The local hobby store have now complemented their PIC chips with a number of Arduino articles.

So anyone where I live who wants to get local gear are best off going for Arduino.

And because it represents an internationally spread standard, there are a huge (H! U! G! E!) amount of projects available for them. Excellent both for people who can't program or solder themselves, and for people who want to start with something working and then extend it, or for people who just want to get an idea and then do it all by themselves.

And no problems with compiler licenses.

A microcontroller development board, or a Arduino kit are both excellent routes. There is a bit of a conceptual difference, so it's best to spend a bit of time reading up on the two routes. But both alternatives are excellent for learning microcontrollers, or for developing simple (and also quite advanced, depending on user skills) robotics solutions.

The big advantage with the Arduino (at least for a large part of the world, or for people with access to a credit card) is the availability of a large number of very specific I/O shields. Such as shields designed specifically for robotics, with GPS, accelerometers, etc. This is especially nice for people who don't have the skills to solder themselves - especially since all new and cool components are surface-mounted. There are sites that sells individual sensors already soldered to small PCB, but it's even nicer to get a complete set of sensors on a PCB that mates directly with the microcontroller PCB.

Richard is just so "do what I did 1970". Just that he did not have GPS, accelerometers, earth magnetic sensors etc with RS232 and I2C interfaces for direct connection with a quite powerful microcontroller - and all code and RAM space you could need to write a larger program than your school studies could afford ;)

It's best to learn microcontrollers on a already working PCB (dev-kit or similar). When you do know how to program them, and do know how to connect external hardware to them, it's time to consider how to design own solutions from scratch. That route saves you a lot of time and frustration compared to starting with just a processor and lots of wires and try to both get your design electrically correct, and correctly programmed and with a correctly working program. How do you debug a program, if you can't even manage to print out any debug info on a serial port?

Professionals may sometimes have to start swimming in the deep water, just because there are no commercially available alternatives. But that is not a fast or economical route.

You have choices, so instead start with something that works, and learn how it works and how to make use of it. Then increase the complexity as you learn. Way more fun. Way faster. And normally way cheaper. Garbage dump components are the cheapest, but there are normally additional costs that are higher than buying something already working.