Swarm Carrier: Aerial Deployment of Groups of Smaller UAVs
This project outlines my work with Northeastern UAV (NUAV) to develop an aerial system capable of deploying swarms of drones for rapid surveying. Beginning in fall 2018 with the construction of a heavy-lift octocopter, the subsequent success of their airframe would pave the way for concept drop tests throughout summer 2019. This article begins by discussing these early 2019 tests and then moves to current developments of fully-autonomous sUAS swarms.
Swarm Carrier was a project that I proposed to investigate the uses of deployable drone swarms for surveying and search missions. The project called for the rapid deployment, from an aerial platform, of 3+ parasite drones which would be capable of performing collaborative missions. The purpose of the project was to extend the operational range of smaller and more maneuverable UAVs with a large, but less nimble, carrier platform. Deployable swarms would increase the capabilities of surveying platforms while not being inhibited my less accessible enviornments. Initial applications that guided our work included collaborative search and rescue missions and rapid mapping tasks.
The project began with a proof-of-concept test that sought to determine if an existing 1000 mm octocopter could withstand changes in mass resulting from the deployment of smaller UAV payloads. To accomplish this, we decided to use commercially available FPV quadcopters due to their lightweight and the experience that many leads on our project had piloting them. Their use was convenient for initial drop tests due to their high-maneuverability and lack of interference with the flight dynamics of the octocopter. With this choice, the development of a payload mechanism to drop a total of 4 drones began.
In the interest of weight savings and preventing mechanical complexity, we avoided modifying the FPV drones themselves to interface with the mechanism. Knowing that the drop mechanism would have to be outfitted to a parallel set of payload rails on the octocopter, I opted to create an individually actuated mechanism for each quadcopter. This way, one or more could be removed depending on the desired payload quantity. The drop mechanisms themselves used a servo-landing gear actuator that interfaced with a 3D printed clamp that was added to the frame of each quadcopter. The servo would interface with this clamp and prevent the drone from sliding down fixed guide rails. The entire mechanism was 3D printed and laser cut out of wood to decrease weight.
The mechanism proved to be minimally invasive and easily loadable using Mission Planner to actuate individual servos. With the completion of the drop frame, the first test sought to determine the altitude the drones and their pilots would need to perform reliable recoveries after drops. Footage from this test can be seen below!
Due to the minimal weight of the FPV quadcopters and drop mechanisms, there was little to no effect on the flight performance of the octocopter. Moreover, the quadcopters were easily recoverable after drops. The success of single drop tests prompted a scale up to 4 smaller UAVs. The following videos show the missions from that day which include stationary, moving, and rapid drops.
The success of the multi-drop tests proved that the octocopter was capable of withstanding the impulses of in-flight payload drops. Deployments with servos using Pixhawk were also trivial but would require revision for subsequent missions. Now that swarm deployments were mechanically possible, the next step was to develop a means of dropping autonomous UAV(s) from the octocopter.
Work on this project paused in the fall of 2019 for a wave of improvements to the octocopter in preparation for heavy-payload tests with AerospaceNU’s NASA Student Launch Team. You can read more about those tests at the end of my post on the development of the octocopter here. Those tests and subsequent thrust analyses showed that the octocopter had a payload limit of 10 lbs while maintaining reasonable flight characteristics and times. This presented issues relating to the octocopter’s ability to carry heavier vehicles, especially after the development of our autonomous swarm drones which weighed 2.3 kg at best. It became clear that the Tarot octocopter would be unable to facilitate the drop of multiple swarms of autonomous UAVs. However, while a larger airframe was obviously needed, it was decided to use the octocopter to perform another proof-of-concept drop test. But this time, it would drop a single autonomous UAV that would stabilize, perform a mission, and then reintegrate with the carrier airframe.
This second proof-of-concept test is the current pursuit of this project as of spring 2020. This mission calls for a system capable of deploying a single autonomous UAV at an intermediate altitude following takeoff. The parasite vehicle will autonomously stabilize, perform a mission, and then land on the Tarot octocopter. To simplify reintegration, the Tarot octocopter will land at an intermediate position during the parasite drone’s mission and present a stationary landing pad on its top plate for reintegration. Following a successful landing, the entire system will return to the original takeoff position and justify a scale up to sUAS swarms as had occurred following the 2019 tests.
This project is still in active development but you can expect continued project updates to be made to this post as time progresses! For now, here are prototypes of the drop and reintegration mechanisms that the system will employ in the pursuit of the above mission criteria.
Following a wave of payload designs over the summer, the drop frame was manufactured and fitted to our club’s octocopter for testing. Designed to house new and larger LiPo flight batteries, a companion computer, and drop mechanisms, the frame proved incredibly rigid and load-bearing, even with 3D printed connectors.
Although lightweight and strong, the payload frame’s 3D connectors showed signs of needing to be replaced. In addition to their non-favorable geometry for FDM, regardless of orientation, one of the corner flanges would always be prone to shearing along the layer lines of the print. At the same time, the composite construction of the frame verified the feasibility of newer and more complex payload frames. This success drove me to pursue stronger and more easily manufacturable connectors for lightweight tube0frame assemblies.
In addition to corner joinery, a new payload dubbed the multidrop mechanism, required T and tetrahedral connectors. I designed this payload to give larger aerial vehicles the ability to drop, store, and reintegrate up to 6 autonomous drones from modular bays. Much like the landing pads and payload frames above, the multidrop required robust joinery with reliable manufacturing capabilities. From these requirements, I designed novel, composite connectors that utilized clamping of carbon plates and 3D printed parts in compression to give reliable fixing methods for all 3 joint types.
Using full parameterization and master modeling techniques, I was able to easily incorporate these connectors into joints of larger frame assemblies such as the multidrop. Thanks to their design as multibody parts in SolidWorks, the multibody drop was surprisingly lightweight and compartmentalized, in spite of it being one of the largest assemblies I have ever designed.
The multidrop assembly itself is a frame of storage bays for autonomous swarm drones. Using an actuated, gravity-fed storage column, the design allowed for drones to be integrated via top landings, uniquely stored in multiple levels, and dropped sequentially using pulleys attached to linear actuators.
My club projects plan on using 2 of these bays centrally mounted within a 30″ propeller octocopter. As the final Swarm Carrier system, this vehicle will be able to deploy a fleet of 6 smaller drones during flight. Substantial prototyping is underway to assembly the first bay before the end of the fall 2020 semester!