that experience to decide which way to go?

PeteBlackerThe3rd gravatar image PeteBlackerThe3rd  ( 2018-11-26 04:48:03 -0500 )edit

Amount of Budget approved for this project is limited to buy only 1 equipment. Also NI is costly (my rough estimation for price of one UAV is around $12 K. That's the reason why I am under pressure.

AbdullahIrfan gravatar image AbdullahIrfan  ( 2018-11-27 00:24:25 -0500 )edit