Takashi Gomi

The idea first came to Takashi Gomi after hearing about a fatal landslide in Japan in 1994.

A dam had broken near the mountainous border of Niigata and Nagano prefectures, creating a landslide that swept away highway construction workers and their heavy equipment. Days later, 14 bodies were recovered from the rubble.

Mr. Gomi, president of Ottawa's Applied AI Systems, Inc., realized intelligent technology could have helped prevent the disaster.

Such situations had already prompted Japan's Ministry of Land, Infrastructure and Transport to install several hundred surveillance cameras along some of the country's most treacherous highways to monitor potential disaster areas, explains Mr. Gomi.

The cameras capture about 30 frames per second and send the images to staffed field offices. Images of several hundred locations appear on a handful of screens, which switch locations every five seconds.

However, disasters risk being caught only after they've occurred because the cameras don't behave intelligently, says Mr. Gomi, who worked at Bell Northern Research and Atomic Energy of Canada before establishing Applied AI on Carp Road in1983.

With more than $2 million in R&D funding, Applied AI has been working with the Japanese government to develop a solution. The company, with 15 staff, derives 95 per cent of its revenue from Japan.

In 1997, the firm proposed installing an "AI filter" in each of the cameras, so images could be captured automatically and viewed in real-time by a field officer.

For example, a falling rock could be a precursor to a larger disaster, says Mr. Gomi.

"Usually, the next day, something bigger happens."

Applied AI's efforts hit a snag last March, when it tested its system in Japan's rugged Gifu prefecture in the presence of several advanced technology personnel from all over the country. The demo was scheduled for 2 p.m. and Applied AI staff had readied a cardboard rock that could be rolled down a cliff face to test the quick wits of the intelligent camera installed nearby.

But just before the demo was to begin, a storm rolled in. Raindrops pelted the camera, followed by snow, hampering the device's ability to react.

"I couldn't believe it. It must have been God's will," Mr. Gomi says with a smile.

Other challenges remained, such as the camera's tendency to initiate false alarms after detecting non-threatening things such as deer, which frequently roam the hillsides. The cameras have difficulty distinguishing between real threats and false ones, he says.

It was back to the drawing board for the company, which hired two new computer engineers to work on the problems.

Despite the snags, Mr. Gomi has negotiated the installation of 10 intelligent cameras on Gifu's highways.

Mr. Gomi is optimistic about the market possibilities, predicting highway cameras will eventually cover every corner of the country, generating images that can be accessed on the Internet.

However, the real business challenge doesn't lie with the technology, he says. Bigger hurdles arise from Japan's system of doing business deals and brokering partnerships, something he describes as "autocratic and hierarchical".

"Unless you're in association with Toshiba or NEC – and I have great respect for these companies – things really don't work there like they do here. Three days ago, I shook hands with (March Networks CEO) Terry Matthews on the plane back from Italy and you can talk. But it's a little different in Japan."

Mr. Gomi would like to partner with Toshiba and has approached the company many times. Recently, his company increased its presence in Japan by setting up a small office in Mr. Gomi's native Tokyo. He has visited Toshiba's headquarters after accepting an invitation there.

"They were impressed with our technology, but it won't turn into a (partnership) right away. It may take another three to four years," he says. "It's a very long process."

In the meantime, the company is negotiating the installation of AI cameras along Japan's riverbanks to catch people illegally dumping garbage in the water.

Other projects include an intelligent wheelchair that can move autonomously. The company says the chair will eventually "know" its environment and the user can ask to be taken to particular locations.

Another project is a robotic assistant for farmers who pick fruit. The machine would move autonomously while the farmer collects the produce, easing the burden of carrying heavy fruit baskets.