ENGR 4530 Controls and Optimization

Dr. Hongbo Zhang

Project Phase 2

26 April 2024

Jack Lockett, Ashleigh Moffett, Christopher Bishop

 Introduction
The aim of this project was twofold: firstly, to delve into the hands-on experience of

constructing and operating a first-person view (FPV) drone, and secondly, to enhance our skills

in tuning a real-world system's PID controller. To lay the groundwork for this endeavor, we had

previously honed their skills in PID tuning and reassembling the Phase 1 drone. Practical

experience was gained through flying the Tello and Phase 1 drone. We also utilized the flight

simulator extensively that was provided in the lab.

The team's enthusiasm for this second project was fueled by a keen interest in PID tuning

and a general passion for drone operations, including construction and flying. Some extra

enthusiasm came from building, tuning, and flying the Project Phase 1 drone. PID tuning, a

significant element of the first phase, involves a control mechanism that employs a feedback

loop to correct errors by comparing a desired setpoint to the actual process value.

A good example of a PID controller is a robotic arm. For example, a robotic arm might

be tasked with picking up a box and moving it to a desired location. In this example there are two

process variables and two setpoints. The first setpoint of the robotic arm would be the location of

the box. The first processed value would be the sensed position of the arm in relation to the box’s

position. Remember the processed value is always being updated while the arm is reaching its

desired position. The robotic arm must repeat the same process when placing the box in the

correct position as well. Compared to the drone the goal is to PID tune it to fly as smoothly as

possible with as much control and precision that a student can generate.

PID can be broken down into three phases. Cruise Control is another good example of a

PID Control Loop that can be used to break down the PID controller. The desired setpoint is the
desired speed and the processed value is the current speed as it changes. The Proportional

Control in this example would be the difference between the current speed and the desired speed.

The Integral Control is the sum of the differences until it reaches the desired speed. Lastly, the

Derivative Control is the rate of change between the sampled differences. Each of the factors

listed are scaled by a gain factor Kp, Ki, and Kd. To get to the desired position these gain factors

should be scaled accordingly. The process of tuning these gain factors is called PID Tuning. If

the factors are too high the output response will exceed the desired set point. Also, if the values

of the factors are too low then the output will not get to the desired set point.

Another example of PID tuning and the rebuilding of something that would be like what

students are doing with the drone is a self-made ball balance. From the complete build from

scratch to major tuning of the PID from whichever sensor a student will go buy. It is a lengthy

and very precise tuning process to get the ball balance to balance the ball as fast as possible just

like the process to tune the drone to fly as controlled as possible.

Ultimately, PIDs use stimuli to determine the response that is needed. It utilizes a control

loop to effectively calculate the most efficient way of getting from the current position or speed

to the desired position or speed. Depending on the difference between the processed value and

the desired setpoint, the PID controller will give a correction in terms of proportion, integral, and

derivative.

Methods

Part 1
 Part 1 of this project required the team to build their own drone to become familiar with

the drone DIY process, learning how to select parts, construct the drone, and then fly it similarly

to the Phase 1 drone. Guidance on which drone could be used was not provided, so which drone

was selected was up to the team's discretion. Some ideas were floated, to include a 5-inch drone,

a whoop drone, and even a fixed wing drone, but ultimately a whoop was decided upon due to

the availability and standardization of parts, cost, and durability/legality issues. From here, the

team needed to go online and shop for parts, and once delivered, the parts needed to be put

together and then tested in flight to ensure that all components of the drone worked together.

Figure 1. Types of remote controlled aircraft usable for the project.

Part 2
The next portion of the project requires the team to tune the PID controller used in the

drone to achieve more stable and rapid responses to user input on the transmitter. This process

required the team to fly the drone for around 30 seconds per trial in horizon mode, ensuring that

the drone underwent ample pitch, roll, and yaw commands to generate sufficient data when

analyzed later. Once the flight was complete, the black box data was uploaded to PID Toolbox,

an open-source software that allowed for the black box data to be visualized as a step response.

Once the plot of the step response was generated, the characteristics of the system were noted,
and dependent upon the undesirable traits of the step response, new PID values could be used to

help the drone fly better. This process was repeated to have 3 trials to have ample data to

demonstrate the improvement of the drone as the tuning values changed.

Part 3
For the confidence of the team members and to relish the end of the semester, all team

members were required to fly the drone. At this point, the team was very confident with their

flying skills, so most of the flights were completed in acro mode, allowing for the team members

to utilize full manual control and pull off unique flights.

Results

Part 1
Building a BETAFPV Meteor 75 drone involves assembling various components and

tuning it for optimal performance. Let’s start with the frame and motors. The frame is built for

durability while maintaining a light weight. First, attach the four brushless motors to the frame.

These motors are known for their high speed and efficiency, which are ideal for a drone of this

size. Make sure the motors are mounted correctly and that each motor’s wires are correctly

positioned to connect to the flight controller. Use the screws provided in the box to ensure a good

attachment.
 Figure 2: Brushless Motors

Figure 3: Frame of Drone

Next, let’s install the brushless flight controller and the VTX Board. This component

serves as the brain of your drone, managing input from the transmitter and output to the motors.

The flight controller has extra features such as an integrated electronic speed controllers (ESCs),

which regulate the power to the motors. Having the flight controller and ESCs in one part saves
weight as well. Carefully solder the motor wires to their respective pads on the flight controller.

Make sure to pay attention to the correct wiring order. Next, mount the Video Transmission

(VTX) Board above the flight controller. The box may or may not have a wiring diagram in it. If

it does not, simply use the internet to find one that works for you.

Figure 4: Brushless Flight Controller

 Figure 5: Video Transmitter (VTX)

Next, install the camera onto the front of the frame. This camera was chosen because it is

light weight and has clear image quality, which is crucial for FPV flying. Connect the camera to

the video transmitter, which will send the video signal to your FPV goggles. Also, make sure to

connect the VTX and camera to the Video Transmission Board above the Flight Controller.

Lastly, attach the propellers to the motors. The drone uses tri-blade propellers that provide a

good balance between thrust and stability. Ensure they are securely fastened and that they rotate

freely without hitting any part of the frame.

 Figure 6: C03 Micro Camera

After assembling all the components, you will need to connect the drone to a computer

for firmware updates and configuration via Betaflight software. This will allow you to tune your

PID settings according to your flying style and conditions. Once configured, your BETAFPV

Meteor 75 is ready for flight. Always perform a pre-flight check to make sure all the connections

are secure, and the drone is responding correctly to your controls.

Part 2
Much like the Phase 1 drone, once it was constructed, the whoop drone was controllable

when it first took flight, although it was visually apparent that there was plenty of overshoot

present when we originally started flying. Stable hovering and smooth landings weren’t very

feasible this way, although these errors can be somewhat overcome with skilled piloting. This

behavior was reflected in our first flight’s data, which showed pitch and roll responses that had

large overshoots and struggled to settle in on the desired input. In order to address these, we

began increasing the D gains for the drone, and it required an awful lot of D in order to get closer
to a better tune. However, several flight trials show that increasing this derivative value helped to

manage the settling time and overshoot compared to the default tuning. In addition, some of the

roll and yaw required additional changes to their integral gain due to a failure to reach an

adequate settling time and some roughness. Admittedly, performing flips may make it more

difficult to ascertain the values response, but the flight performance is certainly more stable after

a few iterations of PID tuning. No significant changes were made to the proportional gain, as the

performance was already good and just needed different D and I values due to the nature of the

step response measured using PID Toolbox. The results of the PID tuning trials performed are

shown below in Table 1 and Figure 2.

Trials Control Gain (P) Control Gain (I) Control Gain (D) Settling Time Overshoot

1 56 50 53 140 ms 35%

2 56 60 59 120 ms 25%

3 56 60 75 100 ms 30%

4 56 60 85 120 ms 23%

Table 1. PID Tuning Results

Figure 2. Roll, pitch, and yaw response for the drone through several flights. Data in the table is

based on pitch.

Part 3
After tuning was complete, each team member has ample time to fly their drone, now with

several hours on the simulator and close to an hour or more on the actual drone. While still far

from experts, the team was able to all fly in acro and was able to pull off neat tricks due to the

confidence that had been developed over the course of the semester. The videos from these

flights are shown below in Table 2.

Flight Team Member YouTube Links

Number Name

1 CJ https://youtu.be/pJbddE5czuY

2 Jack https://youtu.be/kw5qp7MdprA?si=5ZuPvjIxS8l2pcFZ

3 Ashleigh https://youtu.be/8LWYSP0pZ7U?feature=shared

Table 2. Whoop drone flights.

Conclusion
In conclusion, phase two encompassed building our own mini quad drone. Above is the

in depth description of how the team assembled and flew the drone. There were some key boxes

students had to check to complete a successful phase two project. First was the selection of the

drone, next was the assembly, following that was the PID tuning, lastly was demonstrating the

ability to fly the drone with tact and skill in controlled settings. The last three portions of the

project were imperative to developing a deeper understanding of the content and course material

from this semester.

 The building of drones, PID tuning of them, and the flying of them have great impacts on

health, safety, and the welfare of many. This can be exemplified in a situation such as forest

fires. Drones, if able to be flown properly, can save hundreds of lives early on in forest fires by

being able to fly over and map out the surrounding areas on fire and where they are speading

actively. They can also use drones to stop the fires as well as saving pilots and firefighters from

potentially deadly situations by just using the drones to scout it out and stop the fires. Drones can

be used in many other situations to impact many livelihoods for the better.

For the impact of drones on global, cultural, social, economical, and environmental

factors they have numerous. A massive example would be the use of drones in military and a

slight example would be mapping drones used for mapping locations around the world remotely.

In the military around the world drones are being used for warfare, the mapping land and sea,

and other purposes. It really effects innovation and the economy as they are constantly looking

for ways to upgrade the drones to better suit what they are needing them specialized for. This

funds many companies to hire more engineering/rouv specialists to attain their goals.

In phase one the teams main difficulty was rebuilding the drone from when it had been

disheveled and broken in the prior semester. For phase two the team did not have any difficulties

building and PID tuning. Where issues and struggles were encountered was the end where

students had to continuously fly the drones and unfortunately there were struggles on the

durability of the drone. At the end of all the flights the team made on crash unfortunately broke

the camera system so the team is now working on the purchase and rebuild of the drones camera

system. Overall phase two was a massive success for the entirety of the project scope, where

students by far were able to apply all their learning into one project.