Hm... interesting question.

The ceiling test could give a "false indoors" if the robot went under a tree or something, and the light / dark test won't work since both indoors and outdoors can be light or dark from time to time.

Tricky AI problem this: we as humans "know" what it means to be indoors or out, but to define that (which is what you have to do before you can test for it) is amazingly tricky. One of those things that separate us from machines?

Got a link that actually works? That one just gives an error.

Lose the presumably accidental "s" on the end of "pdf"

HK's website calls that a "super strong" servo, so maybe it requires more current than your power source can supply?- especially since you say it works with a different servo.

(PS: the 9v is not a good idea, since that's a 4.8-6v servo.....)

Can you recommend an where I can start looking up stuff like this? I tried using the word or where you put the bars like in VB?

About half-way down the left-hand column of the reference page

You could use a light dependent resistor to check the light: have a look here. Then have a look at this multi LED tutorial.

Nope.... says that link suffers from the infection known as URL:Mal

I can probably by-pass the anti-virus, but then, that's what I bought it for....

How about this. I have a couple of them that tested out fine and work as explained. Just need the killer application to use them in now.

My Avast! anti-virus blocked that site as possibly malicious....

Although my idea does use a micro: just not doing the fan speed thing which is getting a bit complex. A micro-less solution is even better of course, unless the micro is in play for some other part of the project and is "free" in that sense.

Ah, then what you want is to look at a voltage divider, where the output voltage (which I guess then becomes your input voltage) depends on the ratio of the two reseistances.

In fact, a potentiometer is exactly that, where the "mid-point" actually moves along a long resistor. Pots have 3 connections: the "top" being the input voltage, the "bottom" being ground, and the "moving mid-point" is your variable output. Have a look at this tutorial.

I'm not 100% sure what you're getting at, but this might answer your question.

The Arduino does use two pins to read an input: there's a ground as well as the pin you're using. So if you had 3xAA cells giving out about 4.5v, and you poked the +ve with a wire from an Arduino pin, you would get no reading. But if you connected the Arduino ground to the -ve of the battery, then you would get a reading: a digital pin would read a high, and an analog would give you about 0.9 of the 1023 max reading, ie about 920.

Well that might be a bit of a complex solution sonnyyu, where the OP really just wants off and on.

First you need to look for a temperature sensor: there are lots of threads about them, and have a look on Sparkfun or adafruit to see what they sell.

Second you simply hook the motors up in parallel using an NPN transistor, something like this. (Assuming they all work at once... )

Then third, write a sketch to read the sensor and switch the fan on and off. You'd need to think of something to prevent it switching off and on too quickly if the temp is hovering near the threshold, some kind of "band"  idea maybe.

Maybe consider the fans working separately (sounds like you want them working together?) and have the second and third switch in at higher temps, ie if one or two fans is not controlling the temp adequately.

[Mecha]

I think the subject I want to look at is "mechatronics"? Can anyone confirm or correct this notion?

Yep, that's it: combination of Mechanics and Electronics afaik

When you get to the choosing stage, or just want a virtual browse  , have a look at Firgelli.

There are lots of ATX PSU hacks on Instructables..... ATX supplies have 3, 5 and 12V and 100s of Watts.

Presumably the purpose of the indication is to let you know when the robot's going to die? I Googled Li-Po discharge curves and from what I see, the voltage is fairly flat until it drops of a cliff, so looking at the voltage doesn't give a good indication of how long it's going to last.

Other threads on this board have indicated that you really need to monitor the power usage which, when you incorporate time, tells you the energy used and that can be subtracted from the full-charge energy capacity. But I think that's probably a whole project on its own....