Customized solutions for industry, real-world research situations for academia. University/industry partnerships seem like a logical and mutually beneficial relationship. In our 12 years of research experience the n-dim group has found that these partnerships can indeed work well in specific types of situations where the nature of the partnership and the expectations of the relationship are clearly understood. Time must be given to building the relationship, ensuring that both sides understand the problem to be solved and that both sides understand each others expectations.
In partnerships to create design support tools we have used the method diagrammed below, which is essentially a cyclical evolutionary prototyping method:
Further explanation of these ideas can be found in the research report Software doesn't transfer, people do that was written to capture the experiences reported and conclusions reached in a workshop held at the EDRC in 1993.