SAN JOSE, Calif. Prith Banerjee knows what it is like to turn a big university research project into a commercial success. As the new director of HP Labs he hopes to help sort through a forest of R&D projects at Hewlett-Packard Co. to replicate that experience for some of them.
"I have been meeting with researchers in their cubicles and one of the things I discovered is that many of them are working on pet projects with one to three people that have been going on a long time," said Banerjee who joined HP August 1 from his position as dean of the college of engineering at the University of Illinois at Chicago.
"Even if you are 100 percent successful such projects will not result in commercial products. Such small projects are appropriate in academia but in the corporate world you need to take things to the next step," he said.
For HP Labs that next step under Banerjee will involve lots of collaboration and focus.
Today HP Labs' 600 researchers are engaged on as many as 150 generally small projects. Over the next five years Banerjee hopes to pare that back to as few as 30 larger efforts that have a bigger impact on computing research and products at HP. He expects to do that without planned cuts in staff or increases in the group's $150 million annual budget.
"We need to reinvest the resources we have," he said.
So Banarjee has set up a review committee made up of researchers from the labs and HP's business units. Each year it will weed out one-fifth of the group's least promising projects and re-distribute their budgets to just six new projects with average budgets of $5 million each and a three-five year time horizon.
"Over time the lab will be transformed to working on about 30 projects, each headed by a dynamic leader and focused on achieving a specific goal. In this way we will refocus our energy on large collaborative projects with potentially big impact," he said.
A separate committee of 80 HP Labs researchers and 40 technologists from HP business units is drafting a five to ten year strategic plan for the labs. Another smaller group will help pick new projects and technology areas for the labs to focus on as it tries to make fewer, larger bets.
"Before jumping into new problems we want to gauge how big the opportunities are," said Banarjee.
About a third of the work a HP Labs will focus on basic research often in collaboration with universities. HP will publish its list of project priorities to universities and actively recruit leading academic researchers to help wok on them, turning the usual model on its head.
"Typically academics will write proposals and corporations will fund them as a sort of philanthropy," said Banarjee who spent 20 years in academia at the University of Illinois and Northwestern University.
Another third of the projects will focus on applied technology, and the final third will be development projects. A major goal is getting the results of the projects into real products at HP, something that does not happen often enough in corporate settings, Banarjee said.
"I have been seeing for some time in academia where business units are too focused on short term deadlines to catch the new technology," he said.
"I found what works best is to transfer the people with the technology," Banarjee said. "So when a project is done we will create a team of 5-10 researchers working with 5-20 researchers from a business unit, all focused on creating new products, exactly like a startup," he added.
Banarjee experienced that transfer when he took time off from the University of Illinois to found Accelchip in 2000, a developer of DSP software tools.
"It started as a large collaborative university research project with DARPA funding," he recalled. "When it was finished I pursued the concept with some venture capitalists and was able to form a company and help sell its first products.
"Eventually we hired a professional CEO and the company was sold to Xilinx. It was an example of a successful research project that eventual spawned products that wound up in the hands of more than 20,000 customers," he said.
At the time of the acquisition in early 2006, sources estimated the sale at $21.5 million.
Banerjee's second startup, software tool maker Binachip where he still serves as chairman and chief scientist, is seeking a similar exit by acquisition.
There are a variety of ways Banarjee will measure the success of his efforts to increase the impact of HP Labs.
"In basic research you measure that by publishing in prestigious periodicals and conferences. In applied research you measure it by patent submissions and awards, and in development you measure it by successful technology transfer into new products," he said.
But first, Banarjee is off on a two-week trip to visit five overseas offices of HP Labs. His effort to pare back and create more large, high-impact projects starts in earnest in January.