From the article:
"Instead of giving the robot a rigid set of instructions, the researchers let it discover its own nature and work out how to control itself, a process that seems to resemble the way human and animal babies discover and manipulate their bodies. The ability to build this "self-model" is what makes it able to adapt to injury."
Methinks the key phrase here is "
seems to resemble" human and animal qualities. While this sounds impressive at first glance, it's doing nothing more than cycling through possibilities until it finds solutions within the parameters of its directive (similar to a "brute force" password cracker, just a bit more complicated).
If results start to fall outside those parameters (due to the "injury" mentioned), it simply starts the process over with modified options (i.e. one less leg).
At best I'd guess it uses a simple neural network, at worst, nothing more than lengthy "switch/case" scenarios (or the equivalent in whatever language is used). I see no evidence of programming magic here, just a novel application.
Don't get me wrong, I do think it's a nice piece of work (with real potential for practical use), but a bit misrepresented as to having any kind of cognizance.
Just my opinion though, perhaps I overlooked something.
moo
