Historically, Japanese has had several morphemes that change the transitivity of a verb. Most of these pairs involved lexicalised combinations of some verb with one of these morphemes.
The morphemes are:
-(a)su - causative. You can see it in words like ゆらす ('cause to shake', compare ゆれる 'shake').
-(a)ru - passive, or rather, general agent deletion (English's passive implies an agent, this doesn't). Visible in your example とまる ('come to a stop on one's own', compare とめる 'cause something else to stop').
-e(ru) - a kind of transitivity flipper, it can make transitives intransitive or intransitives transitive. An example of the first is さける ('split open', compare さく 'tear'), and an example of the second is つける ('attach', compare つく 'stick to').
Many of these pairs, とめる~とまる included, seem to have had these morphemes added to both members.
I'm not sure this is a grammatical process, at least any more; none of these morphemes are still productive as far as I know. You're probably best off remembering each word as a single lexical unit, especially considering the fact that there's a good deal of variation among what pairs up with what. Just remembering that あげる is 'raise' and あがる is 'rise' prevents you from trying to make non-words like *あがす or *あぐ.
(The Middle Japanese -(a)su and -(a)ru were nidan verbs, and became the ichidan -(a)seru and -(a)reru, which are still extremely productive in Modern Japanese. I don't have an explanation for why verb forms that incorporate them are godan.)