With 2020 wrapping up and all of us here at UAS@UCLA finished with a successful Fall Quarter, I want to take a moment to reflect on the accomplishments the Hardware Team achieved throughout this research challenge.
As the pandemic hit and forced everyone home for the duration of Spring Quarter, UAS@UCLA was motivated to remain an active organization and decided to pursue the NASA Undergraduate Student Research Challenge. We came in with the goal of developing drone swarm technology to indefinitely lift a small payload. Our proposal, titled AVIATA (Autonomous Vehicle Infinite Air Time Apparatus) and relevant to NASA’s ARMD (Aeronautics Research Mission Directorate) Strategic Thrust 6 initiative: Assured Autonomy for Aviation Transformation, was submitted back in June. Specifically, our Hardware Team was tasked with developing the drone swarm system, the frame each drone attaches too, and the docking mechanism.
After a successful proposal, our team first had to crowdfund $2000 in order to receive the NASA grant and continue working on the project. We surpassed this goal and ended up raising over $2,500 in about a month of crowdfunding on our GoFundMe page! Once finished with crowdfunding, we dug deep into the technical aspects of our project, and presented our design in a Preliminary Design Review at the end of October.
Our initial frame design was an octagonal shape, however, due to the benefits of easily mounting an Apriltag (for long range docking identification) and payload to an octocopter-like frame, we opted for the latter design. This design consists of 8 boom arms protruding from a central frame made out of carbon fiber plates. On the top plate would be the large Apriltag, and secured to the bottom plate would be a payload bay. This payload bay is a rectangular box with a door that opens to securely fit a box that holds a 2kg box, or the payload. Landing legs on 4 of the carbon fiber booms allow for the frame to land without damaging the docking mechanisms and payload.
For our drone swarm, we intended on using commercial off the shelf (COTS) hexacopters. In order to generate more thrust from each drone, we opted for using S4 batteries rather than S3 batteries so that we could maximize the thrust output of the drones without having to worry about motor saturation when lifting the frame and payload. This was determined after performing motor thrust tests with both S3 and S4 batteries. We also plan on further modifying the drone to contain the docking attachment and necessary communications systems.
Lastly, our docking system, probably one of the most complex tasks for the Hardware team, was designed with two main systems: the docking joint on the frame, and the docking attachment on the drone. The docking joint on the frame has a “funnel” that guides the drone’s docking attachment to form a connection that does not require power to sustain. It also has two permanent magnets of opposite polarities that are used to align the drone in the proper orientation with the funnels. The docking attachment is connected to the bottom of each hexacopter and contains internal rack and pinion keys that are used to slide into the funnels. It also has battery powered electromagnets that work in conjunction with the permanent magnets on the docking joint for alignment and undocking.
Overall, while we have made significant progress on our Hardware design with AVIATA, there is still much to be done with implementing and iterating the physical hardware. Our next steps involve the structural strength testing of components, further motor/power testing, and testing and increasing robustness of our docking system in preparation for our Critical Design Review at the end of Winter Quarter. Thanks for your engagement this quarter, and we hope you stick with us through the next one!
The software side of AVIATA has continued to work with and improve our simulations to prepare for running our code on physical drones. We’ve had a strong start to AVIATA, and have high hopes for 2021.
The docking team has been focused on implementing the 3 stage docking system previously mentioned - using a large central target to help the drone find AVIATA, then using it to align and fly to the peripheral target, and finally using that peripheral target to actually get close enough to dock. An important part of this process is how the drone decides to move based on where it sees the target. To do this, we are using a PID controller which we’ve spent the last week tuning. Meanwhile, other members of the team are working on implementing the other stages of docking. Specifically, we are finalizing stages 1 and 3, which involve homing in on a target using an on-board camera, and continuing work on stage 2, which involves flying from the big central target on the frame to one of the 8 peripheral targets at our specified docking location. Finally, we are improving our target detection algorithm by generating 3D simulations in Gazebo, a drone simulation software.
Figure 1: The PID tuning process for the controller that regulates our altitude, where the yellow line represents the final parameters we decided to use. You can see that the drone is able to stabilize itself at the desired altitude within 1 second, which is much better performance than any of the other trials.
The controls software subteam, which has become our largest and most complex subteam, has continued to work on the collaborative nature of AVIATA. A few modifications were necessary to the flight controller software, named PX4, to support this, while the groundwork has been laid for inter-drone communication using custom ROS2 messages.This is necessary for having the leader drone inform the follower drones how they should be flying, since they’ll all be attached to one another. All of this work is being done with the goal of running this C++-based controller program on the Raspberry Pi that will accompany each drone.
Finally, the communications team has completed the software-focused research and initial testing of the OSLR-v2 mesh networking library. We have moved on to physical testing by ordering WiFi 6 antennas and plan to use these with Raspberry Pis we already have to set up a proof-of-concept mesh network. This will provide the framework for inter-drone communications that the controls team will depend on, as we hope to have it operational by the time our Critical Design Review comes up.
With 2020 ending, we’d like to thank everybody who’s been reading these blogs and keeping up with our progress. It’s been a very long and eventful year, and we’re looking forward to continuing in 2021. Next year, we’ll be starting off strong by presenting our work directly to NASA. An internal poster session will showcase all of the grant recipients across NASA’s research programs, including the USRC, which we are a part of. After our Critical Design Review in late February or early March, we’ll be integrating all the software teams and the hardware team together to have a final product by June. We hope to see you next year!
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.