Bot Brains Inside

In honor of my lunch this afternoon, I wanted to share a snapshot of the sushi chef (see video below). While these parlor trick robots do make tasty rice patties, they remind us of how early we are in the cycle of robotics. Windows 95 was the moment when personal computing became ubiquitous; in my opinion, it will be the software, not the hardware, that will be the biggest factor in moving autonomy from a carnival side show attraction to mainstream adoption.

This week, Intel made this first step by launching a new compact computer called Euclid based on its Prime Sense vision system to speed the development of robots. Euclid is being marketed as the “eyes” for a future line of humanoid robots. Intel demonstrated the Euclid computer in a robot, which was moving onstage during CEO Brian Krzanich’s keynote at the Intel Developer Forum on Tuesday.

Euclid has a 3D RealSense camera that can serve as the eyes in a robot, capturing images in real-time. It has motion and position sensors that can help the robot move around both indoors and outdoors. An Atom processor provides the computing capabilities to analyze the images and data gathered by the robot (i.e., its brain).

intel euclid intel developer forum idf august 2016

Euclid has wireless connectivity capabilities for communication and tracking. It runs the Ubuntu OS and the ROS (Robot Operating System) set of robot development tools and libraries, which are used to develop many robots today.

Intel’s Euclid is more than just a fancy camera, it is the evolution of decades of development around the most critical piece of the robotic equation – the software or Robot Operating System (ROS). The ROS is essentially a framework of programming tools used to write and develop robot software. It works as an open-source system providing OS-like services designed specifically for robotics —i.e., hardware abstraction, device control, implementation of common functionalities, and data package management.

ROS was originally developed by the Stanford Artificial Intelligence Laboratory in 2007, to support the lab’s projects (specifically the STAIR Robot). From 2008 to 2013, development on ROS primarily took place at the famed research lab Willow Garage. After Willow’s closure, ROS was transformed into an open source ecosystem. Since then, ROS development has skyrocketed.


ROS had the biggest leap in development when it was housed within Scott Hassan’s famous R&D shop, Willow Garage. Hassan is one of the biggest visionaries in the robotics space, and an extremely successful seed investor (his most recent pick was Magic Leap). Hassan started his career working with his college buddies, Larry Page and Sergey Brin programming Google’s original search engine. He was one of the first investors in Google, when he decided to invest $800 in the company 12 days after it was formed it 1998. He then founded his own company: An email-list service called that Yahoo bought for about $432 million in 2000. 

Thanks to that success and his early Google stake, Hassan amassed the kind of money that eventually allowed him to buy office space in Menlo Park before he even knew exactly what he wanted to do with it. Its address — 68 Willow Road — ultimately inspired his new company’s name. Hassan then convinced his friend Steve Cousins, who had hired Hassan as an undergraduate intern at Washington University years earlier, to become Willow’s first CEO.

Hassan planned to dedicate enough funding to the startup to keep about 60 people working there per year, so they began hiring all the top roboticists and researchers they could corral. Initially, the collective focused on personal assistants, driverless boats, and autonomous cars, but eventually focused on building programmable bots.

“We had this vision that came together through the people who were there,” Cousins says. “Everybody brought something to the table.”

Hassan had long been a big believer in open source, so the team started focusing on a common operating system that roboticists everywhere could use to stop wasting time “reinventing the wheel” on every research project. In tandem with the Robotics Operating System (ROS), Willow was building robots to run it on.

In 2011, once ROS had officially infected the robotics community and the initial robot units it built had been distributed, Willow started to focus on other projects and spin-offs. The company started exploring market opportunities for autonomous bots and educating the team about entrepreneurship. Google acquired three of the for-profits as part of its ambitious robotics buying spree.

One of the ideas in particular fascinated Hassan. The team had hacked together an iPad-on-wheels to allow a remote employee to zoom around the office, which it refined into a robot called “Beam.” The possibilities of telepresence (think videoconferencing with robots) inspired Hassan and he spun-off a team called Suitable Technologies in 2011, taking a bunch of WG employees with him. 


That meant that from that point on, Hassan was funding two companies. Willow Garage was burning about $20 million a year. He had realized that getting autonomous personal robots into people’s homes was still a long way off (while ROS could support countless functions, the cost of creating the hardware to handle something even as seemingly simply as picking up an article of clothing was magnitudes too expensive). He finally decided to pull out his investment in 2013 and focus all 0f his resources on the more near-term telepresence market with Suitable. 

“I saw people starting to get restless, so I just decided to shut things down and go, ‘You’re free!'” Hassan says. “And what happened next was, everyone started companies.”

Almost everyone who stayed at Willow until the shutdown ended up either starting their own robotics company or joining one founded by fellow coworkers. Cousins, for example, now runs a company called Savioke, which makes robots that can deliver items to hotel guests. 

“Willow went down but now we’re all like the Phoenix,” says Mirza Shah who worked at WG during its last year and a half, and is the cofounder and CTO of a robotics company called Simbe. “We’re out there building new things, stronger than before.”

All told, Hassan says he poured more than $80 million into Willow Garage. But he considers the cost well worth the reward of robotics progress. And, as always, his sights are set on the future. 

Hassan’s investment is paying major dividends; last year alone, more than $150 million of venture capital funding went to businesses that use Willow’s ROS. 

“A large portion of current roboticists in the world have one degree of separation with Willow Garage,” says Maya Cakmak, another former employee. “I think in the future, when we look back, Willow Garage will be what Bell Labs or Xerox Parc was for personal computers, for robotics.”

This surge has opened up the ROS ecosystem in a way that puts it on the path to becoming the Linux of robotic software. About 9 million ROS packages were downloaded last year, and it found its way into systems ranging from the DARPA Robotics Challenge (where 18 teams used ROS), to NASA’s Robonaut. ROS may soon find itself to be a household name the same way Google’s Android has for smart phones, tablets and IoT.

Brains, even robotics ones, have to count for something until then – here’s another stupid bot trick:


3 thoughts on “Bot Brains Inside

  1. Oliver, you are always on track with this rapid changing market. Your observations are spot on. This is the beginning and its the software and intelligence that are initially programmed, then extracted from experience that will leap-frog the current capabilities in Robotics. Good article. Thanks.

  2. Serious issue, here: Can the sushi robot identify and remove parasites and their eggs and larvae from any raw fish and meats (sashimi) being converted into sushi? Sushi-chefs (human) are trained to search for, identify and remove such critters and are equipped with critter removal tools. Are the robots so equipped?

    Who is responsible for insuring the safety of such food-stuffs?

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: