One of the world’s most storied corporate research labs–the lab that brought the world the pocket scientific calculator, the inkjet printer, the first commercial LED, and, more recently, experimental molecular-logic gates–just got a new director. And after only a few weeks on the job, he’s planning a major shake-up.

The new director is Prith Banerjee, a longtime academic researcher and a serial entrepreneur. He plans to cut most of the research projects under way at HP Labs, the IT giant’s research arm, and introduce changes that he hopes will improve the transfer of tech from the labs into products.

He’ll begin by paring down the number of projects from 150 to just 30 over the next five years, while bringing in new projects at the rate of five or six per year. The cutbacks won’t necessarily mean job cuts, as researchers will be shuffled into new groups after their pet projects disappear. Banerjee says that the regrouping is already happening: research groups are starting to form alliances, and researchers are mailing in proposals that they be transferred to specific projects. “People want to be aligned with the winners,” he observes.

His plan for improving tech transfer from the labs seems to revolve around the idea of establishing temporary organizations within HP that he calls “incubation labs” or “mini-startups”; they will be modeled on his experience starting two tech companies: Binachip and Accelchip. As an academic researcher, Banerjee had been frustrated that, even when he gave away his research software to established companies, the software didn’t make it to market. The companies didn’t seem to know what to do with it. So he decided to start his own companies. “In those startups,” he says, “I transferred the people–the graduate students–along with the technology, and brought in outside people who knew how to build products. And that made it successful.”

“Researchers are good at innovation, but they don’t know how to make products,” he explains. At the same time, “businesspeople know how to make products, but they’re incremental improvements to their current product line.” The secret, Banerjee says, is to bring them together. “When you take the best of both worlds–the innovation and the product expertise–you have an amazing combination.”

Here’s how he sees this working at HP. Research projects will last about five years before they’re dissolved, and the people and assets are redistributed to other groups (which sounds a lot like graduate studies). If, during that time, a project happens to make a promising prototype, this will trigger the formation of an incubation lab. For a period of 12 to 18 months, researchers from the project will be freed from their obligations to do research and publish. And select sales and marketing personnel from the business side will be freed from their obligations to keep current products selling well. The two groups will then sit down and figure out how to make the prototype a sellable product.

Overall, Banerjee wants to mimic the process of spinning a company out of an academic lab, except for the hassle of raising venture capital. Researchers will have the rewarding experience of seeing a project through, from beginning to product launch, and then they can pick up with another interesting research group at HP and start again. Banerjee hopes that this setup, along with new internship programs for putting PhD students in direct contact with HP researchers, will attract some of the best and brightest researchers, and ultimately make HP Labs more productive.