ECE 110/120 Honors Lab Section : Project Proposal: Drone Swarm - Exploring Applications

Created by Unknown User (kanishk3), last modified on Sep 28, 2020

Kanishk Giri

kanishk3

ECE 110




How Sensor-Equipped Drones Serve Our Needs Today?


The use of autonomous or ground-controlled drones has made revolutionary strides in their potential for transportation and observation/surveillance based capabilities. 


There is a wide range of drone-based technologies that have advanced motion-based characteristics in programming and pushed forward the development of sensors

and logistics. These advancements enable drones to fulfill required niche directives that have been employed within various sectors, such as agriculture, construction/infrastructure, and photogrammetry.


Although the path forward in the sophistication of drone technology still holds more potential to explore, I aim to venture into the use of drones on a larger scale with a smaller package: Drone Swarms.





My Goal for Programming a Drone Swarm



In the current state for commercial use, drone swarms are particularly used for entertainment through visual lightshows.


             Intel's Drone Swarm Lightshow Package Service                                          Chinese-based Drone Swarm Light Show Company - Ehang



As spectacular these demonstrations of drone organization and coordination are, I believe these capabilities can be expanded beyond than just entertainment purposes and used for integral information observation and surveillance.


My goal is to program my own set of a drone swarm equipped with sensors for more practical uses in the fields described above, essentially multiplying the scale of a task (irrigation or photogrammetry based) they can complete.

I plan on starting small scale with a only a handful of drones being able to coordinate motion profiling objectives. After incorporating more drones into my swarm, I plan on exploring photogrammetric capabilities in rendering image profiles on various geographical layouts.


Task Coordination for Drones and their Surveillance Capabilities


Fig 3


Drone swarms primarily operate on a hierarchical form of handling information and algorithms commanded by the operator. The swarm is organized to execute commands from a server through a drone "leader", which communicates with a ground-station or server. These "leader" drones directly communicate with other nearby drones and allows the coordination of tasks in this manner of hierarchical prioritization.  


The use of a drone in photogrammetry or LIDAR based rendering for physical urban layouts has been a commercial product for drones in the last decade. These two forms of geometric rendering offers unique pros and cons in using either formto produce images  based on aerial image-gathering. LIDAR can often be an advantage over photogrammetric based scanning as it not only provides 2-D renditions of physical geometry but also develops 3-D models of more complicated structures and terrain.

The problem with incorporating LIDAR with drone swarms is the equipment of the LIDAR sensors. Due to their cumbersome size and steep cost, LIDAR sensor equipment is best suited for lone drones outfitted with heavy-lifting capabilities. Thus, the use of a drone swarm can be best incorporated with photogrammetry, which uses multiple images at various vantage points to overlap images and create high resolution renders. Multiple cameras can be outfitted with multiple drones within a swarm and potentially minimize the time it takes for image capturing an entire area. 


***DID YOU KNOW THAT...****

Edited (9/21/2020)

Apparently bite-size LIDAR sensors exist so I might consider switching from photogrammetry to LIDAR. Problem is that is they will eat have eaten my wallet.


Design and Flow Diagram



This flowchart illustrates a "leader" drone communicating with the main server.

All properties and information the "leader" drones sends is processed to return instructions to avoid collision, maintain swarm constellation, and render images.


Hardware

Motors - Main electrical components responsible for the thrust and gyroscopic positioning of the drone

ESC - Electric speed controllers that control the speed of each individual motor

Gyroscope - Calculates and stores data of the drone's current tilt, yaw, and roll

GPS - Provides drone's position to be used for location tracking

PDB - Power distribution board provides appropriate amount of power to each electrical component

Accelerometer - Provides drone's acceleration and velocity

Camera - Allows for image capture at various vantage points of the drone


Software

Main Program - Processes data from the drone's sensors to process trajectory, positioning, and aerial photography of the drone

Main Server - Responsible for receiving processed imagery and data from drone. Returns commands to survey and coordinate tasks



Parts 

Hardware

4x Motors - Eachine FPV Drone                                                       

1x Electronic Speed Controllers - Eachine FPV Drone                      

1x Frame (3-D Printed or Purchased) - Eachine FPV Drone              

4x Propellers - Eachine FPV Drone

1x Power Distribution Board - May not need as ESC has one built in


RGB LED Strip (Unknown Length) - If I have time to make a lightshow as well (Bonus points?)

Battery

1x GPS Module - IMU MPU 6000 

1x Accelerometer - ^

1x Gyroscope - ^

1x Camera - TBD

1x Bluetooth Dongle


(IF BLUETOOTH/WIFI FAILS)

1x Radio Reciever

1x Transmitter


Bill of Materials listed below for the construction of two Autonomous Drones




Since nano-drones with autonomous/teleop swarming capabilities are expensive (ie Crazyflie: $200), I'm gonna make my own with an FPV drone (minus the FPV part). 




Possible Challenges that I Could Face


Interference


If I were to use WiFi or Bluetooth on a large scale involving communications and signal processing between drones, there is a likelihood that ground testing my project near residential areas could be problematic. Runtime and coordination can be lagged due to nearby WiFi sources or Bluetooth operating devices. As a result, initial tests involving more drones would be likely tested in areas with minimal internet infrastructure nearby to avoid interference.



Learning Curve

I had discussed this project with my sister, who has had experience with programming drones in the engineering field. Before I can realize my goal of achieving a dense (10-20) swarm of drones operating tasks through hierarchical processing, I must be start small with preferably two drones as per her suggestion. This way the learning curve can be more gradual and feasible to master, allowing for a quicker transition from small scale to larger scale swarms.



References

Drone Light Shows Powered by Intel. (2020). Retrieved 19 September 2020, from https://www.intel.com/content/www/us/en/technology-innovation/aerial-technology-light-show.html

EHang | Aerial Media - UAV Formation Light Show. (2020). Retrieved 19 September 2020, from https://www.ehang.com/formation/

Drone photogrammetry vs. LIDAR: what sensor to choose for a given application | Wingtra. (2020). Retrieved 19 September 2020, from https://wingtra.com/drone-photogrammetry-vs-lidar/

Tahir, A., Böling, J., Haghbayan, M., Toivonen, H., & Plosila, J. (2019). Swarms of Unmanned Aerial Vehicles — A Survey. Journal Of Industrial Information Integration, 16, 100106. doi: 10.1016/j.jii.2019.100106











Attachments:

image2020-9-18_19-4-1.png (image/png)
image2020-9-18_19-8-5.png (image/png)
image2020-9-18_19-44-37.png (image/png)
image2020-9-18_20-39-7.png (image/png)
image2020-9-21_8-39-4.png (image/png)
image2020-9-24_20-21-35.png (image/png)
360.jpg (image/jpeg)
image2020-9-24_20-30-0.png (image/png)
Robbie_Rotten.gif (image/gif)

Comments:

This is an extremely, extremely difficult project. I'm not sure if the sheer complexity of this project is clear in your proposal.

First of all, having these drones maintain high-precision 6-dof positioning/tracking is an enormously complicated task. It is not clear at all from your proposal how you intend to do this. You list an IMU (gyro/accel) along with a "Camera TBD", but with no explanation of how this will be used to locate the drones with sufficient accuracy. SLAM/VIO/reconstruction/etc is a very very very difficult task, and ranges from "nontrivial" to "extraordinarily difficult" depending on the platform, algorithm, and other factors. I might say "impossible" given the resources and time.

I hate to be a downer but this project really does not seem doable, especially given the lack of background research for how to position drones using VIO/etc. Message me if you want to continue with something like this, but it really does not seem feasible.
Posted by fns2 at Sep 30, 2020 23:40

After looking into specific autonomous software for drones, I found hardware/firmware that is compatible with Mission Planner, which is a bridge from Ardupilot. I simply didn't revamp the reason for choosing the IMU's in my proposal. Mission Planner offers GPS/Way-point navigation through Google Maps with optimizable control. I believe there is also opensource tuning for controlling drone's path within Mission Planner but I will have to look deeper into it. 
Posted by kanishk3 at Oct 01, 2020 09:35