My exact same thoughts (at least the PWM part )
The problem with lowering the voltage is that it doesn't reduce the motor speed. We supplied 12V, then 10V, then 7 V and the speed was pretty much the same but it we supply less than 6V, it will stop rolling and we are also using a LM7805 in order to supply 5V to the sensors.
Now, would it be a good idea to reduce the current supplied to the motors? I guess it would but I don't know how to achieve it. Maybe placing the motors in parallel so the voltage would be the same and the current would be half the current value when moving.
I would love to know your opinion on this idea.
Thanks!
Motors can become very complicated.
The voltage you supply to the motor. E.g. 12 volts, will allow it to build up speed to a certain RPM (the RPM is NOT fixed, it depends on load and other factors). But motors take time to build up the RPM (and hence speed etc). So although you are giving it 12 volts, its RPM may be a lot lower at this exact point in time, as it will take time to accelerate up the RPM which it can potentially reach with that voltage (and current drive capability).
I.e. The motor is not at its final RPM for that voltage/load, until the robot has reached the max velocity for that motor power/voltage/current value. This takes time, and may not even happen, if the load/gear-ratios don't allow it. I.e. the RPM may be held back because the motor has not got enough "push" to accelerate any more. Anyway motors get rather complicated, when you get into the finer details.
This is why I keep on rattling on and on about using Microcontrollers (N.B. I fully agree that it is probably WAY, WAY too soon for you to use MCU's yet. As it needs lots of knowledge and experience etc, to use them well, and they are rather complicated until you know about them).
But the MCU would be able to use quadrature decoders (or other methods), and hence know/read the exact wheel/robot speed. N.B. The wheel rotation speed can be different between the left and right hand sides, and it can even be different from what the robots velocity is (e.g. Wheel spin/slip etc).
Once you know the wheel speeds, robot velocity (estimation) and can PWM the motors inside an MCU, using a PWM'ed H bridge motor drivers, plus position sensors (line sensors), and maybe even accelerometers and gyroscopes etc etc.
You can create/use algorithms to make sensible choices about how you will react.
What you are trying to do, is tweak the robot (line followers) performance, using rather simple/crude means, which are mostly ignoring the wheel speeds, robot velocity/direction, etc etc. Hence in many cases your attempted tweaks will not work well.
You can limit the current, but it can reduce the torque too much, and potentially stop/stall the motor and/or give it too little torque. My understanding is that PWM gives better results, because during the "ON" parts of the PWM duty cycle, it effectively gives the motor (potentially) the FULL current, and hence tends to maintain reasonable torque WITHOUT stalling the motor, too badly.
This is why (if I understand/remember things correctly), drill speed controllers, are also best done using PWM techniques or similar. Otherwise the drill bit becomes too weak (too little torque), to successfully drill into the materials that you want to make holes in.
tl;dr
You may have to accept the fact that without using a microcontroller, and extra sensors and better ways of controlling the motor. You are going to be rather limited in ways that you can improve its performance.
But don't let me put you off too much. There are going to be improvements to be had, I would imagine.
Going the microcontroller route becomes EXTREMELY involved, complicated and time consuming, going well beyond the kind of technical resources and time, I expect your team intend to put into this project. People who do work like that, are usually already Electronics and/or software experts, to at least some extent. But even beginners sometimes have a go at stuff like that. There are also kits available (probably not suitable for your studies as it would probably be considered cheating).
EDIT:
Please DON'T be too put off by my mentioning MCUs. There are many, many ways of doing Line followers, which people use which DON'T involve mcu's, and their performance can be quite good, maybe even very good. In principle line followers, is simple enough to do, without MCU's.
I'm just way too biased in favour of MCU's, and this bias seems to be leaking out too much in my answers.
Sorry!