Are we seeing the first steps in which robots take control of their own evolution?
There is no cooked meat in nature. I mean, sure, an animal might get struck by lightning and get instantly cooked. Or maybe a forest fire makes a BBQ out of some small woodland creature but overall, cooked meat is almost impossible to find in nature. But why am I talking about this on a blog dedicated to robotics? Well, it’s because we humans managed to find a way to cook meat a very long time ago. In fact, cooking our food is considered one of the most important steps in human evolution, a step that allowed us to get more nutrition out of our food and expend less energy while digesting it. It also opened up the door for us to start eating foods that we could not digest before, like potatoes, beans, eggs, and much more.
Anyway, I think what I’m trying to say is that we made our own fuel. We didn’t just scrape by with the energy source nature provided for us; we enhanced it, we manipulated it. You could even say we hacked it. And that made us the dominating specie on planet earth. And now… we are doing it again.
The Agricultural revolution changed the course of the human race, then the Industrial revolution did it again, and now we are deep into the digital revolution. In fact, we are so deep into the digital revolution that we take for granted some pretty unbelievable facts - like the fact that we made sand do math 😲 we literally took silicon out of the ground and combined it with electricity to produce calculating machines. And in just under 70 years, these machines evolved from merely crunching numbers to the backbone of our entire society. What started as a glorified abacus is now responsible for the way we communicate, our entire banking system, the way we navigate, and basically how we do everything else. Computers have become much more than just tools - they operate autonomously, make split-second decisions and do anything from managing our traffic lights to keeping our planes in the air to even making actual works of art!
In the past decade alone, artificial intelligence has made such great leaps forward that it can now make meaningful conversations, it can look around and make sense of the world around it, and thanks to advancements in robotics, it can now navigate autonomously and manipulate objects in 3d space with amazing accuracy and dexterity.
I don’t know about you, but over here at Unlimited Robotics, we believe in the robotic revolution. Actually, we believe it will have as much of an impact on humanity as the Agricultural or Industrial revolution had, if not a greater one. We believe that robots hold the key to jumpstarting the human race into an era of peace and plenty. Where all our needs are met, and our society becomes free of dangerous, dull, and repetitive work. The robots are here, and they are ready to serve us.
However, while advances in artificial intelligence, machine learning, and robotics have been fast and impressive, battery technology is still dragging behind. Modern electronics, such as electric cars and the newest smartphones, all rely on batteries whose chemistries are still largely created manually through trial and error.
A recent study, however, demonstrated that artificial intelligence can direct robots to quickly find novel, cutting-edge battery formulations. Scientists have to test out a wide range of potential components when using conventional methods to develop new batteries, which can take years. The need to accomplish multiple conflicting objectives, such as longer life, more capacity, quicker charging, and improved safety complicates this.
In the new study, scientists aimed to speed up the development of batteries by combining a robotics platform called Clio with an AI called Dragonfly to automatically determine the ideal combination of battery components. In the new study, the system experimented on its own with the solvents ethylene carbonate, ethyl-methyl carbonate, and dimethyl carbonate, as well as the lithium hexafluorophosphate salt.
The robotic system used pumps to inject different mixtures of solvents into pouches that contained a graphite anode and a lithium nickel manganese cobalt oxide cathode. According to the researchers, "the system decided what to do; there wasn't a person telling the system what to do."
The system independently identified six electrolytes in 42 experiments spread over two working days that allow for faster charging than a typical electrolyte composition. This method came up with the new chemistry six times faster than it probably would have through a random search.
The researchers point out that their system probably performs more experimental measurements each day than the typical human operator and uses roughly 30% as much lab equipment. They claim that in the future, their system might end up being 20 to 1,000 times more effective than hiring people to do this work.
So, much in the same way that humans “Hacked” nature to get better nutrition from the same ingredients, robots might also do the same to design better batteries. This might be one of the first steps in which robots will end up designing better versions of themselves and, in essence, taking control of their own evolution process, much like we did all those years ago.
