This was originally posted on my personal Medium blog @selamjie, but I thought it would be good to share here too for those of you who may still be in high school and interested in robotics. Now that I have graduated, this is the type of post-college wisdom I might share from time to time on Admissions :)
When I was in grade school, I distinctly remember not really knowing what different engineering professions looked like. By high school it became somewhat clearer, but honestly, not all that much. By sophomore year of college, I had to declare my major. I knew I liked robotics and what that meant academically, but even at that time, I didn’t really know what it meant to have a job in robotics.
I find that this is even more so the case for people whose parents had professions that are far removed from STEM. My mother has a small business doing tax and accounting, and she served a bunch of other small businesses in the Chinese community. Some people did similar work to my mom — insurance, real estate, professional services that people who didn’t always speak the best English could rely on. Others had restaurants, liquor stores, massage parlors.
All these small business owners had kids, and as you might guess, they often told them they should be engineers, among other professions. But I’m not sure many of us really knew what being an engineer meant.
So that’s why I decided to write this post. I’m hoping it will become a series — “What a Career Looks Like” — where I can interview people in other fields, too, but we’ll see. For now, I’ll just start with my own profession, robotics.
What is Robotics?
‘Robot’ is, when you think about it, a very abstract term. I’ve always found it funny that the Chinese translation for “robot” is “机器人” or “electric person”, because not all robots are humanoid.
A robot is just an automatic machine. A 3D printer could be thought of as a robot. A regular printer could even be thought of as a robot. Some robots have no bodies — Twitter bots and customer service chat bots, for example. Some robots don’t need electricity.
I guess the reason that robots are first thought of as human, as just “electric people” or “机器人” is because of what people intend to do with them — to replace the labor of a person, but still accomplish a task. Before the printer or printing press was just pen and paper, and everything was written by human hands.
I start with this very broad overview because the modern field of robotics is equally broad. It’s a highly interdisciplinary field with a lot of different applications. I can give you my perspective on what a robotics career has practically meant to me, but don’t be afraid to think outside the box. Your base discipline, that which you are most familiar with or have a Bachelors in, does not necessarily have to be one of the three traditional fields in robotics, which are Electrical Engineering, Computer Science, and Mechanical Engineering — some are in more specialized fields like user interaction design or human-computer interaction design. Some are in animation or even biology. But the three traditional disciplines are the ones that I will focus on in this blog post.
I’ll oversimplify these a bit as a summary:
Mechanical engineering is the study of applied physics mechanics — “how to build things that won’t break”
Electrical engineering is the study of applied physics electromagnetism — “how to use electricity to control, count, and process”
Computer science is applied mathematics and logic — “how to build things that make sense”
Typically, when you go to college and want to study robotics, it is advised that you are familiar with all three of these disciplines, but realistically, most people get most familiar with “two out of three”, although you can even do a lot with just one. Where you get with your “two” might surprise you. The core technology of impressive robots like those from Boston Dynamics, purveyor of viral videos, or the MIT Cheetah, relies very little on machine learning techniques and much more on electro-mechanical interfaces, sensors and motors that can rapidly respond to changes in the physical environment. Those who choose electrical and software can produce powerful sensors that are smart about storing and relaying information, like the wiring of a robot’s “nervous system”.
I tend to be in a strange spot — mechanical and software, which seems like the least likely pair. And this is because it’s also the area people think about the least: how your brain knows to move through the world. Humans take this for granted — we take for granted that we know to grab a cup around the center rather than some physically impossible point on the very edge, that we know how to pick up a tennis ball. This is actually an intuitive understanding of physics that we all have, and that we all learn as children through trial and error, through experiments (discovering gravity by dropping pasta on the floor as an infant, for example). Robots don’t have that understanding, and those at the intersection of mechanics and software try to find ways to establish that physical awareness. This is particularly important as a safety measure — a typical “dumb” robot arm like those used in car manufacturing does not know whether there is a person who could get hit in front of it, because it may not be adept at “feeling” its physical environment.
One of the things I love most about robotics is that it is always changing. And therefore I, too, am always changing. I’m early in my career so I must constantly adjust and refocus and think about what things I’m really interested in and what applications I’m truly passionate about. But this does make it difficult to outline, or even imagine, all the different present and future career paths in robotics. So if you really want to pursue this, keep your ear to the ground. Look at different job postings and descriptions, and if possible, talk to those in the industry when you can (most people I know would be happy to mentor a young roboticist!)
Right now there are a few big industries in robotics that have grown dramatically over the past decade. Some of these are:
- Driverless cars
- E-commerce robots
- Surgical robots
- Safety and security robots
E-commerce robotics is now an enormous industry. E-commerce order fulfillment, when you click “checkout” on your cart at Amazon, Walmart, or even your favorite online clothing store, is a massive logistics operation that requires a lot of different people (or robots) handling a lot of different physical components. There’s stocking, sorting, packing, unpacking, packaging, labeling, scanning. Even after your order has left the store warehouse, the shipping company, too, uses robots in its pipeline to sort packages and route them to the correct place. These operations have traditionally taken place at enormous warehouses, but now, with the help of e-commerce robots, more distributed models are occurring, maybe with more small or medium warehouses and “order fulfillment centers” closer to cities to enable fast transportation. This industry is full of innovation and constantly evolving, sped on by the trend of more and more people shopping online and expecting fast home delivery.
I once worked at an e-commerce startup, and I now work in the safety and security space. Safety and security robots perform tasks that are too dangerous for humans, such as bomb disposal or the “search” part of search-and-rescue in an emergency. FLIR and Clearpath Robotics, for example, fall into this category.
Right now, I’m still learning myself, so I haven’t been very particular about the industry I work in. I’m happy as long as I receive challenging assignments that I can learn new skills from, and I’m glad to say that right now, I have learned something new virtually every day of working since I graduated college.
Technology itself does not necessarily have a particular industry. “Manipulation” is one well-established discipline within robotics — this means picking up and handling objects. Manipulation is used a lot in the e-commerce industry, but the general concept applies to medical and other applications too.
If manipulation deals with robot arms and hands, “locomotion” informs the design of legs. “legs” is a broad term here, including wheels, wings, and rotors. These are the primary physical fields, and these two broad terms include most forms of physical movement.
Other fields include:
- Human-Computer Interaction (designing robots that are human-friendly and easy to work with)
- Natural Language Processing (software focused, involves getting a robot to understand human speech)
- Machine learning (involves getting a robot to understand a pattern)
- Soft robotics (designing robots that are squishy)
(A reminder that all these lists are far from exhaustive)
Here’s an example of what typical day-to-day tasks in robotics looks like for me. This is a general description of all the jobs I’ve had, and is not particular to any specific one (to avoid telling you the proprietary things I was working on ;))
- Internal meetings. In any engineering field, or any field where you have coworkers (more or less all of them) you need to be able to understand and collaborate with each other.
- External meetings. In most of my jobs I have had to have conversations with customers or clients to understand what they need. Sometimes this client is a business, sometimes it is a government agency. It is never an individual consumer; much of robotics right now is business-to-business technology, but things like smart speakers and driverless cars can be consumer technology.
- Research. This includes everything from reading Stack Overflow posts about specific approaches to reading journal articles so I can try to replicate a technique for doing something. It may also include learning about a new software library or looking through catalogues of electrical components or off-the-shelf mechanical parts.
- Coding. I primarily use Python, but I’d like to learn C++.
- Math. I implement mathematical equations as algorithms — this is sort of part of ‘coding’, but there is some separate time spent to think or perform calculations, and sometimes if the algorithm is derived from data I must produce numerical models, separate from the coding part. Only the end result or equation is actually coded into the algorithm.
- Experimentation. Ok, this I *can* tell you about. I once worked on a tick-collection robot (yes, ticks the insects). I had to put some lab-grown, diseases-free ticks in a bin and time how long it took for them to escape, because we couldn’t have them escaping our collection bin out in the field. Let me tell you, those bugs are fast (also no one else on my team wanted to do this). But there are more typical experiments too — when I worked in a manipulation lab we did some tests to see how long the robot’s reaction time was when it tried to reach a blocked position (how long it took to realize there was an obstacle).
- Writing. Typically reports, either internal or external. Documentation is important. At a previous job, I also wrote grants for funding, and I won about $500,000 of funding. If we did a good job on the first parts of the grants, that initial $500k was eligible to turn into almost $4 million over 2 years. My point is, writing is important for engineers :)
- Debugging. Equal parts frustrating and satisfying. I do a lot of this after implementing a new technique I’m not familiar with. You’ve probably done this for your parents, when some device in your house won’t work, and honestly, while professional debugging takes longer and is more complicated, it’s not that different at a high level. It mainly involves resolving why an outcome is different from your expectation, and hopefully, finding the way to fix it so that the outcome is what you want. It’s a frustrating process but super satisfying in the end! (to me, at least)
- Hands-on work. I work on robots with bodies, and the test/development setup usually has at least an arm or other hardware. I also have mostly worked at small companies. This means I do a lot of hands-on work — assembling components, modifying or prototyping them, stripping wires, soldering, all of that. It’s taken me a while to learn this always takes more time than I think it will.
- Lunch. I myself am not a robot. A girl’s gotta eat 😉 But in all seriousness, social time with your team is important. It helps everyone work together more comfortably, efficiently, and of course, it just makes the whole thing more enjoyable when you know and like the other people you work with. I’ve been able to learn from older coworkers or bosses and exchange notes with coworkers at my same level. Which is why my #1 rule is to never say no when asked to lunch at work. It will always be worth the $10–20 — I think of it as a part of my career development.
Robotics Needs You.
Robotics is a diverse industry, and it needs equally diverse people. We’ve seen an explosion in automation recently, so much so that it’s become a political topic and a regular news item. I remember being excited about robots as a kid because I wanted to invent the future, and I loved physical, touch-able things (I was never meant to be purely a software engineer). Now I’m excited about it because I see that this is true — people in robotics have direct influence on the future of how everything is made and transported, on how people interact with their environment, on a potentially huge, structural piece of technology that will change the underlying systems we all rely on. I’m sometimes worried about the field for the same reason.
Robotics can often be seen as scary or intimidating, but it’s not so frightening. The implications of what humans do with robots scare me more than the robots themselves — and that’s also why I believe we need all kinds of people in robotics, to make the “electric people” as representative of the concerns and needs of the real people as much as possible.
When I first graduated from college, I wasn’t sure if I made the right decision. A lot of job postings in robotics seemed to demand master’s or PhD degrees. I finally got a lucky break when a PhD student in the lab I worked in spun off an e-commerce robotics startup.
Startups are startups, so I was laid off when the company faced some financial difficulty. I spent three months unemployed, wondering if robotics was what I really wanted to do. I got one job offer I turned down, with no other options, just because it didn’t feel right. I felt a bit listless. Again, I noticed lots of positions demanded advanced degrees. I had tried my best to make myself a sort of “senior engineer on sale”, someone who took advantage of the resources of my school to make it so I could, even with just a Bachelor’s, be considered a good candidate for jobs in robotics. I started doubting myself a lot, worried that I picked the wrong major, did too much “breadth” and not enough “depth”, was too interdisciplinary to be taken seriously in any one discipline. I had lots of “maybe I should have just been a software engineer” moments, or “I should have gone straight to grad school” moments.
But in the end, I did find places where I could both do exciting work and also have work-life balance — and at the end of my search, there were three job offers I had to choose from. Even with just a bachelors, I was able to do interesting research, like the tick collector project, and I partnered with professors at universities on their research. I have never had a boring job, and I have never hated what I do. While I have sometimes questioned the importance of a particular task or technology, I’m more convinced I’m in the right place now. Regardless of the specific industry or application, all technological development, by anyone, is a contribution to what the future looks like. This hasn’t been an easy path, but I feel more confident that this future needs me.
And it needs you, too.
Now that I’m a Real Adult, I write mostly on my own blog, medium.com/@selamjie. See you there!