Weeks 8, 9, and 10

Sorry for the lack of updates, I was preparing for the VEX Worlds Championship, which took up a lot of my time.

I did, however, finish all my testing and analysis. For my last modification, I decided to do something that would completely destroy my quadcopter, because, why the heck not.

I wanted to create a more aerodynamic body for the quad, so I melted through the center of all four of the arms.

This meant that the air would not push down on the body, but instead pass throughout.

Unfortunately, the heat somewhat disfigured the arm and the quad became a lot more unstable, it doesn't event fly up straight anymore. However, I did all my testing, and got the result of 1.198 seconds for the time it took, a trial is down below. I believe the reason it had the opposite effect is because some of the energy was used in the horizontal motion instead of the vertical.

Weeks 5,6, and 7

Well, the last three weeks have been quite hectic. My work in MATLAB began somewhat well, but I don't think I was able to obtain value from my simulation attempts. I tried to study from pre existing simulators and create my own but it did not end up well.

Above are two screenshots from my quadcopter simulator. As you can see, there are many issues ranging from the UI to the actual model, even the math is not quite right. I finally decided that the time I would need to invest in figuring out how to make my simulator effective was not worth it, as I could get more valuable data from real life testing. However, I do believe that simulation could be very helpful in this particular project, and is something that I would choose to pursue further if time permitted.


As seen in my last blog, this week's work involves tipping the propellors. Many airplanes have tipped propellors, however, many drones do not.

The purpose of a tipped wing is quite simple. It increases efficiency. It has a similar effect to what is commonly known as a ducted propellor, which are used in airplane turbines.

Above is a diagram of the airflow around a ducted propellor. A tipped wing follows a very similar principle.

I took the cut propellors from my earlier experiment, and bent them so that a centimeter was pointed up. I then ran 20 trials, one is included below.

The result ended up being 1.152 seconds, which supports my hypothesis.

Weeks 3 and 4

Hello!

Most of weeks 3 and 4 have gone into learning simulations and MATLAB. Although I have nothing to show for these 2 weeks, the skills I have learned will allow me to create valuable data and results I hope to display in the coming months. However, I have devised another hypothesis I hope to test in the next week. Here is a quick teaser for what is to come

Weekly Wrap Up #2

I spent this week transforming my last weeks' results into scientifically significant data. I began by performing a standard trial. I ran 20 trials to create a more accurate data collection. Before I began, I realized that a one-meter height is not enough to reflect changes in the quadcopter's properties, thus I little over doubled the distance. I collected data from when the rotors began spinning to when the base of the quadcopter reached the top of the door.

The data for the standard trials is below. It took the quadcopter 1.029s to reach the top of the doorway unmodified. 

The next experiment was ran using the cut propellers from the last experiment. A trial and the data is also shown below.

The final experiment was run without a hull. The data and a trial are below.

Weekly Wrap Up 1

So the first week has been amazing. I got a lot done and so far my results are looking good. I spent the majority of the first week working out the math so that I could finalize my hypotheses.

On the left, you can see a basic thrust calculation that can be used for any aerodynamic body with a propellor. The calculation defines thrust as a function of thrust coefficients, air density, effective area, radius, and angular velocity of the propeller. Using this calculation, I was able to finalize my hypothesis that thrust, and therefore vertical acceleration, is proportional to the radius of a propellor.

After I had worked out my theory, I wanted to tests to confirm whether or not my hypotheses would hold true in what I call a "mini-experiment." Before dedicating 20 trials and hours of video analysis, I wanted to run 5 small trials and figure out whether the theories even made a significant difference in quadcopter performance. My setup was quite simple. I placed a 1meter mark on a door, and using frame by frame video analysis, I calculated the difference in time once the quadcopter's propellers began spinning to when the base of the drone had crossed the mark. Before I could test any modifications, I had to do a standard trial to have data to compare to. One trial video has been placed below so the setup can be displayed.

After all the videos were taken for the five trials, the 5 trials of data were compiled into a table and statistical analysis was done. Below is the data from the standard experiment. The final calculated value was determined to be 0.749s. This means that the unmodified drone took 749 milliseconds to reach 1 meter in height.

 

After the standard trial was complete, it was time to start modifications. First I wanted to test my radius hypotheses so I cut one centimeter of each side of each propeller. Then I ran the same experiment again. Below is one of the trials.

I ran the same analysis as in the standard trial as well as the same method of statistical analysis. The final result came out to be 0.785 seconds, which proved my hypothesis correct. Since the radius decreased, the drone produced less thrust and therefore accelerated slower.

I concluded the week by working on another hypothesis which is a bit more well known. Obviously, the mass of an aerodynamic body causes a decrease in acceleration. However, I wanted to prove it mathematically and in the real world. Thus, I set out to prove my theory mathematically first.

What the math proves is that while a body is hovering, the thrust produced must equal its weight. While this is obvious, it was quite tough to derive it. I found the thrust equations as a function of mass and power and solved for thrust in order to derive it. However, what this relationship proves it that acceleration is increased when thrust is larger than the weight (and drag, but that was ignored).

To prove this expirementally, I decided the easiest way to prove it would be to detach the hull. The hull is a frame that some quadcopter come with. Essentially, it acts as protection for the propellers in the case of a collision. Below is a trial without the hull.

The data from these 5 trials is below. As you can see, the quadcopter without the hull attachment is significantly faster than one with, by almost 0.1 seconds.

This week was quite successful. Next week, the plan is to transform these results in to more scientifically valid expirements.

Hello!

My name is Mannat Rana, and I am a senior at BASIS Peoria. After being a captain of the robotics club for so long, I have wanted to challenge myself in a more realistic region of engineering. Lately, drones have been getting much attention; they are used in the army for surveillance, used domestically to take pictures and videos of memorable events, and are even being used by Amazon to deliver light packages. Thus, I wanted to make these processes more efficient, and I felt the best way to do that was to optimize the drone's vertical acceleration so they can get to the desired height as quickly as possible.

My project will involve modifying an off-the-shelf quadcopter to be as fast as possible. I will change properties of the quadcopter and see how it affects its flight pattern. These modifications include hulls, rotor blade length, the length of arms, and possible more variables if time permits. Once I have collected my data, I will try to mathematically explain why the observed trend occurred. Hopefully, this project will combine the many years of learning math and building robots into a realistic engineering problem.