Well, almost. An English speaker is fairly likely to pronounce sh- correctly, and certain to pronounce s- correctly.
I think pinyin is hard to understand for a layman. Eg how many people said Guangzhou or Sichuan correctly on the first go?!
O and wo are not ambiguous in Hepburn, as they are being explicitly differentiated.
Another thing that Hepburn doesn't help with is は, transliterated as "wa" in the case it's the enclitic particle, instead of "ha".
I did some research on Japanese wikipedia (四つ仮名) and found that most uses of di and du were changed to zi and zu in 1986: ふじ(mountain) and ふぢ (flower) are now both written ふじ.
There is a list of words that have changed and ones that haven't.
The government decided that the particle wa should be written ha in 1955:
助詞のはは,はと書くことを本則とする (an actual law)
I don't really have a point, but I now know a lot about kana usage.
ISO 3602 makes a lot of sense to me because it is unambiguous in principal, but also allows phonetic spelling when necessary.
With the number of variations of romaji I have seen, even in official usage, it's a wonder why they bother.
I have found Japan to be a place that makes a lot of rules, then makes following them optional (then saying "there is no excuse" when you break them and get caught.)
You also might make apparently strange choices so that the romanisation reflects certain relationships between phonemes or within the language's native writing system. For example, the nihon-siki and kunrei-siki romanisations of Japanese render し and じ as 'si' and 'zi', even though they'd be pronounced more like 'shi' and 'ji', because し is usually grouped with さ ('sa'), す ('su'), せ ('se') and そ ('so'), and じ is grouped with ざ ('za'), ず ('zu'), ぜ ('ze'), and ぞ ('zo'). The Hepburn romanisation of Japanese, on the other hand, renders them as 'shi' and 'ji' for better pronunciation by English speakers, but this obscures the relationship to the other syllables. It also introduces an ambiguity when transliterating back into Japanese's own script, as there are actually two other symbols in the same family which can be used to write these same sounds, which Hepburn can't romanise differently as it needs to present clear pronunciation.
Pinyin provides another example. It has the three groups of initials (leading consonants) z-, c- and s-; zh-, ch- and sh-; and j-, q- and x-. An English speaker would pronounce none of these correctly, and they don't match any European language. But there's an internal consistency here within Pinyin: the first group are alveolar consonants, the second retroflex, the third alveo-palatal, and the first initial in each group is an unaspirated affricate, the second an aspirated affricate, the third a fricative. It's providing useful information which is lost if you try to make it so it's pronounced correctly by English speakers - which is rather fruitless given there's some phonetic distinctions in there which don't exist in English.