1

General Robotics Talk / Re: Robots should be made of wood!

« Last post by Zero on Today at 11:14:17 pm »

That's a sustainable Terminator!!

Weight and flammability shouldn't be a problem: think about radio-controlled aircraft (motors are hot).

It could be really beautiful.

Yeah... this will never happen, but such a project would have been so cool.

2

General Robotics Talk / Re: Robots should be made of wood!

« Last post by LOCKSUIT on Today at 11:01:40 pm »

Or, wooden robots in a 3D simulated world in VR. Such, customizable wooden textures.

It's breathtaking.

3

General Robotics Talk / Re: Robots should be made of wood!

« Last post by Don Patrick on Today at 08:54:16 pm »

If you look at that picture you'll see what the real problem is: It's that so many roboticists think that a skeleton with an excavated skull bearing a death grin is appealing.

4

Robotics News / Re: Using Bitcoin to prevent identity theft

« Last post by keghn on Today at 08:39:01 pm »

Ever wonder how Bitcoin (and other cryptocurrencies) actually work?:   

5

General Project Discussion / Re: Ahead Network Parser: developing an option towards NLP and some other thingies

« Last post by ivan.moony on Today at 07:15:26 pm »

How long it actually takes you will largely depend on how long you're willing to waste your time trying to reinvent everything yourself like I did. Once you realize that no matter what you invent somebody else already did it and did it better, then you start learning from others and you make much faster progress. It's only then that you stand a real chance of inventing something that actually is new. Hubris is the biggest obstacle to progress. I wish I'd learned that when I was a lot younger.

Hubris... there is another thing which is following me for a long time. How much invents there are stored in the whole vast of the Internet? As much as Internet could be a life savior, that much it could bombard you with a vast of unnecessary propositions among the ingenious ideas that lose the track in the sea of informations. It is almost impossible to find what you are looking for on today's web. I tried Earley, was slow, read about CYK, was too slow, heard of LR and PEG, turned out that they are not handling CFG, checked out GLR some time ago, decided it is too complicated, and finally found out papers on GLL, sounded great, but it was to late, I was already ahead with compact parse forest and my version of algorithm. Lastly, I checked out GLR again, just to find out that it is not that much complicated. Is it only me, or is it that hard to find existing solution satisfying, so I tighten my sleeves and do the work myself. Maybe I'm just out of luck, maybe I'm just not born with a lucky star in my sky, but that much often I can't find what I need. I would be the happiest man in the world if everything I seek and hopefully find could fit into my plan, but it doesn't. I don't know anymore is it curse or bless, but something unexplainable is happening with my research, I'm telling you. I'm always forced to find out my own solutions, so I do the only thing I can: reinvent the wheel. It happened with functional programming, and now it happens with parser.

[Edit] Hubris... I guess you're right, I don't settle with average solutions. 

6

General Robotics Talk / Re: Robots should be made of wood!

« Last post by korrelan on Today at 05:08:27 pm »

https://i.kinja-img.com/gawker-media/image/upload/s--V9d0WTDx--/c_scale,f_auto,fl_progressive,q_80,w_800/1260617945207929487.jpg

 

7

General Robotics Talk / Re: Robots should be made of wood!

« Last post by Don Patrick on Today at 04:38:16 pm »

It's a nice idea, but weight and durability, not to mention flammability near potentially frying electronics, are some reasons why most robots aren't made of wood. You could clad the surface with thin sheets of lightweight wood just for aesthetics or make plastic with fake wood grain of course.

8

General Robotics Talk / Re: Robots should be made of wood!

« Last post by WriterOfMinds on Today at 04:22:25 pm »

wooden bots could be DIY open-source community-driven projects, instead of cost-prohibitive plug-n-play toys...

In my own experience of personal robot-building, the structural parts aren't the primary cost driver.  Motors and electronics are the things that make the exercise become expensive fast.  And there are lots of options for DIY manufacturing with plastic; you can CNC mill that too, or use a 3D printer.

One thing I do appreciate wood for, vs. many plastics, is its rigidity.

9

General AI Discussion / Re: Do any neural network training schemes vary the nonlinear activation function?

« Last post by keghn on Today at 04:18:10 pm »

A visual proof that neural nets can compute any function:   

http://neuralnetworksanddeeplearning.com/chap4.html



Which Activation Function Should I Use? : 

10

General Project Discussion / Re: Ahead Network Parser: developing an option towards NLP and some other thingies

« Last post by ivan.moony on Today at 03:16:48 pm »

I investigated GLR parsing. Ahead network parser is almost it, but not quite. The difference is that GLR is bottom-up, while ahead network parser is top-down. Ahead network parser also constructs the look ahead table elements on the fly, and ahead symbols are stitched to symbols, while GLR stitches ahead symbols to sequence positions. Thinking about speed, GLR parser could be somewhat faster,  but I kind of like a simplicity of ahead network parser. I'd almost abandon my design in the favor of GLR parser, but what a heck, a simplicity is an advantage too.