can be your best friend!
Rubik Cube robot solvers are magical pieces of hardware. The level of sophistication and effort that is put into them by, more often than not, engineering students is astounding.
Some of the robot solvers that we’ve seen at this point defy logic when it comes to solving time. Fractions of a second are all it takes for some of these machines to make a 20 move solve, which you wouldn’t think is possible.
Advancements in Rubik Cube robot solvers are being made year on year, and we’re going to talk about a recent addition to the roster that takes the challenge a bit differently.
Rubik Cube Solvers Are Single Purposed
Up to this point, Rubik Cube solvers have been single purposed. The whole point of the machine is to solve a Rubik Cube, and nothing more. That’s all well and good, but it makes you think about the other applications that this kind of processing power could have.
That’s where Open AI enters the equation.
With traditional solving robots, it’s generally a box-style mechanism is a rotating connection attached to each side of the cube.
It has one single purpose and is strictly designed to solve a Rubik Cube as fast as possible.
Open AI’s robot solver does things a little bit differently. Rather than just being a Rubik Cube solving machine, it’s actually a robotic hand that is capable of solving the cube.
It’s only one hand, meaning that the solution is incredibly slow, but the AI programming is capable of doing it, nonetheless.
The Bigger Picture
Why does the fact that the Open AI robot solver isn’t strictly for solving a Rubik Cube matter? Well, it matters because it reinforces the notion of reinforcement learning in artificial intelligence.
The hand was exposed to a variety of stimuli that it had never seen before, and it actively adapted to that stimulus.
Things like being prodded by a stuffed giraffe were done to the hand in an effort to test its reaction, and it was found that it could actively change the way it was solving the cube in response to that.
The robot hand was designed to mimic that of a human and is operated by a pair of neural networks.
These networks are trained exclusively in simulation, hence why the display that it has given with Rubik Cubes is so impressive.
There is a whole host of technical data on the Open AI project over on the official website. There are papers, graphs, stats, and discussions on AI learning that are all absolutely fascinating.
We would be here all day if we talked about it now, though, and it also doesn’t exactly pertain to Rubik Cubes.
That being said, if AI technology is something that interests you, it’s more than worth giving the official paper a read.
Why Do It, Though?
The technology is impressive, as is the ability to solve a cube, but why do any of that in the first place?
Well, Open AI believes that the kind of reinforcement learning that has brought the machine this far lays the groundwork for general-purpose robots in the future.
The idea is to develop a robot that is capable of complex manipulation tasks and solving a Rubik Cube is a perfect example of that.
The initial solve happened in 2017, and in those short three years, the technology has continued to come on leaps and bounds. That being said, there is still a long road to go before Open AI reaches its end goal of a general-purpose robot that is fit for household use.
What Does This Mean for Rubik Cube Robots?
The impact that this technology has on single-purpose Rubik Cube solvers is minimal, in all honesty.
Despite being both impressive and exciting for the AI and robotics world, it doesn’t impact the Rubik Cube world beyond being an impressive feat.
It only solves the cube 60% of the time, which goes down to 20% for a maximally difficult scramble.
Even when the machine does solve the cube, it does so slowly.
The technology powering the AI is a lot more diverse and complex than that powering actual Rubik Cube solvers. This AI was designed to learn and grow, whereas cube solving AI has one singular mathematical purpose.
AI technology is never going to be able to compete with the solving systems that are already being used today. There have been machines that have solved a Rubik Cube in a fraction of a second, which is a number that is continuously chipped away at.
It’s unlikely that an AI of this nature is ever going to be able to do that, especially when it comes to physically perform the movements themselves. Perhaps if the cube was on a program, it could keep up with solving algorithms.
As it stands, though, the Open AI robotic hand, while impressive and exciting, presents no real developments for the robotic Rubik Cube solving world.