I attended the TechCrunch Robotic+AI conference at UC Berkeley last month. I’m in tech, but not in this kind of tech; I went to learn more.
The conference had a number of interesting speakers and exhibitors, and gave me a lot to think about. Here are a few take-aways….
Solution in Search of a Problem
The conference was stuffed full with super-smart people who want to build robots. What many would-be robo-teers don’t have is a clear idea of is what, exactly, the robot should do. I’ll admit I was a bit surprised by this early admission, but when I thought about my experience in application development, I agreed that the need for product owners, designers, artists, and SMEs with strategic vision is essential to the successful project/product. Building robots is no exception.
The verticals that will first benefit from robotics: Big agg (already on the leading edge); manufacturing (of course), healthcare (which could benefit from more technology in general) and the home which, given the complexity and commonality of house and yard work, is the most difficult to solve, but potentially the most lucrative.
The irony is that in a room filled with “hard science” people, what they desperately needed were those fluffy “soft science” majors: artists, sculptors, designers, anthropologists, occupational therapists, science fiction writers — someone to guide the linear thinking robo-teers to other life forms. Liberal arts is art for a reason.
Robots Need to Complete the Task – Not Create New Ones
Colin Angle, CEO of iRobot, had perhaps the best insight into what it takes to bring a robot to market successfully. Colin left college with a burning desire to build robots; however, he did not become successful until he “started selling selling vacuums ….” He cautioned that it’s not enough that a robot be cool, it cannot add additional tasks or complexity if humans are going to buy one. Robot designers must understand that simply automating the existing task is not enough – they must completely re-think how the task can be accomplished – by a machine – a machine that doesn’t usually have legs or hands.
I was a Gen1 Roomba owner. I loved that the Roomba vacuumed my house, but what was the point if I had to charge, clean and find the thing? Not surprisingly, iRobot had this same realization, which drove improvements in subsequent iterations. The Roomba is now self-docking, charging, cleaning – and cats still love it.
I’m not sure how many cats will be riding their latest product, Terra, a robotic lawn mower, which looks to be thoughtfully designed in addition to being super-cool. No lawn in my Mediterranean landscape, but I did buy some iRobot stock.
Another “yeah, duh!” moment for me was Nvidia’s VP of Engineering, Claire Delaunay introducing the Isaac SDK robotic platform. If you think back to the pre-smart phone dark-ages, you’ll recall that it was the release of the common mobile platform(s) that provided global developers the ability to build flashlights into the future. Robotics is in a similar state of evolution. Claire explained that delivering robotic intelligence has been deterred by a lack of unified platforms for software and hardware development.
Nvidia’s Isaac SDK robotic platform wants you to come build, and there were quite a few people at the Nvidia development workshop who intend to do just that.
Robots Are Coming for Your Job
Ahhhhhhh…..Not anytime soon.
Perhaps the best thing about this conference was the reaffirmation of the amazing human body. The ability to walk, balance, process and instantaneously reprocess new information, muscle memory. No robots for this – not even close. At the end of the day, AI is still an algorithm.
Consider the task of sorting tomatoes or picking strawberries. The ability of the human hand to immediately detect the softness, firmness of the fruit, adjust one’s grip, scan swaths of fruit moving on a belt, detect blemishes, defects, size, ripeness. Toss a sub-par piece of fruit aside (and we’re talkin’ two hands here) while having a conversation with a co-worker. The millions of decisions and actions taken by the body and brain (not to mention breathing and keeping your heart beating) is simply miraculous.
Robots are good with hard stuff – no strawberry pickers here – consistent sizes, repetitive motions, they all have wheels, they need smooth surfaces. Legs can navigate irregular surfaces, climb stairs, slide into Warrior 3. Robots can’t rebalance or adjust to change; they must have consistency. If the task has multiple variations, the robot’s effectiveness is limited or terminated.
The ability to replace humans is much less realistic than the ability to augment their movements. This is where prosthetics and other Iron Man devices really capture the imagination.
For me, the most compelling story was presented by Manmeet Maggu CEO from Tréxō Robotics. Manmeet explained that his journey began with the discovery that his nephew had cerebral palsy. While searching for solutions that could help him walk, he was disappointed to find that exoskeletons were designed for adults not growing children.
Manmeet and his friend Rahul (both studied robotics at the University of Waterloo) set out to change that. Their current product is designed to help children with disabilities walk. The technology provides the wearer a repetitive, physiologically correct gait, which enables a child to exercise by walking.
I loved everything about this. Unlike a Battle-Bot, it meets a real market need. It can be adjusted to each individual. The product exists IRL, and can be purchased or leased. Well done.
China, not surprisingly, is leading the way in robotics technology. There were cautions to be protective of IP, and that those who choose to partner with Chinese companies should not skimp on lawyers. VC in robotics is also challenging. The ROI period for a robotics start up is long (10+ years), and their burn rate can be much higher than a software startup due to the complexity of industrial design, prototyping, and integration. This isn’t Angry Birds – you need to be ready for the long haul.
Brevity dictates my take-aways focus on the robotics side of the conference; however, the AI talks were also compelling. Be warned: Fake news will join Deep Fakes making the policing of social media platforms even more challenging. No doubt, we will see these forces invade our 2020 election cycle like a hoard of White Walkers into Winterfell.
Implicit bias, and other discussions of ethics are important topics of the day, and could easily be a whole separate conference.
While the Berkeley campus was a bit inconvenient, it likely kept the cost of the event down, and was a nice change from usual hotels and convention centers.
Looking forward to Disrupt in October.
* * *