WOW, great response. Thanks!
First of all, let me explain what exactly do I do. I work in cable industry and we have so called CV (continuous vulcanisation) tubes. That is where you put insulation on bare wire. Now, every once in a while some insulation material is stuck to the inside of the tube and it becomes hard as an rock (500C is in there). Then it starts to damage the cable that scrapes on this which is not good, of course. We have to be able to locate this problem so that we can se where and what is causing damage to the cables. Now, brand new pipe robot costs over 10000$ which is way to much for couple of looks inside those tubes (once per month max). Now, since this tube has a shape of catenary, has almost same diameter all the way, doesn't have bends, it's basically clean (except those couple bumps which I'm looking for), so no water, dirt or other nasty environment, it's relatively short (150m, although I only have to go in about 60-80m) I though that this can be done much cheaper than 10000$. I was thinking under 2000$ (not including PC of course). Industrial pipe robots cost that much because they can go into tubes of various diameters, vertical, through bends, in water/mud/what not, further that 150m, work 24/7 etc. So, when I remove all those things, I think this can be done for under 2000$.
Design would be very similar to this:
So, what I need now is a way to control those 3 drive motors, get a video from CCTV camera and control PTZ.
@jackwp
- 4 twisted pairs, 8 wires
- I didn't plan any charging. Well, around 15 mins would be great (driving, paning/tilting, camera and LED's)
- Well, here is what I thought. One pair from PTZ camera to PTZ joystick for controlling PTZ. That is basically plug and play solution. Second pair for video. So, two baluns, one on each side of the cable. On control end some specific RS485 capable monitor or laptop with RS458-USB converter. Again, plug and play. So, now I left with 4 wires (2 pairs) which I thought could use for control somehow.
I was thinking about having external power through 2 pairs (so, 2 for VCC and 2 for ground), but I'm not sure about it. 150m is a lot, and simple small LiPo should be better solution.
Yes, one person will be helping with wire. He will push the wire in the tube and pull it out (as much as possible). Also, I was thinking about using Kevlar strip to pul the robot out if necessary. Kevlar is light and very very strong and I might have some at my disposal. 
@MAS3
Good idea with charging, but I don't think it will be necessary.
@cr0sh
Yes, DC motors would probably work fine.
Ethernet would probably work, because as I said, I actually only need to go in about 60-80m, 100 max (second part of CV tube is for cooling the cable and we never have problem there).
Thanks for the suggestions, I will look into that.
@Robin2
All assumptions are right! PC would be great for control, but 3-way switch is an option as well. Although, I would like to have linear control over motors as oppose to ON-OFF.
150m = around 4kg - however, CV tube is going down (about 18 degrees) so robot shouldn't have much trouble with that. We can pull it out backwards.
Wireless is not an option because this is completely metal, grounded, tube. Also, it's not completely straight so antenna at the entrance wouldn't work either. Otherwise I would already do that using my RC equipment 
Yeah, robot stuck in a tube is not an option either. 
My biggest problem is that I don't know much about programming microcontrollers. So, complicated solutions where I need to program a lot is not an option. However, if there is a solution that involves Arduino, but there already is complete code which I just have to put on, I shouldn't have a problem doing this.