We at Unmanned Aerial Systems at UCLA welcome the new year, and are very excited looking forward to pushing ahead with our projects!
In the mechanical airframe sub-team, we have most recently been looking into sourcing our drone fleet. In our initial proposal and designs, we are using hobbyist hexacopters, particularly the F550 model, which had used to be produced by DJI, and has since been taken over by 3rd parties since DJI discontinued them. This has been one of the issues with finding a quality and reputable supply, as supply is both rare and expensive that fits our specifications.
Since we are outfitting our drones with additional equipment such as a Raspberry Pi and our custom docking mount, as well as being able to lift the payload frame, we need a stronger propulsion system. In switching our onboard battery design to 4S (16V), the drones can achieve satisfactory performance.
This has since been our main focus as both the controls and docking sub-teams are nearly ready to begin hardware testing with cooperative control and computer vision tracking, respectively. Another mechanical airframe subteam involvement with these endeavors has been the design of a mounting adapter that connects two drones together through carbon fiber tubes to test the minimum cooperative control setup.
Figure 1: Two different mount designs to connect two drones with carbon fiber spars to test minimum cooperative control.
Our next goal is to finalize our drone fleet design, order them, and assemble them. As a priority, we will first be figuring out shipping the current drones we have on hand to the sub-teams to conduct their testing. These tests are preparing us for our CDR (Critical Design Review) this quarter, before we begin integrating each sub-team’s subsystem for our tech demo hopefully in June!
The software team has been hard at work writing code for each subsystem and testing via simulation. We have successfully demonstrated distributed control as well as vision-guided flight using simulated AprilTags. Mesh networking needs a little more tweaking before we can fully test it, so stay tuned for when that happens.
Firstly, here is the distributed control simulation, demonstrating a leader drone commanded to takeoff and land (left), and a follower drone copying its attitude & thrust commands (right). Communication is implemented via ROS2.
And here is our docking simulation, with the 3D world shown on the left and what the drone sees from its perspective on the right as it approaches the AprilTags.
Obviously, this is a long way from our final product. One limitation of the available simulation tools is the inability to connect multiple vehicles in a rigid body (and it’s already very computationally intensive). Previously, we had simulated this using a simplified model which did not incorporate the actual flight software. To put these things together, there is hardware on the way which will allow the first real-world test of cooperative control. The temporary test structure using two drones is shown in this week’s hardware blog.
Potentially, the full rigid structure could be simulated along with the flight software, but this would require significant modification of the already complex simulation tools. Because we are able to conduct real-world tests, the benefit of simulation is not very appealing given the cost of implementation.
We are in anticipation of the results of this first real-world controls test, which will inform us if and how we should implement alternative control methods -- there are many potential ways to accomplish cooperative control, and we want performance to guide our decision on which to use.
These results are based upon work supported by the NASA Aeronautics Research Mission Directorate under award number 80NSSC20K1452. This material is based upon a proposal tentatively selected by NASA for a grant award of $10,811, subject to successful crowdfunding. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of NASA.