Heavy-lift UAV Design and Modification
In the fall of 2018, myself and the other project leads of Northeastern UAV (NUAV) proposed Swarm Carrier: a project to research deployment of drone swarms from a large UAVs for rapid surveying. Such a project demanded a large multirotor with a heavy payload capacity as well as the design of compact swarm vehicles, substantial software development, and actuated payloads to facilitate drops and reintegration. This page focuses on the development of the large multirotor but you can find more information about the swarm drones and payloads on my other project blogs!
The first goal of the project was a proof of concept drop test to determine if large UAVs using Pixhawk for control could withstand changes in an airframe’s center of mass and overall weight in flight. It was decided that if a sufficiently large mulitrotor could sustain flight during these changes, as a result of parasite vehicle drops, then the project would be further developed. The test called for an airframe with a payload capacity of ~2 kg which was the estimated combined weight of 4 FPV quadcopters that would act as the test vehicles. The airframe’s design would have to be simple and cost effective to make sense for us as student group with the intention of using it for future tests outside of the Swarm Carrier project.
To meet these criteria, I decided to go with a commercially available frame: the Tarot Iron Man 1000. With an assembled weight of just 1.6 kg and a capability to mount 15″ propellers, the airframe could achieve a thrust of up to 8 kg depending on the power systems used. In addition to being lightweight, the airframe was cheap, simply designed, and had ample payload space. Other parts included a Pixhawk 4 flight controller or autonomous missions and 8 x 4114/320 Kv brushless motors for an octocopter configuration to achieve maximum lift capabilities.
The frame required modifications for mounting electronics and making wiring connections durable. Because we had started with an existing airframe, the first step of the project was to reverse engineer its components and re-create the assembly in CAD.
After assembly, the task of wiring the motors and ESCs was conducted. For the first version of the airframe, no power distribution board was easily mountable so a custom solution was required. Moreover, fixtures to mount the ESCs under the motor mounts would be needed to safely pass wires into the arms and mount LEDs to indicate vehicle orientation in flight. I chose to modify the arms with XT60 connections to allow their interface with a central, octagonal hub of power connections that would distribute power to the motors and flight electronics. This design would allow the arms to be easily removed for maintenance while having a high-amperage PDB to sustain the draw of all 8 motors. This early PDB design can be seen below:
Other custom fixtures included an early soft mount for the flight controller and payload-facing FPV camera. With the PDB scheme manufactured, the wiring of 8 motor-ESC subassemblies was also completed. At this point, wire management had become a substantial issue as each arm contained ~2 m of wire, with all 8 converging at the central hub which had very little space to pass signal wires through. This made assembly fairly difficult, especially due to the fact that the entire frame body was only rigid when the plates completely sandwiched the arms and closed off all access to the central wire hub. These thoughts would motivate future modifications to the frame but for now, assembly continued.
At this point, the project hit a major barrier due to issues with the first assembly. Initial motor tests were unsussessful with inconsistent twitching and faulty connections. It was suspected that delicate ESC signal wires had been damaged during the assembly of the central frame so multiple dissassemblies and tests were conducted. In the end, I designed exterior arm covers to contain the delicate ESC signal and ground wires to simplify wire routing. This let us easily troubleshot ESC connections, check ESC firmware, and perform motor tests.
Using Mission Planner and an ArduCopter install, the airframe was configured and the motors mapped to the correct octocopter configuration. Following successful plug-in and motor tests, all that was left to do was the field test!
Flight tests were a success, until the last flight, and marked NUAV’s first practiced usage of autonomous waypoint missions for airframe testing. The final weight of the vehicle was ~ 5.5 kg with all electronics and a 16 Ah 6s LiPo flight battery. Without an additional payload, flight times of >15 minutes were achieved and stable flight observed with basic and spline waypoint missions. However, just as we started taking the vehicle’s performance for granted, an unfortunate crash occurred. On the final mission of the day, the octocopter dropped from ~5 m after I forgot to raise the throttle before switching the vehicle back to manual for landing. But to our surprise, the damage had been absorbed by 3D prints that we were using to hold the landing gear to the base plate! The drone had survived the day intact but a full systems check was still in order and already, plans for the next wave of improvements were being discussed. Here are videos of the waypoint missions and crash from that day!
Now well into the summer of 2019 timeline of this project, development shifted to the proof of concept deployment test which used FPV drones to assess the octocopters reaction to in-flight payload drops. Feel free to check out details on that project here as the rest of this article focuses on improvements made to the octocopter!
Fast-forwarding to fall 2019, NUAV had successfully deployed drones from the octocopter and was now seeking improvements to its design in support of heavier payload tests. The first of these included a reliable means of attaching the servo-actuated landing gear and replacing the central PDB hub. The reason for the replacement of the custom PDB was due to reliability and accessibility issues. Specifically, the XT60 interfaces were tough to align during the already difficult assembly of the Tarot airframe. The custom PDB also had no ability to organize the ESC signal and ground wires which left them vulnerably mounted to the top plate. A commercial solution was found that allowed all 8 arms to plug into a central hub with pinouts for signal wires, power to flight electronics, and AS150 plugs to the flight batteries.
While ideal for wire management, the new PDB had no stock means of mounting to the octocopter. In addition to this, a lack of mounting options for the servo-actuated landing gear prompted me to design a custom base plate to be waterjet cut out of carbon fiber. The plate features mounting holes for the landing gear and PDB as well as a large access hole to pass wires up through the center of the existing Tarot frame.
The base plate was cut out of plywood on a Full Spektrum laser cutter first to test fits and hole iterations. Tolerances were made dynamic so that a test waterjet cut of various hole sizes could determine the correct design criteria. The plate mounted the new PDB under the frame which was now refitted with lower payload rails to mount the battery and payloads. FDM printed landing gear and plate supports were also designed to support the sides of the new base plate which protruded outwards to bolt into the landing gear (the black plastic brackets on the side arms in the video below).
The purpose of the Fall 2019 modifications was to support the octocopter’s use as a versatile test platform for the entirety of our aerospace club: AerospaceNU. Having announced that goal, we were approached by Northeastern’s NASA Student Launch team with a request to use the vehicle to test deployments of rocket payloads. The payload to be tested was a guided parafoil vehicle that would be ejected from the team’s rocket, deploy a parachute, and perform a flight mission before landing on the ground to conduct further experiments. The vehicle itself was a large cylindrical object, coupled with an actuated deployment frame, which weighted ~10 lbs. This presented serious challenges for the usage of the airframe which had a maximum payload capacity that was yet to be determined. Furthermore, there was significant concern that the payload would act as a lever arm under the octocopter, causing it to roll or pitch until irrecoverable. Yet, the airframe had never been more ready to pursue such a test, due to the recent base plate and PDB modifications, and we were apt to push its limits. Therefore, our projects decided to collaborate on a series of heavy-payload drop tests.
The semester concluded with a full-systems drop test of the guided parafoil. The vehicle and drop actuator was wired as a servo to be controlled from a telemetry link to Mission Planner. During the test, the octocopter would be flown by a pilot and co-pilot handling the ground station and actuation of the drop mechanism. Preliminary thrust tests showed that the octocopter was theoretically capable of lifting the payload’s weight but not without affecting the ease of handling the aircraft in flight. Tests went forward with this assumption and a custom landing/takeoff frame was built to service the drone and payload before flight. Here are some pictures and videos from the day of the test!
On the day of the test, the vehicle was reduced to <10 lbs to maximize the probability of favorable flight dynamics. Following tests of the octocopter’s new frame and PDB configuration, a successful lift test was performed with the payload! While tolerancing of the drop mechanism caused the payload to detach itself upon landing, little damage was incurred. Crucially, the drop of the payload only slightly affected the altitude of the autopilot which quickly corrected and was easily recoverable. While ultimately, we were unable to deploy the parafoil due to issues with its parachute deployment mechanism, flights showed that the octocopter was capable of reliably lifting >8 lbs!
While successful, the drop tests definitely pushed the maximum capabilities of the octocopter airframe. Flags included a reduction in flight time and very poor handling with the payload in manual flight modes. But, while the roll and pitch authority was definitely decreased, it was not unmanageable in either manual or autonomous flight modes. With the conclusion of the semester, the next phase of the octocopter’s development would be its specific application to the Swarm Carrier Project.
After the NASA Student Launch drop tests, the Tarot 1000’s modifications were complete and the vehicle deemed capable of pursuing more advanced missions and payloads. Subsequent payloads included an antenna array for locating crashed rockets and additional drop mechanisms to support NUAV’s new swarm UAVs.
The fall 2020 semester saw a new wave of flight tests of the octocopter, this time for testing swarm and new rocket payloads. The first of these included single drops of autonomous UAVs and a high altitude, high capacity drop of a parachute deployment mechanism!