This summer, I interned for ModalAI again, just like last year.

Last year it was fun, but this year turned out to be a lot more fun…

Firstly, because there is a really good coffee store (I wrote a post about it) that I discovered right next to work. But that doesn’t really matter, although it was great.

ModalAI was awesome, again, yes.

This year, I ended up working a lot on the Autonomous Mobile Robot (AMR). It’s meant to be a development / demonstration platform for using the RB5 flight deck on ground vehicle. After developing VOXL, which uses a snapdragon processor, the voxl line switched to using qualcomm’s RB5 processor, which later became Voxl2 - which is out today.

The AMR runs a different firmware than the drones and makes use of chirp sensors which aren’t used on the RB5 (or Voxl2) drones.

The electrical interface is pretty simple, but I won’t go too into detail since I’m not sure if I’m allowed to (better safe than sorry! 🙃)

The RB5 is the main processor / computer on board the AMR. Motor drive is control by talked to an Arduino nano, which serves as the midway interface between the RB5 and a brushed motor esc. A simple script converts Serial communication to a pwm signal that the esc interprets as a movement direction and speed.

Since I helped with the development of the AMR, I got to work on lots of different elements to designing; mechanical, electrical, and software. That was pretty cool, since I got to use a bunch of cool tools and gained a pretty good understanding of how these different elements work together - specifically, in a professional setting.

In this sense, working at ModalAI was especially meaningful since it allowed me to gain a much better idea of how large teams can actually work. This may seem trivial to some, but as someone who primarily works on things by himself, the way that different people - specializing in different things - can work together, was very interesting to me.

In general, I discovered that companies work much slower than I expected, or at least on a much longer timescale. More people means that more time is required to get things working since no one knows exactly the status of a project at one time, but it also means that everyone works slower and more elements of a project are considered. As a result, things are done much more thoroughly than, I think, an individual could do it.

This was very enlightening to me. I discovered the true importance of communication - how else are you supposed to know that you’re not overlapping with someone else?

But I also learned a really interesting conflict when working with other people: when to solve a problem yourself and when to bother someone else with it. On the one hand, solving it yourself keeps the project going, but if you can’t solve it yourself, then you’re holding it up. And consulting someone else will almost always hold it up, even slightly.

Althought I’m not sure how to express it here, I think I’m much better now at determining when to attack things myself and when to consult others. I usually did things myself, but whenever I had any doubt towards a problem after researching it, I’d consult someone else. Given that everyone at ModalAI is really good at something, there was almost always someone who had a better solution than me; and that gave me instant opportunities to learn.

It’s through this problem that I learned to use a variety of tools at ModalAI, like lasercutters and 3d printers. How to focus the laser on the cutter for different plastic thicknesses; how resin prints can be dyed in the middle of their curing (which is seriously cool, by the way). How antennas can be placed for maximum efficiency. When and where to include multiple grounding layers in printed circuit boards.

What was cooler than that, though, was the impromptu discussions I’d have with employees there - people who have worked in technology industries for decades. It’s incredible how enlightening such lighthearted discussions can be. I learned ways that wheels can be positioned to be maximally stable when climbing different terrain (uphill or rocky surfaces, or even stairs!) and how basic networking works. We talked about the infinite ways that we’ve crashed drones, ways that drones have failed, and ways that we can fix these mistakes in the products ModalAI makes. Everyone there is an enthusiast, in some form, of what they are working on.

But another cool part was getting to use ModalAI products, since they’re so cool! I learned to use Voxl and to run docker on it while using its services. And to use the cameras on Voxl, its most identifying part. I also got to learn ROS - yes, finally, yes!

ModalAI was awesome. I loved it, and I hope to go back again, maybe once I’m in college. There is so much to learn, and think that this is partly because the company is so small - there is always something to be done, something to be learned; something I can do, something I can learn.

Somehow, I feel like Google or Amazon or Apple or Facebook can’t compete with that.