Doubts about the design of a mobile line follower robot

I'm an enthusiastic hobbyist of robotics and i want to make my first line follower robot, i've learned some important topics about microcontrollers, programming in c, so i think i have the enough skills for development of this project.
So my questions are.

1. Which is the best actuator between a dc motor with gearbox(gearmotor) and a servomotor.

2. I am planning to add encoders on the wheels, someone could suggest me what type of encoder is the most suitable and why

3. How can i know the minimum and maximum torque that i'll need for the dc motor or servo

4. what is the most suitable structure for the robot considering the aerodynamic.

Thanks

Or you can look it up on the web??

Example
http://www.zuccante.it/informazioni/Microrobotica/sandro/MaterialeRobot/OTTIMO building-a-line-following-robot.pdf


Encoders use a 360 degree circle marked as you would see inside a printer with infra red sensors with input into gpio pin and count Giving you 360 counts = One revolution

Maximum minimum torque of motor is the specification of the motor or its current sensing and monitor via adc input to micro

Then you also want Fuzzy logic as well


Aero dynamics Ask a and aeronautical engineer possibly want Nasa Are you planning on making this fly too in which case have a word with Richard Brandson perhaps not his balloon wasn’t to successful i think it went up in smoke In any case if you do decide Richard brandson ide get a parachute and two way radio you might need it His cellpone airtime isnt that clever neither is his Virus sofwtare provide by his ISP

You have o be a little cheeky now and again dont you

Probably a gearmotor is cheaper. It's commonly driven with a simple h-bridge. Unless this is high performance the h-bridge could be driven from software. Make sure both high and low outputs to one bridge can't be on at once or your h-bridge will be short lived. I've made software bridges in a few applications.

Quadrature phase incremental encoders are inexpensive and give waveforms that can be decoded into fwd and reverse but require fast counters and/or an encoder IC. I used simple logic to decode the quadrature into up and down pulses and input each into a counter in a microprocessor, cascaded in software to 16 bit. The position was the unsigned difference between the encoders. Overflow did not affect the position.

You probably don't need to know position and may not even need encoders. I imagine you could move each motor at a set speed and slow one as you moved off the line.

If you just want to follow the line you can keep it simple, the elegance of servo control. If the line is on the left side you need to slow the left motor, that's it. You will only have to adjust the gain (how much do you slow the motor)
Absolute encoders are much more expensive and give absolute position for one revolution and then either reach a stop (watch out for that)or overflow so you must use either software or hardware to count revolutions.

Torque depends on your application. On the flat without a lot of friction the torque will mainly affect the acceleration and deceleration. Most any motor will be powerful enough in this case. If your motor's too powerful it will be hard to control.

Unless you are going faster than I can imagine or operating in a wind tunnel aerodynamics is not an issue.

I use stepper motors for most of my applications. Aerodynamics doesn't come into play until you get up to ridiculous speeds. 4 wheels in a skid steer arrangement, or two with a free following caster work well. think zero turn lawn mower. Read "Making Things Move" by Dustyn Roberts - "http://tinyurl.com/makethingsmove(Making Things Move)":http://tinyurl.com/makethingsmove