Who Is Open AI?
To start off, if you do not follow the Open AI on any social media, I highly recommend that you do. They post some really cool stuff, and create some amazing developments in the robotics field. Not to mention the famous rubik’s cube solving robot that we are talking about today.
Open AI is a non-profit, San Francisco based company that’s goal is to promote a future with AI that benefits humanity as a whole. On of their imperatives is that they are a research company, not a corporation. This makes it interesting as a frequent concern regarding the advancement of AI is who will own it. Although, they received a $1 billion pledge from Microsoft this year, and $1 billion when they were founded from a multitude of investors.
And if you care to dig into their entire lineup, you will be absolutely blown away. The amount of intellect and ingenuity that is part of the Open AI team is mind-boggling to me. I am the biggest fan of this company, and am very excited to see what they manage to do in the future.
If any of you Open AI folk ever need somebody to act like a star-stunned weirdo in your press, email me.
Why Make This Robot?
I have brought up this robotics development to a few friends and family, and what seems to happen is they think this is just a stupid Rubik’s Cube Solving Robot with no practical purpose. But, this becomes increasingly interesting the deeper that you dig into the underlying signals from this event.
There have actually been robots for decades that perform specific tasks, but the issue was normally somewhere in the software. For example, they could do the same button press in a string, but it was really just programmed to go up and down. Now, they are trying to code software that will provide the robots with dexterity. So, the Rubik’s Cube Solving Robot is an example of Open AI working around the limitations of a button-pressing robot.
How Did They Do It?
We are going to try and describe their process without going into too much complicated lingo. Because, once you dig into what they had to do in order to program this robot for it’s purpose, you start to run into some very abstract ideas.
In essence, what they had to do is create a simulated world where the robot had to solve the Rubik’s Cube in increasingly difficult environments. This process is called Automatic Domain Randomization, and it freed them from having to hand-craft the difficult environments on a subjective level.
The issue that they had was also that you cannot go too easy, or too hard on the robot. Because if that is the case, then the robot cannot learn efficiently. It has to learn the effects of friction, elasticity, surface area of the Rubik’s Cube and many other real-world obstacles. Finally, Open AI has introduced random obstacles into the robot’s environment in the videos to show it’s adaptation ability.
All in all, this is a very interesting development in the robotics world, and it was done by an amazing company. I recommend you head over to their website for some in-depth details. Also, check out their social media profiles for updates on their projects.
Check out the Blog Feed for other posts that you’ll enjoy, or head straight back to the Homepage for other features! Also, check out another AI article I recently did. I am interested in hearing what you think of the advancements, and what your opinion is on the progress of robotics as a whole. Thank You all for stopping by, and I hope you have a wonderful day!