R&D can be misleadingly glamorous when you first hear about it. Here's a vague glimpse at how mathematics might be used in industry:
http://www.maa.org/mathematics-and-operations-research-in-industry
At some point you will need credentials to get into either industry or academia. A simple B.S. degree in Math, Physics, or Computer Science will suffice to get you to the next level (provided that you work hard and solve problems well). I would use the time during your B.S. degree to get research experience or internship experience, starting with your 2nd year. You will usually need to become acquainted with your professors before you can get involved in their research, which means taking classes in the subjects that you are interested in and doing well in them. Take as many classes as possible in every subject that interests you; this way you will be able to refine your interests.
Keep these four questions in mind during your undergraduate experience:
1) how can I get research and/or internship experience?
2) who will write me the most effective letters of recommendation? how will I earn their respect?
3) how can I keep my grades at a top level?
4) what tests or credentials are required upon graduation in order to be admitted to grad school or to get a job?
After your bachelors degree, the next level is either an entry level position at a company (where you likely will not use very much math), or a graduate degree in one of the three subjects listed above. After either an MS or a PhD, you will have the credentials to get into a job where you might use mathematics and/or physics, though it probably will be on a very specific subject with a very specific skill-set. If you maintain a wide variety of math/physics interests, a professorship will likely allow you to study the interesting things; however, an industry job will allow you to make more money.
SpanishOmelette said:
Jsut as a side-note, where does dynamics of non-linear systems (I.E. Chaos) fit in? M. Physics or A. Mathematics?
Nonlinear dynamics is a broad subject that is difficult to penetrate deeply. At the forefront of modern research, nonlinear dynamics can be incorporated into just about every physics subject. It requires a very strong background in physics, calculus, statistics, differential equations, and numerical methods/computation.