I'm a beginner too and I have some advice on the $50 robot project. If you have to buy batteries, the AVR programmer and need to expand your tool kit in order to build the robot it can quickly become a $200+ project. I'm also thinking the shipping charges weren't factored into the cost of the robot. While the 'ghetto' portions of the robot save money, in reality they save relatively little compared to the servos and micro-controller board parts.
In terms of tools, For instance I have an el cheapo Weller soldering iron that is useful for soldering extension cords together and little else so I'm replacing it as soon as I get my next paycheck. A competent temperature controllable unit starts in the $80 price range and just goes astronomical from there. A breadboard for prototyping would be great and they start at $9 for a mini one you could place on a robot for testing. For bigger projects I would get a trainer kit , but they start in the $150+ range, unless you happen to find a used one that a school is selling off.
I did some guessing and while I'm glad I have 'most' of what I need to get started, a good kit consists of a tool box full of stuff that you don't find in your mechanic's tool box. Along with that, a decent breadboard, temperature controllable soldering iron, small multimeter and you can be in business for around $200-$300 if you watch what you buy and buy good stuff no matter what. Later, an O-scope, electronics trainer, a decent Dermmel, and a home built CNC machine should be on the wish list.
Having said that, the next robot project would be fantastically cheap to build if you go through the pain of the first one; I'm guessing as little as $15-$20. I even see the possibility of eventually building one at no cost if re-use of parts is considered.
I've also begun to embrace the idea of scrapping junk appliances and electronics for the components as the potential to save is considerable. Just a CD-RW drive can yield a couple dozen caps, a micro-switch, micro-audio plug, and 2-4 motors depending on the model and suitability of the motors used. And let's not forget that older is better for scrapping purposes since the new stuff is mostly surface mount these days.
So, the real question is, do you want plug-and-play for the speed of development and low risk of massive failure, or do you want to really learn how this stuff works?