Can a fruit fly navigate 100's of miles like a UAV? Does that make it less intelligent? You are confusing some issues with sensors and physics with intelligence. I've build CAD algorithms to do automatic routing of integrated circuits and printed circuit boards. The complexity of the mazes in this case far exceed anything a fruit fly would ever encounter in the real world. In their own fashion those algorithms far exceed the intelligence of a fruit fly.

For practical applications it is necessary to build and integrate sensors as well as brain. For a measure of intelligence you can run simulations and planning while giving something virtual physical properties and sensors input. Once again, I think you could easily exceed fruit fly "intelligence" in such simulations.

Finally, I find it amusing that many people like to keep changing the bar on intelligence and basically whatever artificial creations can do is defined as non-intelligent and whatever something alive can do that artificial creations can do is defined as intelligent. Auto-pilots keep getting more sophisticated, and it used to be though that piloting a craft took intelligence, but I guess now that auto-pilots can pretty much do everything it is no longer intelligent. Same with chess playing and myriads of other things. When will people realize it is just a matter of time. We are able to pool our own intelligence and combine much knowledge that then may be handed off to computer software. Software that does medical diagnosis now probably exceed virtually all doctors abilities in a general sense.

The barriers keep falling and Moore's algorithm keeps doubling and the number of software/design engineers keep increasing. There will be fewer and fewer things that humans do that is unique as compared to what artificial constructs can do. I find that with each order of magnitude of computing power increase, there are usually significant knew applications that can be accomplished. This happens roughly over the course of a decade. Given the current state of sensors, robotics, and the first primitive integrations of all of those, I would say in another 10 years, you will be amazed at the complexity of interactions robots will be capable of.