this sophisticated AI boot camp teaches robots to move (realistically) like a human or T-rex 

it takes practice to get to carnegie hall, but apparently it takes even more for AI to simulate realistic human movement. motion control remains a large hurdle for programmers/roboticists — a problem that can be summed up as the creepy repetition that AI can’t seem to shake. despite hours and hours of painstaking programing of the physics of movement, computer bots still can’t deceive audiences into thinking that their movement is real. however, UC berkley seems to be on the right track in fixing this bug. its research has resulted in a sort of sophisticated ‘boot camp’ for AI, which could just as well teach robots to move like jagger — or any other human for that case. it combines reinforced learning — a process of refining AI intelligence through trial and error — with motion capture — which allows AI to compare its own movements to an actual reference.

all gifs courtesy of berkeley artificial intelligence research
AI testing his parkour skills

the berkeley artificial research lab essentially wrote an ‘if statement’ that perpetually rewards the AI for mimicking the real life example more accurately. after millions of trials, the AI behaves accordingly, completing backflips, cartwheels and other sophisticated motor skills. the program makes the AI learn to move almost in the same way that a dog learns a trick — even perhaps in the same way that a human learns. in fact, this programing theory is partially based on human psychology. over millions of tries, the result is an AI movement almost indistinguishable from the reference. 

illustrations of the progression of motor skills

it’s hard to realize the significance of this work from these gifs alone. as we’ve probably seen similar animated movement before, it’s important to distinguish that now in this example, the programs are not simply doing as they were programed, they are operating on a strong foundation of knowledge resulting from countless experimentations, and can take this knowledge to different environments. the possible uses are wide ranging, from game design, VR, robotics, animations etc, further blurring the line between the virtual world and reality. sure some dystopian future from film or fiction is coming to mind right about now, but there is no need to be nervous — at least not yet.

the movements of a lion simulated here

the movements of the T-rex