Robots cannot disobey, but they can harm without knowing.
A simple example is if a worker falls in a fully automated car smasher. The "three laws" would prevent the robotized machine to continue operations. A robot today, wouldn't know the difference and if programmed to smash, it will smash on mark, every time, no matter what is in the pit.

So no, the robots cannot disobey. But as soon as you introduce AI and the ever changing meaning of good and bad, an AI routine could assign a "bad" value to a command and therfore skip its execution. This isn't what you think of disobeying, but it's as close as it will ever get.

Still, the three laws are flawed.

If you follow human commands, at the end humanity will self destruct, with or without robots but quicker with the latter. At the end the only real commercial use of humanized robots will be for sex, as that is the only safe application. Think of coin operated machines.

The non-humanized robots, think of mainframes, core set of procedures, are already used in many industries, but they all have redundant safety switches... and lastly, they all have one human being supervising their operation. I don't see this changing, but I could be wrong.

As for fear of the unknown, I have to admit I have not yet bought the I-roomba (or whatever the robot round vacuum cleaner is called) because of fear. Yes, I fear that my kids will leave on the ground somethign that might sooner, rather than later, brake the robot. I never have been too keen abotu throwing money away, and with three kids I have done that one too many times already.

The only robot I can live with is my cell phone, because I can turn it off when I don't want to be bothered ;)