In software engineering startups, you generally go to a room with a computer or bring your own laptop, and write code. I'm interested in how robotics startups work: Is there a separate location for designing the robots? Take for example, Anki. Do they have separate research labs for designing robots? How does a robot get from a single design to being manufactured?

I couldn't find a better place on SE to post this (the startups business section is defunct): Please link me to another SE site if there is a better place to ask this question.

  • 1
    $\begingroup$ To be honest, probably off-topic for here, but I couldn't think of anywhere better... so it stays :) Welcome $\endgroup$
    – Andrew
    Jan 6 '14 at 20:07
  • $\begingroup$ An alternate way to read this is "what are best practices for testing robots"? Startups are one place where this happens. This seems like a good question for this site. $\endgroup$
    – Ian
    Aug 20 '15 at 15:30

I work at a robotics startup. To sum up my long answer below, robotics startups are slower, more expensive, and often more distributed in location than their more famous software counterparts.

Robotics or hardware startups do have a lab for design. It holds owned or rented equipment: 3D printers, machine shop lathes and drills, soldering stations and power supplies, and ONLY 1 really nice oscilloscope that everyone has to share. All the engineers will have a laptop- they may often share licenses for the expensive software like MATLAB and SolidWorks. They work outside the lab whenever possible, because the lab has no windows. Sometimes they make work from home.

The main Engineering work of a robotics startup is to go from a prototype to product. Signs of a prototype that needs to be commercialized include pieces made from

  • Arduino boards
  • RC Hobbyist servos and batteries
  • cardboard
  • duct-tape
  • 80/20 framework
  • Legos
  • 3D printed parts
  • Hand machined metal parts

These parts, for reasons of cost and scaleability and reliability, must be replaced with custom molded plastic, metal, composites, cheaper PCBs, standardized fasteners, and consumer grade batteries and power supply electronics. This process is months to years long and uses at least 3 engineering disciplines. It's very expensive.

At the same time, the Sales and Marketing and Finance employees (and also Engineering employees if the company is any good) are working to move from an idea to a business. This selling and fundraising is where the money to do the Engineering work comes from. You'll notice that Anki site you linked to is billing the product as a toy, not as a military transport robot. This seems obvious, but it has significant engineering implications for what your robot will be expected to do, how long it will last, and how much you can sell it for.

All these people are busily burning up money, sometimes to the tune of millions of dollars. To fight some of this battle, most robotics companies outsource A LOT to consultants and contractors: a consultant might build your website, design your business cards, redesign your plastic shell to be injection molded instead of 3D printed, test your power supplies for safety, and write your user manual. You should NOT outsource complaints handling and tech support initially, because if you do, you'll lose a lot of valuable information about your first products.

As a startup, once you are actually ready to produce a product (Double check- if there are still pieces of your robot made from the bullet list above, then you aren't ready), you will probably hire a Contract Manufacturer to make your first robots because you can't afford all the equipment or employees to make them yourself. You will pay them to follow your instructions to make your robot and test units before they are sold. You will discover, over the course of several months and dozens of angry user phone calls, that your written instructions weren't clear enough and robots have shipped broken. You will fix it.

Your coworkers and consultants and contractors are located all over the country and world, not in your lab. In the case of suppliers and contract manufacturers, someone at your company is paid to visit them and audit their work, and report the results back.

All this costs more time and money than the stereotypical software startup. It rarely includes extra money for Silicon-Valley-style perks such as gyms and gourmet smoothies[1]. The logistics of a robotics startup are close to insanity. The payoff margins are smaller than software, and the costs are higher. About the only good news is that no one can bit-torrent your robot for free, and because it is so hard to build a sustainable robot business, there are fewer people willing to fight the battle.

But, I get to say I design robots for a living, and that's worth a LOT.

-Crazy System Engineer apologizes for long answer.

Note 1: The perk of free coffee IS included. In my professional opinion, no one has EVER made engineers pay for coffee.

  • $\begingroup$ Presumably no money has been raised at this point yet? $\endgroup$
    – Guy Sirton
    Jan 1 '14 at 4:34
  • $\begingroup$ Actually, none of the above works at all without raising money first. In my company's case, the technology prototype alone was enough to raise a small initial seed round of funding. A CEO was recruited very early, and a BIG part of his job was (and is) raising money in several subsequent rounds. $\endgroup$ Jan 1 '14 at 20:47
  • 1
    $\begingroup$ From your description it sounds like you're a bit underfunded which is a shame. It's really hard to compare to "software" startups because those range from a small web site to large scale efforts. If you have a 100 person software startup or a 100 person robotics startup you're probably spending most of your money on salaries. Best of luck! $\endgroup$
    – Guy Sirton
    Jan 1 '14 at 22:13
  • $\begingroup$ @Guy Sirton and FrisbeeGirl: Thank you both; your answers are very informative and helped a lot! $\endgroup$ Jan 2 '14 at 1:08

I can share my experience at a startup where we built what was essentially special purpose robots. I'm sure there's a wide spectrum though. A lot depends on the complexity of your robot or machine and the target market for it. Is it something that will be manufactured in 10's of units and sold for millions a dollar a piece or something that will be manufactured in the millions and sold for hundreds of dollars. There are many other factors as well.

  • We had mechanical engineers, electrical engineers, software engineers, physicists, system integrators. When you're at the "garage" stage you may have 0-1 of each but the general ideas still apply.
  • Physically we had some lab spaces where the machines would be built/integrated/tested and around them was the team work area/offices.
  • You start from a concept/high level system design.
  • The system will typically be decomposed into separate parts/assemblies.
  • Typically the design is iterated on. Some assemblies may be designed, built, tested and then changed. The system design can evolve as well.
  • If you walked into our labs there'd always be some piece of the machine being run through cycle testing or some other functional tests. Things clicking and humming all over the place ;-)
  • In our startup we didn't make any parts in house. We had some close machine shop contractors manufacture all the parts we needed for development. I know some startups will have a machine shop with a CNC mill/lathe for rapid turnaround of parts.
  • We did do all the assembly and integration in house.
  • Over time the design is finalized and various production drawings, procedures, and BOMs are made for the entire machine. Those include all the manufactured parts, the off-the-shelf parts, things like printed circuit boards etc.
  • Throughout this software is being developed side by side with everything else and needs to change to match the changing system and requirements (that's my job :-) )
  • There are various other business activities ongoing as well. Talking to customers, HR, finance...

It's not as easy to prototype as it is with pure software. Arguably if you're building a small web site you can bootstrap your business with no investment. If you're building a robotic company it's probably not possible.

I hope this gives some idea of my experience. Feel free to ask more questions.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.