I believe that there is another side to this, although the points so far are true. I am a computer science major and have been following the development of robotics for over a decade now. As such, I personally see the companies that would readily sit down and work on something autonomous pick projects more along the line of Nao, ASIMO, iCub, or BigDog. From what I have seen, there is more interest in either humanoid high intelligent robotics or military grade robotics. There is more money, grants, and programs in those fields. And, since most of these companies are relatively small (like Boston Dynamics which was recently bought by Google) they do not exactly feel the need to spend time on "toys". Example, when the Nao robot was first developed they were very interested in making a model available to robotic hobbyists and such. But now only research institutions, universities, or people who work for the company are allowed to purchase it.
Also, remember Ugobe went bankrupt. This has happened to a few other companies like it or Innvo Labs. I think some people are afraid to put something like Pleo into production because they are not completely sure there is a market for it. So, we are stuck with Wowee toys called "interactive robots".