Sending Robots to School in Virtual Reality


It is no secret that we are seeing robotics and virtual reality race to mass adoption at break neck speed. What we are witnessing, at a now extraordinary pace of progress, is the mass democratization of virtual reality and robot technology.  In the past, we would celebrate new technology and the new capabilities it offered knowing it was probably only available to the well funded and privileged few. That era is over.  The tools, applications and frameworks necessary to work with cutting edge robotics technology and virtual reality is cheap and available to the masses.  All scientists have equal opportunity.  Democracy has come to breakthrough tech.

With both robotics and virtual reality, the wave that is suddenly upon us has actually been building in the background for some time.  In case you have not noticed, we already have a lot of robots at work in our economy. In fact, the segment of the working population experiencing the most growth in employment is robots.  Globally, 2014 alone saw 29% growth in the demand for robots. They come to us with a skill set and profile we could only dream about previously.  They are dependable, consistent, productive and do exactly what we ask. Simply put, robots are the new talent.

Many people still have an antiquated notion of robots being merely a technological fad that will accomplish nothing more than the cold-hearted theft of desirable jobs from good and hard working people.  This is a flawed paradigm.  It is time we stop thinking of robots as novel technology, and we start viewing them as a meaningful and growing segment of our talent pool.  And as with any other employee, how we train and teach them will be the biggest factor determining our success.

First, some context.  Oxford researchers estimate that 47% of all jobs in the economy will be performed by robots over the next two decades.  The drivers of this phenomenon are many; from a perpetual desire for cheaper labor to the perpetual advance of cheaper and more complex technology, these forces have been building in the background and it feels increasingly like we are on the precipice of taking the next step. Technology has never been cheaper, humans have never been more expensive, and the economy has a voracious appetite for anything that can deliver volume globally, meaning that anything enabling greater economies of scale is in peak demand.

It is a mistake though if we simply view this as the rise of the robot age. These robots are essentially our new employees and the field of robotics represents a fertile and fantastic pool of new potential talent.  The best part?  This talent needs no perks and little supervision.

In fact, this is a talent pool which enters offering its own benefits back to the business. They come cheap, never call in, their retention rates are fantastic, and their output is consistent and predictable.  This consistency and predictably is the long sought yet elusive holy grail of Michael Porter style management concepts that started to gain steam in the 1980’s and have persisted since.  If you are a boss, leader, business owner or large corporation, this is the promised land.  The last 30 years have seen management theorists introduce countless diagrams, models and concepts designed to help managers develop and deploy more consistent, automated and algorithmic processes in the name of quality and consistent output.  This has brought us to the dawn of an era where the desirable employee is no longer human, in fact, it’s easy to think of a robot as essentially one big automated process or algorithm in motion. We are able to atomize the most desirable aspects of our best employees and program them into efficient machinery.

But there is one catch.  As we know, a future workforce is only as good as the education system that outputs it.  This has been true in the human dimension since the dawn of economic time.  And our emerging pool of robot talent is no different, although it might take a bit to get used to the idea.  We need to create systems and infrastructure to educate this emerging and eventual half of our workforce.  Companies across the globe will need to solve for robot education and training.  One can envision a world in the distance where everyone is staffed by robots and their training is a source for even greater efficiency and ultimately, competitive advantage.

To solve this, we need to scale how we teach and train robots.  In other words, we need to send our robots to school.

Luckily, for that we have virtual reality.

Let’s start with some highly complicated scenarios.  When you think about the most unpredictable and challenging environments scientists deal with, space certainly comes to mind.  This week we saw news out of NASA that they are training robots in virtual reality environments.  Several friends of mine who have worked in industrial applications for virtual reality for years remind me that these capabilities have existed for sometime now.  They have seen waves of interest in virtual reality come and go.  What is different now?  The rocket scientists are now using Sony Playstation VR.

Let that sink in.  NASA training robots and humans on how best to work together. In virtual reality.  In space.  On a Playstation.

And the truth is, this will mean that humanoid robots will eventually replace humans on space missions.  At least that’s the stated goal from NASA.  It will also mean we are no longer putting these humans in harms way, it will mean a larger talent pool that can take on more missions, and because of that, it will accelerate progress.

In addition to VR available on a Playstation, a host of tools are rushing to market to assist in widespread robotics development. One clear example is Gazebo.  With a tag line of “Robot Simulation Made Easy”, Gazebo is an application that is free, supported by an energetic community, and offers an environment to test algorithms, design robots, and simulate real world implementations. Gazebo is an exceptional tool.  Their simulations run off everyday desktops and offer sensor heavy immersive and 3D simluations.

Scientists at  the Johns Hopkins Computational Interaction and Robotics Laboratory have begun programming industrial robots with an Oculus Rift.  They are working in full scale virtual environments on technology that will be available to the masses soon.  Their current use cases include a heavy focus on training robots for surroundings and environments that are unsafe for humans.  With a little imagination, we can also see use cases at the nano level.  With modern and widely available virtual reality technology, we will also be able to train robots for environments that are not only unsafe, but also currently impossible for humans to enter.

Regardless of the merits of the use cases though, the moral of the story is clear.  Scientists at one of the most prestigious Universities in the world are training robots with an Oculus Rift.

Smart companies are launching products built on these cheap (mostly free) platforms and giving us a glimpse of the future.  Rethink Robotics is one such company.  Much like the Gazebo platform they were designed on, they produce robots that are intelligent and collaborative.  They describe their robots as being designed for the real world.  The paradox is, they are increasingly being designed and trained in a simulated one. Rethink is delivering large volume and building for a global economy as we see demand for industrial robots soar in places like China.



To date, we have seen robots gravitate towards industrial applications. Companies like Amazon are hosting contests inviting teams to compete and help Amazon tackle meaningful robotics challenges. Most teams are using off the shelf software and require an environment that requires rapid iteration through repeated simulation.  Just the type of thing virtual reality is useful for. A highly accurate 3D environment also allows robots to be prepared for unpredictable settings and for changing environments.  In other words, like any good attempt at education and training, it helps them grow and learn and evolve.

The next wave and true revolution is on deck for the service sector.  Places like McDonalds are rolling out fully automated restaurant concepts and whether they find additional motivation in an ever escalating minimum wage or not, they seem driven to reach scale as quickly as possible. Consider McDonalds a very loud canary in the coal mine. It is hard to approximate how big this gets and how quickly it gets there, but it seems like the McDonalds on the corner being fully automated helps us all get used to the idea.

The use cases start to write themselves. A world in which we each have our own personal drone to handle household tasks is well within reach.  And in that case, the training script is flipped as virtual reality can help educate the human on what it feels like to be close to a drone. This is exactly what researchers at the Nicer Robotics laboratory at the University of Illinois have done with an Oculus Rift.  They see many potential implementations, including robot drones as companions for the elderly not only helping with chores but also providing much needed social engagement. They  are able to use an Oculus Rift to help people experience a virtual simulation of what life with a robot is like to help lower ambiguity, ease fears and ultimately, accelerate adoption.

In every case, virtual reality can be deployed to accelerate learning. Eventually, having access to a virtual imagination might be the most effective way for robots to learn.  Recently we were introduced to Darwin, a “robot toddler” that lives in the lab of a Professor at the University of California.  Darwin was built with a series of algorithms embedded that were inspired by the human brain.  He is programmed to effectively “imagine” what he needs to do, before actually accomplishing the task. The process was designed to mimic the process children follow when they learn new tasks.   Darwin is in the news as the scientists behind it demonstrated how it learned to stand and walk.  Because of the “virtual imagination” Darwin has, it is far advanced over previous attempts at robot learning that relied heavily on a mostly binary trial and error approach.

The technologies on their own are awe inspiring and impressive.  But it is at the intersection where these technologies are massively democratized and able to be combined that is gets truly exciting.  When you combine robotics with virtual reality, and maybe sprinkle in a little nano tech, well, the future we have all been promised and thought was possible comes into view a little more clearly. It is with these tools we can build virtual schools and training grounds for our robots in 3D virtual reality enabled space. We are now functionally able to leverage one breakthrough technology application to train another.

But is isn’t just the functionality.  Its now the availability.  That is what makes now different.  Cheap and sophisticated tools in the hands of brilliant people all over the globe is probably the most dangerous combination of all. Robots are the new talent, virtual reality is the new training platform, and other technologies like nano tech open up entirely new worlds.

In the end, we all stand to benefit from this new era.  And like the eras that came before it, everything relies on our ability to teach and train those who will do the heavy lifting and hard work.  That should feel familiar.  Its just the fact that the new employee learned in an entirely virtual world and can’t ever call in sick that might feel a touch different.



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s