Media is awash with robots falling over. In all fairness, the electronics and subsequent programming algotithms to allow a robot to walk are becoming increasingly sophisticated, however, when you see one fall, it's not a pretty sight. Arms and legs flailing, bits of bodywork being smashed off and ejected, it's quite worrying. There's so much focus on the difficult job of retaining balance, that it seems the engineers have overlooked how to let the robot catch itself in a fall. It's like the robot isn't allowed to fall at any cost, but they do, and they will.
Russia's debut humanoid robot was no exception, the look of disappointment not only on the creator's faces, but on the robot's as well.
So, here's my suggestion. Robots fall over, people fall over, let's accept the fact. But why can't the programmers allow the robots to do so with some grace. Surely, the sensors in the robot can catch the fact that the robot has gone down? The robot responds to the fall with a series of extreme moves to help retain balance, but there will be a point of no return where that will no longer be possible. I think this is where a new algotithm should kick in.
When a person falls, they stick their arms out to try and catch the fall. Why can't the robot do this? Move the robot's limbs into a damage limiting mode, letting the robot continue it's fall, then let it pick itself up in a calm manner. It's that simple. When the robot continues to flail as if to still get the balance right when it's already fallen, it's quite disturbing. If a robot falls over, calmly, picks itself up, and gives a small I'm OK wave, I'm sure that will induce far more sympathy and acceptance.
