Summer Robotics Internship
Noa/h Milivojevic, Robotics actuation r&D Intern
4 minute read
When KMI began posting positions for our summer 2024 internship positions, I requested a position for prototyping robotics systems. As I interviewed applicants for the internship, Noa/h stood out due to the passion he showed for all aspects of the design cycle, matching a lot of the skillset we were looking for, and Noa/h certainly went above and beyond our expectations. As his KMI mentor, I had the privilege to witness Noa/h's self driven motivation and his genuine curiosity towards how things work, often resulting in a fresh set of eyes on older designs that was invaluable to the engineering team. Noa/h also showed an excellent capacity for adaptability, handling all the curve balls and picking up new tasks from his own initiative after completing his tasks in record time. The following paragraphs are Noa/h’s own words describing his summer at KMI and the excellent work he accomplished while here. - Preksha Sanjay Madhva, Robotics Engineer.
The plan – Individual Actuation;
So what even is individual actuation and why is it important? [note: I’ll assume the reader is relatively familiar with the general KMI conops and REACCH, and if not, you can learn more here.] Long story short: the REACCH mechanism KMI had mainly been looking at up until this point was an underactuated mechanism: a mechanism that can have many different output states for any one input state. Your first reaction might be to think that this is a step backward – why would you want a mechanism that you have less control over? In reality, underactuated mechanisms can actually exhibit really complex behaviors and even react to the environment on their own. This helps a lot in simplifying control software and in some cases making a mechanism even more predictable by eliminating possibilities for software bugs. It can even reduce the mechanical complexity of a system by removing the need for more motors, or other electrical components, and sometimes has a built-in “soft touch” that becomes important when handling delicate objects.
However, sometimes you do still need that sort of granular control over every aspect of your system. That’s where individual actuation comes in, and that’s what I was brought on to do: explore the options for individually actuating each joint in the REACCH mechanism, compare that with the existing underactuated methods, and gather more data on what the KMI team could explore further in the future. Sounds like a pretty solid plan, right?
How things went;
Like all plans, reality didn’t match what was planned, but like a well-thought-out plan, reality included more than just the original plan. I’ll start with what was planned: the individually actuated tentacle exists! It’s a fully contained system, all you need to do is plug it into a power and serial port (usually just the USB port on your laptop) to send actuation commands over to the tentacles, and the tentacles will do all of their movement things. The arm has some basic sensing capabilities, so it can tell when and how it’s holding on to a target. I’m honestly coming up a bit short on things to say about it since the final product does a pretty good job of meeting all the requirements set out at the beginning of the summer. It’s honestly a lot of fun to play around with, especially doing a little tug-of-war with the responsive touch feedback and being able to adapt to changing conditions.
Things that didn’t go according to plan actually ended up all being positive;
I got to spend a bunch of time working on the ISS REACCH hardware as well as attending the International Space Station Research & Development Conference (ISSRDC) with KMI. Helping out with ISS REACCH ranged anywhere from just checking over some ISS safety requirements to ensuring compliance or helping out with wiring and finish electronics, all the way to getting to design and fabricate some of the interchangeable capture cube face plates to simulate different types of materials. ISSRDC was nice because it allowed me to come full circle, helping present a similar poster to the one that initially got me interested in KMI at ISSRDC last year, on top of just getting to see what everyone else has been up to in space.
Other things that I learned
One of the things I was looking forward to most, other than getting to build space robots (though honestly, just the space robots alone would’ve been cool enough already), was learning to work in an actual company, rather than the student teams I was more accustomed to throughout high school and college. Student teams and companies tend to have very different pressures on them: student teams get free labor, so are incentivized to take on as many team members as possible before going under, whereas companies with paid employees have to think a lot more about who and how many people they have on their team. In practice, this means going from six people all working on the same component at the same time to one person working on six components at the same time. One of the bigger adjustments I had to get used to was not getting the constant feedback of five other people looking over my shoulder as I designed something, which was definitely a relief at times, but it meant I needed to get used to packaging up when and how I got feedback and then also asking for feedback in the first place.
Closeout;
I would definitely do this again. Working with the KMI team was a lot of fun and I still haven’t gotten over the fact that I got to work on any one of the projects I worked on this summer, let alone working on all of them (and the weirdest part, people even wanting to pay me to do any of that in the first place). Unironically, my only regrets are not getting to spend more time working on these projects and with this amazing team of people.
Recommended to read next: Summer Internship With KMI