OK. In your description above, you used the analogy where the potential looks like an English tophat. The quartic hilltop model is a particular case of this. Essentially, if you graph y=1-x^2 and compare to y=1-x^4, the latter is more flat at x=0 than the former. It is a quartic (degree 4) hilltop.
The 'potential' in all of these measures the amount of energy for various strengths of the inflaton field. As in your description above, inflation requires a large value for zero field strength and a smaller value as we move away from zero.
The various potentials just are different mathematical ways to accomplish this that are motivated by other physics.
Most of the models for inflation do not focus on identifying the inflaton, but merely describing its properties.
The R^2 model adds another term to the Lagrangian for gravity. Such additions have been proposed independently of this, so it isn't an unreasonable thing to pursue. In this, there is not a new particle, just an adjustment to how gravity works at large field strengths. But it does give a period of inflation and allows reheating in a natural way (so that inflation is stopped).
Another good source of information:
http://pdg.lbl.gov/2016/reviews/rpp2016-rev-inflation.pdf