In short, as a starting learner of Japanese I acknowledge that the problem you describe exists but think that 'really bad' is an overstatement and it is almost inevitable that your brain associates your old knowledge of English (or other languages) to the new language.
No you don't. I type in zhuyin every day. The sounds line up in order of the alphabet, starting with ㄅㄆㄇㄈ, in the left column. It's actually easier than touch typing English.
It would be like if English speakers didn't use "qwerty" keyboards but instead used "abcde" keyboards. Touch typing would be far easier to learn. My computer doesn't have zhuyin printed on the keyboard, since I bought it in California, but I type zhuyin on it every day. Even before I could touch type, it was easy because it was in order.
In college I was taught Pinyin and we were given instruction on the correct pronunciation of each phonetic sound, positioning of the tongue etc., and the teacher spent time drilling us on that.
What matters is good instruction. Using Romanization doesn't encourage more correct pronunciation. Roman and Cyrillic alphabets are used for all manner of languages with completely different phonologies.
Learning IPA on top is an extreme but still useful third point of attack and again, it's trivial compared to the difficulty of actually learning to hear the sounds clearly. It must have been a year before I realized how different the d in pinyin and the d in English were.
Don't get me wrong, Romaji has useful purposes, such as representing place names in Japan for foreigners, etc. But as a learning device for learning Japanese, it's not a very good one.