Siu Lun

This article got me thinking, and I was going to post it over at TheReg as a comment but it turned out more like an essay.

A nice summary of what the robotic industry has in store right now, but I want to point out that:

“His ultimate goal is a self-driving car, a vehicle which would be a safer and more attentive driver than humans, freeing our minds from the roadway, speeding traffic by packing cars closely. This indeed would be a pivotal social and technological revolution”

Indeed it would revolutionary, but do you think the world will ever get to this stage? Even when “truly” autonomous agents exist that are capable of driving cars, there exist possibilities that things can go wrong, don’t forget A.I. is the building of “artificial” intelligence, as it is defined by our mind. Hence whatever human mistakes exist today, it will no doubt also manifest in AI and automation. Unless the government around the world changes their entire transport system to automated control, such that a partially controllable environment exist. The software not only have to deal with normal driving manures but also lousy drivers, which further increases the possibility of even more fatal accidents.

For “simple” task of working in a factory environment “statistical inference software” is good, because the environment is controllable but in the real world, “statistical inference software” will not be near enough to control anything as it lacks adaptability. Now I present to you three scenarios, would you rather:

1) Drive your own car?
2) Let the controller be a statistic based software program?
3) Let “your friendly A.I.” pal that has grown up with you drive?

For me, 2 will never be my choice, as it is a deterministic system and I do not believe that the world itself is deterministic. (i.e. I do not believe in fate)

3 is a possibility, depends how much I trust my childhood A.I. 1 is definitive, although I’ll probably be better off going for 3. Some of you might want to know what 3 is? your friendly A.I. is a program that develops alongside you, just as you develop knowledge as you grow older, but it will not develop any human endeavour such as woman, alcohol and drugs, but is able to understand friendship, responsibility and the rest.

My point here is that statistical inference software are only good for static, deterministic and environments and many who are developing with the idea of one day realising an ambitious dream such as a robotic football team will not succeed unless they change what they’re researching on. The research should be on evolutionary systems and automated learning, i.e. how can we build an A.I that is capable of learning and adapting. Instead of trying to analyse every single little possibility that may arise for a certain application of automation.

I’m not saying it is not possible however. It is possible, but imagine first, having to analyse all possibilities of what happens when driving a car, then research into how to tackle every single possibility, then trying to program it and test it, and then….. trying to integrate them all into one single system and test it…. AND THEN .. you have it, that’s one car system that may work, how long would it take? 100 years maybe? maybe more? … but more importantly… how much money? would the end development cost be worth it, considering it is just for one application.

And then you have to do the same thing to boats, submarines, aircraft etc… the problem with most people is that they see things in the short term, “what can I achieve today”, but when you’re talking about a system of automation to the scale of global transportation, there are much more to it than our eyes can see.

Lastly, I don’t like the title, ‘Why do robot experts build such lousy robots?’, it’s not that they’re lousy, but it’s because the world is too complex. It’s a bit like saying “Why do game makers today make such lousy games?”. Let’s face it, games today are the same as they were in terms of game play as when they first started making games back in the 80s, the only thing that has changed is the graphics eye candy, but this is precisely my point, the eye candy on robots are becoming more extravagant by the day, and in contrast to the games industry, the robotics industry has made significant progress every year in automation.

The aim of the robotic industry to create truly automated agents is like the dream of a games designer to create a truly immersible real-time world that can allow the entire gamers population to play whatever they want in.

Couldn’t really find anyone to help out with the site, so it’s all down to me to make it and update it, but currently I’m too busy developing websites at work and for others. So this will be postponed until I have the time.

With the launch of the EU domain, I’ve aquired a pretty nifty and short one so anyone should be able to remember. I plan to use this site a build a site about Robotics and A.I., things I’m very interested in, hence why I’m at the University of Reading where Captain Cyborg resides.

I’ve got a lot of motivation in this field, possibly because of Gundam, possibly because of Chii, or maybe I’m just a said lonely git who wants to expand on this field so that one day I am able to program a bunch of good friends? Who knows. lol

In any case, since I’m busy with my third year exams I shant be working on it until start of June.

This is the intro from XIRO.EU

“XIRO.EU is scheduled to launch around end of July and will become the net’s greatest resource site to find anything regarding to A.I. and Robotics, bringing you news from around the world in the advancement of artificial intelligence and robotics. The aim of XIRO.EU is also to rise awareness of the growing field of robotics and A.I. and initiate debates amongst the common people on the benefits and consequences of such research, giving you insight into the latest in research development.

We’re looking forward to the launch date!”

By we, I currently meant I, but if anyone who reads my blog ever is interested in programming A.I. and making robots etc… drop me a mail and we might be able to come up with something. =]