Swarms in Aerial Robotics

Hi all!

As in preparation for next week discussion, here are some questions to the aerial vehicle community who are working with teams or swarms.

  • For which application? Research, Surveillance, Shows, etcetera?
  • What are the pain points you are facing?
  • What is the limiting factor of communication?
  • Which are the important values to monitor for swarm managment?
  • And… what ever you’d like mention as part of your experience :slight_smile:

Hi Kim, hi all,

“Swarms” is a somewhat controversial term, from what I heard. We should define it in the first place I think. Usually authors remain large in their definition :
a swarm of UAVs is a set of aerial robots that work together for a specific goal.” [1]
"A swarm is generally defined as a group of behaving entities that together coordinate to produce a significant or desired result." [2]
But some other authors argue that “swarm” implies decentralized control, as it is the case in nature.

In my lab, we are working on decentralized flocking for autonomous, communicating multi-UAV systems, so I’ll talk mostly about that :slight_smile:.

  • What are the pain points you are facing?
  1. We would like to simulate both the robotics component (physics of the UAVs, command/control using PX4 Autopilot, obstacle sensing / avoidance, etc.) and the network component (quality of UAV-to-UAV and UAV-to-Ground channels, multiple data flows cohabitation with different QoS, etc.). But we lack an “off the shelf” co-simulator for robotics and network integration (that supports DDS). I am developing such a simulator as part of my Ph.D., can talk about that if that’s interesting.
  2. Positioning in a GPS-denied environment remain a pain-point. We are using decentralized UWB distance measurements atm. Other approaches exist.
  • What is the limiting factor of communication?
    I’ll stick with Bekmezci’s analysis of what is different in FANETs opposed to other MANETs [3] :
    • Mobility degree
    • Frequent topology change
    • Must support peer-to-peer and converge cast traffic
    • Longer distances between nodes
    • Different requirements for different sensors
      In general these points have several answers but people work mainly on improving the antenna systems and the routing algorithms implemented in the swarm.
  • Which are the important values to monitor for swarm management?
    On the top of my head :
    • Number of connected clusters
    • Fully disconnected UAVs (i.e. lost UAVs)
    • Average speed / maximum speed of the swarm
      But actually I’ve never really operated a swarm of UAVs in real life, so that’s all theoretical :stuck_out_tongue:

Note that many of these questions arise for our specific use-case of autonomous mission deployment, everything is different if you allow yourself to have a central base station connected to all the robots, or if you allow GPS, or fly heavier UAVs, etc.

See you tomorrow !

[1] X. Chen, J. Tang, and S. Lao, “Review of Unmanned Aerial Vehicle Swarm Communication Architectures and Routing Protocols,” Applied Sciences, vol. 10, no. 10, p. 3661, May 2020, doi: 10.3390/app10103661.

[2] M. Campion, P. Ranganathan, and S. Faruque, “UAV swarm communication and control architectures: a review,” J. Unmanned Veh. Sys., vol. 7, no. 2, pp. 93–106, Jun. 2019, doi: 10.1139/juvs-2018-0009.

[3] İ. Bekmezci, O. K. Sahingoz, and Ş. Temel, “Flying Ad-Hoc Networks (FANETs): A survey,” Ad Hoc Networks, vol. 11, no. 3, pp. 1254–1270, May 2013, doi: 10.1016/j.adhoc.2012.12.004.

1 Like

Thank you @tbalaguer for such a elaborate answer! I will include some of these in the slides for the discussion today :slight_smile: (with credit ofcourse :wink: )

Indeed, there are some clashes with the term swarm and perhaps robotics researchers that study emergence of behavior might not like it that much. It might be better to call it multi-uav or multirobot coordination. But as you suggest, to give it an definition in this context might just be good enough.

Let’s discuss more at the meeting today!

Hi Kim, we cover a lot of these points in our recent paper about the DARPA OFFSET program.

Taranta, E., Seiwert, A., Goeckner, A., Nguyen, K., & Cherry, E. (2023). From Warfighting Needs to Robot Actuation: A Complete Rapid Integration Swarming Solution. Field Robotics, 3, 460-515. https://doi.org/10.55417/fr.2023015

Unfortunately I’m not sure if I can include more detail here, since the paper had to go through a security review. Hopefully you find it helpful though!

edit: @tbalaguer has good analysis. I would add “shared medium access” to the list of pain points.
@KimMcG are you US-based by chance?
Also, @tbalaguer, I’m developing a similar simulator over here. Perhaps we can discuss.

1 Like

I’m located in Sweden currently, since my company Bitcraze’s office is there :slight_smile: what is the occasion?