A step closer to a fully automated workforce
Society has always been fascinated by robots. Innovative technology has enabled the mass production of smart bots for industrial and consumer markets, making them even more a part of our lives. The new generation of advanced robots already work in various different sectors, performing manual tasks, sifting through data and interacting with humans. Robots have been employed on production lines and in medical institutions for years, but developments in Advanced Robotics have taken their capabilities to a new level. Social robots, or ‘sobots’, are being used in customer services, greeting consumers from behind reception desks and providing room service in some hotels. Mechanical co-workers, or ‘cobots’, are slowly being equipped with AI software, making them more like their social counterparts. Now, robots can teach themselves how to perform certain tasks without explicit programming. The implications of self-learning bots are staggering as it is, but what happens when one robot can teach another? How will this change the field of Advanced Robotics?
Developers have been exploring shared robot knowledge for years. Cornell University and Brown University were among the first institutions to successfully get two robots to communicate. Through Cornell University’s platform RoboBrain, a Baxter robot at Brown University was able to complete the same task as a research robot called PR2 – despite being several hundred miles apart as well as totally different models. Robot communication has now been taken out of research labs and into real-life applications. Fanuc, an industrial robotics manufacturer, has partnered with deep-learning technology company Nvidia to create self-teaching machines that feed their knowledge into a shared neural network. Other robots can access the network, adding more information as well as learning from it. Robot co-operation may well feed continued paranoia that machines will eventually decide to eradicate the human race, but there are some real benefits to consider before writing the concept off as frightening. Imagine, for instance, that an industrial cobot is working the production line when a human colleague becomes trapped in a machine. The cobot realises that something has gone wrong, and using its industrial capabilities in conjunction with shared medical knowledge it is able to stop the machines, identify the severity of the injury and alert the foreman. Similarly, if domestic social robots (like Pepper and Buddy) could fix electrical appliances, that would make them far more useful – and therefore more attractive – as consumer items.
How disruptive are self-teaching robots?
Humans continue to be usurped by robots, even in jobs that were previously thought of as safe from automation. Armed with exponential knowledge, bots will become a very powerful tool. They will be able to adapt quickly to new situations, providing an even better alternative to human workers. Ironically, the rise of omnipotent machines could negatively disrupt robotics. People and organisations will only need to invest in one robot to access all robotic capabilities, which could affect sales. As much as super-knowledgeable robots give power to businesses and consumers, they also gives more power to the robots themselves. It might be a little far-fetched to imagine sneaky robots plotting the demise of humanity over channels like RoboBrain, but it’s not something that should be completely ruled out. Developers need to make sure that they can control the exchanges that go on between machines. Without some form of monitoring, bot-to-bot communication could cause serious problems. If the recent escape of a self-teaching Promobot from a Russian testing ground is anything to go by, perhaps programmers should work out how to cap the flow of knowledge between machines. Platforms also need to be secured against hackers and cybercrime – imagine what havoc a cybercriminal could wreak with that kind of interface.
From a business perspective…
Bot-to-bot communication is an incredibly useful tool, as one machine can simply share its knowledge with another… Suddenly, you don’t have to worry about programming the second machine because it already knows what to do. The more bots know, the less resources will be spent on making them smarter, which will save time, effort and money. It takes days for a programmer to set up a robot, but with AI this could take just a few hours. Initially, robots teaching other robots sounds like a good thing. However, for robotics companies, the situation is more complex. Firms that have made a distinct choice between sobots and cobots will be left disorientated as the market changes – industrial robots, for example, might lose out to more ascetic social bots that have the exact same capabilities (as well as a cute face). If all robots can do the same things, the focus shifts to how well they can do it and how accessible they are. Therefore, the challenge for businesses remains the same – create the best product for the best price.
Robot-to-robot communication clearly has merit, especially in working environments where co-ordination is key. If one robot can teach another, companies won’t need to expend resources on external programmers. All robots will be equally as useful, which will fundamentally change the market by making it more homogeneous. The distinction between social robots and mechanical co-workers will be all the more ambiguous, which points to hugely capable, hybrid robots with a wealth of knowledge to hand. It’s not hard to see why this is slightly worrying, but programmers can address these concerns by setting up the right security systems to monitor and protect robot-to-robot exchanges. Without the right controls in place, machines will essentially have free reign over all of the world’s data. . . and if that doesn’t terrify you, then it should.
Could robot-to-robot communication enhance your business strategy? Are platforms like RoboBrain at risk of cybercrime? Will social robots and mechanical co-workers eventually become the same thing? Share your thoughts and opinions.