The Maiden Voyage

Part 3 of 3

Hey folks! I am excited to present the third and final installment of my journey to build a flying Dumbo. =)

Recall that I left you on a cliffhanger in the last segment with the fully assembled, but yet to be tested Dumbo.

It seems worthy of flight, but there’s really only one way to find out…

Place your bets now.

As you can see, the damn thing barely came off the ground, wobbled like hell, and then crashed. (Admittedly this is better than blowing up, but hardly counts as flying.)

So to figure out what went wrong, like many tinkerers of the 21st Century, I went to the Almighty Google. After about an hour of searching, I determined that one potential cause for the wobble was out of balance propellers. The weight of each blade of the propeller could differ and had the potential to throw off the balance of the drone. I had no idea how to even out the weight of the blades so I did a bit more searching and found this thing called a propeller balancer. (It balances propellers.)

Just in case you are curious, this is a propeller balancer:

Does it fly after balancing the props (“props” is the cool way to say “propellers”)?

Nope — It’s still a wobbly mess. (And yes, I know that the videos are exactly the same. That’s because balancing the propellers didn’t do sh!t to fix the wobble.)

At this point, I was fairly discouraged. I didn’t know how to fix the wobble and I was running out of things to Google. I had a long list of non-drone stuff I needed to do and found an excuse to table the drone indefinitely.

For the next few weeks, whenever people asked about how the drone was coming along, I would answer with some version of: “I’ve been too busy with school and haven’t had a chance to make any progress.” But I knew that answer was bullshit. I was just making excuses because I didn’t know how to fix the damn wobble.

One night I was fed up and decided to finish the drone. I gathered my tools and parts and marched down to the IEEE electronics lab at UCLA. While at the electronics lab, luckily, I was introduced to someone who had previously built a drone. He advised that the PID controller was the culprit of the wobble, not the propellers. Taking his advice, I spent the next 5 hours learning how to configure my transmitter and receiver to allow for in-flight PID tuning.

Eventually, I had everything configured and took Dumbo out for another test flight.

Finally, it worked. After about 5 minutes of flying, I ended up getting the drone stuck in a tree. It was awesome.

(Eventually I got the drone back.)

What I Learned

Throughout this journey, I have learned a lot. Here’s a few of the most important things:

1. Something that seems small and insignificant can easily turn into a major road block if you’re not prepared. (RE 3.5mm bullet connectors)
2. Knowledge is only a Google away.
3. If you don’t make the time, you’ll never have the time.
4. Share what you do with others. It’s hard to give up when you have people waiting for progress reports.

What’s next?

As mentioned previously, this was only the first of many experiments. The end goal is to build a fully autonomous drone which can be sent instructions via a mobile interface (i.e. my iPhone). To successfully build this flying machine, I will need to tap into a myriad of domains: embedded systems engineering, computer vision, artificial intelligence, robotics, mobile networking. However, like any large system, I have to build it up one piece at a time.

In the next experiment, I will be focused on programming a rudimentary flight controller. This means that I must:

1. Read the raw values of all sensors (accelerometer, gyroscope, magnetometer, barometer, and receiver)
2. Combine the raw sensor values into something meaningful
3. Use the “something meaningful” from above to determine the appropriate RPM for each motor

Thanks for sticking around for the Dumbo build — stay tuned for the next experiment.

Until next time, friends.

— Chris