Spent half a day today at legendary research hub SRI International, for Xconomy-organized robotics forum (see full agenda here), listening to an impressive lineup of industry pioneers of mass-market appealing robotics talk about their businesses. Some speakers were still physically on stage, others embodied inside telepresence robots, of course. And answering to a recurring moderator question if robots will take away human jobs with a recurring “no”.
As Steve Jurvetson (yes, we keep having these sweet Estonian reunions) put it well in the final venture capital panel: it would be absurd to think that “we should pull Excel out of organisations, because we would create more jobs when people tabulated numbers manually again.” The times they are a-changin’, and for sure not back towards a robot-less world.
See brief notes from all the sessions (and a bunch of videos of cool commercially available robots in action!) below the fold.
- Discussion about automation outpacing the human employee adaptation & skill set upgrades (and thus the unemployment issues being structural, not temporary) is moving from fringes to mainstream economics
- Parallel to PC industry: PCs first penetrated the back offices of various industries (bookkeeping,etc) but quickly spread from there to industry-specific front line applciations.
- Low cost manufacturing has been moving around for last 60 years. Mistake to think of it as stable, we are running out of places to move as the quality of life (and cost) goes up in previous place.
- Since 1961 industrial robots have been rather mindless, following pre-set trajectories precisely. 70% of them in body shops of auto industry today – dangerous for humans to be around. They are like mainframes, come with no useful software, ordinary people can’t interact with them.
- “looks” (with virtual eyes) towards its next activity to make it more predictable for humans around it
- no universal hands yet, you can switch “fingers” – and many customers are 3D-printing ones they need for specific tasks
- As the blue collar work force grows older, they will go from manual, repetitive tasks to becoming robot trainers. There are no young new workers really flowing into these jobs anyway. Percentage of people under retirement age is 77-82% in US/EU/CN, will drop to ~65% in 30 years.
- 3D printers of today still lacking: melding plastic & metal, embedding electronics, higher than linear speed, usability for tooling, better CAD.
Panel on “Will robots take human jobs?”
- We will solve a lot of humankind’s large problems this decade: clean energy, water purification – and you put those two together: food. However, those breakthroughs still need us to solve for changing needs in legal, ethics, different workforce…
- A lot of nuances with culture and labels (Hollywood, cliché of robots taking jobs): people’s reactions are diametrically different before using “a robot” or “an assistive machine” in their plant.
- Agriculture parallel – employment going from 70 -> 2% of population. Was John Deere socially irresponsible to build tractors? Counterargument: people had 200 years, many generations to adapt to better tooling in agriculture – modern robotics improves in months and mere years.
- “I can see the robots going on strike – stop the 3D printers, they are taking our jobs away!”
- Driving and flying robots (Google cars and increasingly weight-carrying drones) about to disrupt logistics massively. On demand delivery of a cup of Starbucks coffee, a FedEx package or drugs in sub-Saharan Africa without a single human job.
- When building a robot startup, don’t focus on ROI that comes from just replacing the cost of people. What can you do better, faster, gathering new data…
- Builds autonomous robot delivery systems for hospital use (get food, drugs, linen and other supplies automatically to patients) and real-time tracking software to control them. Can use doors, elevators, biometrical authenticated secure compartments.
- 400+ robots in 130 hospitals. Average hospital does 3-5k deliveries/day. TUG robots 11 million cumulative deliveries to date.
Panel on Telepresence
- Double: For remote workers, having robotic presence in the office takes out the friction of scheduling time blocks – you just “walk around”, join a water cooler conversation, attend a birthday party… feel more as part of the team. After a day of novelty wears off, the robot just becomes that person for colleagues – why it has come out is important to have one device per person.
- Romotive Romo: not that much about of telepresence, but built a robotic iPhone holder ($150) for 12-year-old versions of ourselves. Shipping 1500/month. Some cognitive dissonance for adults that “oh, grandma is running around the floor and now hid under the couch” – but kids don’t find it that weird.
- Revolve’s Kubi: a neck for tablet. Video caller just taps on the remote video feed and the “head” on the other end turns towards that.
- All using the powerful commodity hardware (tablets, phones) that used to be the expensive part of robotics. Just building the missing bits. Enables a lot of (affordable!) disruption in hardware.
- Physical presence in the same room (even if emulated) is even more important in emotional use cases, communicating with kids, etc, rather than just business – for latter Skype works.
- Interesting use cases: sign language translations (can turn to see hands best), a sick kid attending classes at school. A single great receptionist “manning” several low-volume office locations virtually through robots. People visiting more museums around the world. A boss walking around several offices without flying.
- At home you want a robot designed in a way that you want it to play with your kids, not think “it will enslave my kids!”
- Summary article & video at Engadget
- Quit his job at Wired to co-found the company with a 19-year-old highschooler from Tijuana, Mexico.
- Building fully autonomous drones out of open source hardware.
- Automatic takeoff, landing, click on waypoints on the map and define actions for them. Designing this sort of dance with a flying camera is an artistic undertaking.
- “Humans should not operate heavy machinery, drunk or not”
- First step: BlimpDuino, assembled lego-based kits with kids on dining room table 3 years ago. Now Tijuana & San Diego plants created from scratch, 100+ employees.
- Agricultural drones are the biggest business. Lowest regulation, highest economic value in next few years. Crop survey, spraying. Market size in 2016 is 10X than public safety (2nd largest category; and everything else)
- Farming is an information problem: bigger & bigger farms, with fewer & fewer people. We spray pesticide on an entire field because it is… June. We should do it only if centimetre-resolution images show an outbreak of some part of the field.
- If you get a drone right for a 9-year-old named Max, you have it right for farmers. The reverse is not true (safety, ease of use)
Panel on Robots in Unconventional Workplaces
- QBotix builds robotics for Solar farms. Single robot runs on a track, runs around and can control and adjust the direction of 1200 solar panels. Needs to last 25 years. Alternative would be 1200 separate motors on every panel.
- Bot & Dolly builds tools to use robotics in creative arts, starting with film. Have a dozen multi-joint robotic arms, commodity tech + their own software layer. Can create complex and perfectly repeatable shots. People joke being replaced as a cameraman, but if the shot requires a steady hand while moving 4m/s, hard to argue.
- Gradually moving to other creative areas: architecture labs, solving some problems with furniture production. Eventually some of the solutions could return back to “traditional” manufacturing.
- Harvest Automation builds tough, simple, mobile robots for mundane agricultural tasks. Filling the gaps in human workforce, but also doing more than they used to (for example, moving plants around the greenhouse regularly for more effective growth)
- Restoration Robotics builds surgical robots for medical use, especially hair restoration. Precision robotic delivery enables many other things in the future – the tip of the robotic arm can carry hair, a laser, scalpel…
Keynote: Remote Presence Robots in the Clinic
- 9 million home robots sold. Consumer robotics $6.5B in 2017. Roomba now the most sold vacuum cleaner in some markets (Spain)
- iRobot PackBot crucial to shut down Japanese nuclear reactor post accident.
- Robots are still expensive to build, if you want them not just to work for a demo, but handle the stress of a professional using it every single day. Millions if not 10s of millions R&D needed.
- CBS video story on robotic telepresence used in medicine
- RP-VITA: nurse can dispatch a robot by giving voice commands, it will self-navigate to the right person by face detection or RFID detection to perform basic monitoring.
Panel: Who’s Investing in Robotics?
- DFJ is in ReThink Robotics and QBotix, Bain was in Kiva Systems (exited to Amazon)
- Robotics are like generalised AI as a category – would have been great to invest in this over last 20 years, but it feels only now Moore’s Law is getting us to a place where all these opportunities truly emerge as investable
- Robots (for ex: Kiva) are often useful not just as a single entity, but as a system, an infinitely scalable group of robots working in sync that makes them so powerful.
- 4Moms Origami stroller that folds and expands completely at a touch of a button (think Transformers)
- The reasons why you should have kept away from robotics as an investor have been the same as they used to be for cars, rockets, etc – capital intensive, complex to understand, etc. All of that is changing – you can now think of them as software with a little hardware on the side. Much of everything becomes dematerialised.
- If you asked to name 2 most innovative car companies in the world, you probably hear Tesla and Google. Who is the third?
- Human-replacement ROI seems most obvious, but you’ll often find that customers actually have another reason to buy robots (fast scaling, inability to hire…)
- You don’t think that “we should pull Excel out of organisations, because we would create more jobs when people tabulated numbers manually again”