Video game development may jump to a new level thanks to the work of a University of British Columbia computer science graduate student.
Mike Vlad Cora’s research looked at ways to make video games learn and adapt their behaviours, so as to create a more challenging and entertaining gaming experience for players.
He accomplished all this, while also laying the foundation of a system that would save video game creators months of production time.
Cora’s research was done as part of Accelerate BC, an internship program that connects up-and-coming researchers in the province with BC companies for short-term, applied research projects.
Acclerate BC was established in February 2007 by MITACS, a research network based at Simon Fraser University, which seeks to link businesses across Canada with university expertise to solve key industry and societal challenges.
Cora’s project was focused on helping Vancouver game developer Next Level Games apply artificial intelligence (AI) to the company’s games.
Next Level Games is a third-party video game development house, providing software to various game publishers. Their customers include Activision and EA.
Cora developed the foundation of a system that would automatically learn to do things like selecting the best animation to play in response to certain player movements. Think of a hockey goalie responding to an approaching player with a different angle for each shot.
To achieve this effect, he took some advanced AI algorithms and experimented with ways of attaching them to existing game code so that they could automatically control a wide array of behavioural parameters.
This sort of work is usually done manually, often taking developers months of fine tuning to reach levels of realistic action that makes game play satisfying.
“This can be a very time consuming process,” said Cora. “So I tried to see how we can speed up this process by taking some of that manual work and make it automatic.”
Edoardo de Martin, Studio General Manager at Next Level Games believes the innovation will allow the company’s creative people to spend more time focusing on the quality and experience of the game, rather than getting animations to queue up properly.
“We figure it could save up to six months production time. That’s a huge time savings for us.”
But Cora also believes his research has wider implications, and one of his goals was to simplify the AI algorithms so they could be used in many other applications.
Still in a gaming context, he says “if the AI adapts to different players’ styles then you’ll always have a new experience when you play it, and much more variety in the game.”
Such benefits are not uncommon, although they are usually difficult to quantify.
The Conference Board of Canada has been tracking public-private research collaboration for some time and concludes that they produce a range of benefits that go far beyond the specific goals of the project.
According to R. Sandra Schillo, principal research associate at the Conference Board these benefits include: the real-world experience student researchers gain, new programs and courses universities develop, and the expanded knowledge base that research provides.
And Cora’s own research – and breakthrough – could be added to that list.
“We’re not able to do a lot of research in our industry,” he explained. “When we’re working on a project, we have to work to a deadline, and the publisher doesn’t really want to pay us to do research that might not work. They want us to work on projects that have a definite outcome.”
But he said when researching the “academic side of things”, he could “just explore all the interesting bits that people have been working on. That was very useful to me, and fun.”
The chance to participate in some pure research was one of the key benefits of the collaboration according to de Martin.
Project participants, he said, took gaming concepts out of the video gaming world and developed them in a more generally applicable way, creating a theoretical framework that could apply across industries.
“We could then take that theoretical framework, optimize it, and customize it, for the particular game genre we were working with.
A second major benefit for the company was the investment in human capital Cora’s research and learning represented. He stayed on at Next Level Games to continue develop AI applications for gaming.
The positive experience of working with MITACS has whetted Next Level Games’ appetite for further collaboration, and students working on future projects could also become future employees.
“It’s great for us as a business to recognize that academia has this purpose,” said de Martin, “and for us to reconnect with it. The experience allowed us to step back from daily business concerns and take a fresh look at the way we do things. And that’s always a good thing.”
