AI in the now… AI and robotics…
So far I've come to believe that AI changes our lifes but what is AI capable of?
What will change and what is fantasy, for now at least?
Let’s take a look at Kuka one of the worldwide leading manufacturers of intelligent industry robotics.
Dr. Rainer Bischoff is head of research with Kuka is one of the leading exports in this field.
https://www.kuka.com/en-de/future-production/research-and-development/research-interests
He and his team are working on a new generation of robots that learn independently.
Much like a child they let their robots learn by trial and error. One of the tasks used here is identifying toy building blocks and sorting them.
“The robotic system has taught itself how to grip the toy building blocks. This means there was no human involved in teaching the system how to handle this task. This means it started to try to grip the pieces and wasn’t successful at first, like a child. It’s success rate in the beginning was around 1-2% but monitored itself constantly while doing this. Through this constant self monitoring the robot then was capable to identify when a picture, taken by his optical sensors, worked with a certain kind of grip and when not. By utilizing this strategy he, and in this specific lab constellation, is now capable of managing this task with a success rate of over 90%.
But what happens if you pitch in a new object that the robot hasn’t seen and gripped before?
Künstliche Intelligenz und die Veränderungen in der Gesellschaft (2/2) |...
Well, the described learning process takes place here then too. He’ll try again and again until he found a strategy that works for this new item too. At first it fails again and again because of weight and form differences. But every failure is a step forward in learning until he finds the right strategy for this new object. They will maybe let the robot 1,000 times over and over and then the results are fed into the neural network of the AI that controls the robot.
Intelligent, self learning robots. They identify factory parts/work pieces and assemble them for instance. With the help of AI they are capable to adjust themselves to their environment.
But the development of this is still in it’s infant state.
Dr. Bischoff explains by his favorite example: “Chess is my favorite example to show where we are with this right now. Today we have computer and AI that are capable to beat chess grandmasters but what we don’t have up to date is an robot that goes to an shelve, takes out the chess board and the chess figures that are placed in a box, takes them to a table and sets up the game with board and figures by opening the box, taking out the board and placing it on a table and taking out the figures one by one to set them up on the board correctly to begin the chess game. Any 6 year old kid is capable of this but up to date no robotic system can do this.
This means everywhere where this physical, practical intelligence is needed we are still failing today. He believes it will take another few years at least to accomplish this.
Nonetheless the machines are improving and are getting “smarter” constantly.
In this ad there are still tricks used to make it appear as flawless.
The Duel: Timo Boll vs. KUKA Robot
But this robot here from the university Tübingen (Institute for computer sciences) has learned ping pong and shows what already is possible in the real world.
Cognitive Systems Table Tennis Robot
How long will it take until robots are simply better in some areas than we?
Well, robots are already better than humans in many areas. More specifically in areas in which a highly accurate repetition of tasks in needed, where strength/power are needed, and a high precision is needed. Where robots still aren’t as good as humans is where sensory comes in to play.
Dr. Bischoff believes that we have at least 10 to 20 years in front of us until any meaningful comparison, also just for some areas, between humans and robots are thinkable.
Humans use all their senses and we can do more than robots but the robots begin to learn!
Again my two Satoshis on this...
I loved to see that there are big misconceptions about the progress that has been made in intelligent robotics. AI and robotics are often mixed up in the public perception imo. Looking into this gave me a better understanding where to draw a line between the state of the art achievements and pure fiction.
The longer and the deeper I look into the AI possibilities the more sobering-up in my understanding takes place. This pushes away some of my concerns but also increases others dramatically.
Why can't I kick the feeling that we are kids playing with matches?
At last, here my favorite robot/music video! Enjoy!
AUTOMATICA - Robots Vs. Music - Nigel Stanford
Articles in this series so far:
AI in the now... China, "security" and "efficiency" vs individual freedom?
AI in the now… autonomous driving cars… status quo... ethics?
Idea, credits...
I've been researching in this field out of pure curiosity for the last years with some, not much, effort.
I've read a lot of different papers from researchers/universities all over the world for ongoing research in AI.
Because of a lot points of contact to my day job in information security I also always try to look at this from an infosec/opsec/privacy perspective.
BTW...
I've found a recent documentation about this field from Ranga Yogeshwar and the German public broadcasting service NDR very interesting and balanced.
I picked up their red thread for this series of articles.
Here the links to their 2 part documentary (in German):
Künstliche Intelligenz und die Veränderungen im Alltag (1/2) | Doku | ND...
Künstliche Intelligenz und die Veränderungen in der Gesellschaft (2/2) |...
Let me know what you think about this in the comments or in an tweet reply!
Gif from my friend https://twitter.com/smilinglllama!
Our world is evolving at a rapid velocity. Thanks for sharing.
Posted using Partiko iOS
My pleasure! Thanks for your vote and comment!
Yep... sometimes it's hard to keep track!