Life as a Small Robotics Manufacturer

A few years ago, I dipped into the world of OEM (original equipment manufacturer) robotics. It was a natural consequence of my entrepreneurial spirit, experience designing with microcontrollers, and the fact that barriers to market entry were getting lower. In this multi-part article series, I’ll give you a look into the mind of the small manufacturer. The article isn’t intended to be an MBA course on how to build a business. You will likely find some useful information in that regard, but you may also find a lot of holes. And, the process of getting into business is often rather unclear and disjointed regardless of the overall level of qualifications had by the entrepreneur. If this is your dream, don’t let that stop you, but be warned and be alert.

Hardware is not as hard as it used to be

Around the turn of the century, word on the street was that “hardware is dead in the USA.” No one was going to be designing or building things in this country anymore. But then came the likes of the open-source Arduino and the maker movement. Traditional silicon companies like Microchip and Xilinx started releasing free or inexpensive versions of their tool chains to compete with open-source projects like Eclipse. PCB CAD became more accessible due in large part to EagleCAD. The doom-predicting pundits found themselves surrounded by hardware entrepreneurs.

Related:Moving from EagleCad to KiCad

Today, pretty much every aspect of the design-to-manufacturing ecosystem has maker and unfunded-startup-oriented options. And with the Internet, global tools are also accessible. What this means is that I could design, layout, fabricate, and order prototype quantities of parts for a very small amount of money. I’m not sure there has ever been a time when it was so easy to start a hardware business.

There is a difference between a product and a business

The Pet Rock was a fun idea (you may need to Wikipedia that one) and it made some money for the inventor, but it was a product, not a sustainable business. Kickstarter is paved with great ideas and products that were not enough to build a business around. What all of this means is that you can’t just have an idea. You must think through to the end game. You need to understand how much profit you will make per unit and how many units you can build.

Engineers love to take swipes at the business folks (often with good reason), but those business people are just trying to make sure there is enough money coming in so that everyone can make a living. A dollar profit per unit when you can build 1,000 units per month is not a business. Nor is $1,000.00 profit per unit if you can only build one per month. Before getting started, you need to understand the profit, the number you can build, and how many you can sell. 

Related:Video: How MOSFETs Work

You must also look beyond your initial idea. You will likely need a product line instead of just a product. Some customers want a complete solution, and some, especially in the hobby world, want to mix and match components based on capabilities. A successful business is an overlap of your passion and skills with what a customer is willing to pay for.

In my case, I started with an idea for a custom robot board but quickly determined that I would need several sets of boards. My first idea was a Microchip PIC microcontroller-based robot brain board. It was great for me, because I had a lot of PIC experience and had programmers and tool chains. However, it didn’t take long for me to realize that I was not my ideal customer and I had little luck getting interest in my PIC boards. The real ideal customer needed something with a lower barrier to entry. Hence, the Arduino platform as the base of my first product line.

Robot platform number one

The Arduino has a broad entry-level ecosystem, so I assumed that it would be easier to generate interest. Since the Arduino is open source, I was able to take the Arduino UNO or Leonardo schematic and just give it a different layout. I started with the Leonardo because it uses an Atmega32U4, which has a built-in USB bus interface. Doing so would save a few parts by not requiring a separate FTD USB chip and accompanying popcorn parts.

Related:Have a Craving for Raspberry Pi?

Dual Arduino compatible robot boards. DUANE BENSON

The basic Arduinos are based on 8-bit microcontrollers without a lot of code space or computing horsepower. They are awesome as learning devices but it’s easy to run out of resources. To mitigate this, I took a multi-processor approach. Each board would be an entire standalone Arduino Leonardo with custom accommodations for its designated function. The sensor board had four spots for ultrasonic distance sensors, with three forward-facing at 45 degree angles from each other and one rear-facing. The sensor board also has a MOSFET to be used for controlling a master power relay, if one is needed.

The motor drive board had the same Leonardo core, a Toshiba dual motor driver and a connector dedicated to an Adafruit WiFi module. The motor board is equipped with the system voltage regulator and power switch. Both boards come with I2C, SPI, and five GPIO, each with their own power and ground pins.

As a set, they fit nicely on a round or octagonal robot platform. It was an easy to build and program dual-Arduino Leonardo compatible, WiFi robot platform.

Platform two, the Raspberry Pi

Robot number two came about with the Raspberry Pi Zero. I took a different philosophy with this set, in that the brains would be an off-the-shelf Raspberry Pi Zero and the various functional boards would stack using the Pi bus. A stack of boards in the Pi Zero form factor allows for a nice vertical two-wheel balancing robot.

Self-balancing Raspberry Pi Zero robotics stack with 3D-printed structure. DUANE BENSON

PC board number one is a blank used as a motor mount. Next up the stack is a custom USB 2.0 hub. I needed it for a WiFi dongle. The later Pi Zero W has WiFi built in, so I can skip the hub board with the newer Pi model. Above that is the Pi Zero and then my LiPoly battery charger. Above the charger, I put the dual motor driver board. The order for most of the boards to this point isn’t critical. However, it does get more important now. The accelerometer board comes next. You want the accelerometer as high up as you can to give you a greater moment from the balance point. The top board is a bus board. It has pins for the Pi bus and a set of ten I2C pins, five being 5 V level and five being 3 V I2C level.

“Bare metal” microcontrollers

PIC-based small balancing robot. DUANE BENSON

I also have a set of PIC microcontroller (MCU) based boards. These were the ones I had the most fun with, but as I mentioned above, had the least commercial success with. I put together a simple bus, which I called the “Q bus” or “quick bus.” It carries signals sufficient to drive two motors, I2C and power and ground. It allowed me to have a variety of MCU boards, motor drive boards, and sensor boards and mix and match.

What’s in a name?

With a robust product line, it was time to pick a company name and get down to business. I’ve been in the business world for more decades than I care to admit, and I’ve had some pretty solid successes. However, my biggest weakness seems to be in naming things. I can’t really say that I ever got the hang of picking good names.

Names are important. They become your identity. A name must be memorable, different than everyone else's, and not be embarrassing to say. Arby’s restaurant calls its horseradish sauce “Horsey Sauce” and as someone older than 12, I would rather pick a different place to eat than ask for Horsey Sauce. You don’t want that. Unfortunately, my naming skills aren’t much better.

One of my first entrepreneurial attempts goes back to the early 1980s. At the time telephone communications were largely based on hard-wired analog electronics. The idea of knowing who was calling you was still in the realm of science fiction. But the transition to digital had been started behind the scenes, and I heard rumor that phone companies were starting to send the calling number downline to arrive at the destination phone prior to the ring signal. After about six months collecting specifications by postal mail, I had what I needed. The signals were being set as a DTMF (dual tine multi frequency) analog signal—the same encoding used by the twelve-button touch tone phone keypad.

I designed and built the circuit and came up with a name for it. Being the nerdly person I was (am?), I came up with “Private Line Tracing Device,” or PLTD. A few years later when the Bell system remnants introduced a commercially viable version of the device, they called in “Caller ID,” which in retrospect is a much friendlier name. The lesson is to know your audience. PLTD probably would have been fine for other nerdy people, but most of the potential customer base would have seriously frowned on anything related to tracing a call.

The first name I came up with for my robotics company was “BioTrip.” I thought it would reflect a vision of the journey the world is taking in transitioning from biological intelligence to robotics. As a venture capitalist pointed out to me, what it evoked was either a plague or a charted camping trip in Costa Rica.

In the end, I went to market with the name “Steel Puppet.” The inspiration came from Isaac Asimov’s robot novel, The Caves of Steel. Nobody publicly laughed at the name, nor did anyone tell me to my face that it was stupid, but they didn’t comment on how great a name it was either. I think I can call it not terrible, but I wouldn’t venture to say more than that.

Stay tuned for more about life as a small robotics manufacturer

Keep an eye out for my next installment. I’ll cover what comes along with the move from prototype to production and what to do when the time comes to actually sell and support a product line.