University developing artificial intelligence in planes with NBIF funding

FREDERICTON – If planes and cars could drive themselves and communicate with each other it would save time and human lives, says Howard Li.

The University of New Brunswick electrical engineer is working to develop artificial intelligence in drone planes and ATVs so they can map out the terrain below and around them.

“We need to figure out how to make multiple ground vehicles and multiple aerial vehicles work together because eventually they will be fully automated,” says Li.

He says this would be especially useful for military operations or fighting forest fires.

“We (would) have aerial vehicles to monitor the situation and we (would) have ground vehicles to deliver all the necessary equipment,” he says.

“(The planes) will be able to let the ground vehicles know if there is any danger around,” he says. “Even if we have roadside bombs we will not lose the lives of good people.”

Li’s research is one of 57 projects receiving funding from the New Brunswick Innovation Foundation to cover student research positions, the independent but government-funded organization will announce today.

With money provided in part by the provincial department of Post-Secondary Education, Training and Labour when the program launched in 2003, the foundation is spending $630,000 to fund 70 positions in the province.

“The funding enables them to train students to develop skills and experience that can be applied to the workplace,” the foundation’s chief executive says, adding it builds capacity in research labs across the province.

Li says it’s funding like this that makes his research possible and he’s only going to need more of it.

Since coming to UNB in 2007, Li’s research has cost more than $300,000 and he expects it to hit half-a-million dollars within a year or two.

Four range-censors to allow the his remote control planes and helicopters to read the terrain below cost $12,000 but one good enough for a military drone costs $100,000 each.

Commercial autopilot software for a “toy” plane costs $70,000 and many of the model aircrafts and ATVs available on the market are also quite expensive, so Li and his team of student researchers have been developing the hardware and software themselves.

Just a year ago many of the half-a-dozen models he is experimenting with were merely plans on paper.

Carl Thibault, a master’s degree student in electrical engineering working for Li, says the biggest challenge is setting up the communication architecture between the automated vehicles.

“So you can have essentially multiple robots in different places collecting different pictures and different data,” he says. “We want to increase autonomy so that you don’t have to have someone driving it from a remote station all the time, which is what current UAV technology is.”

Thibault says it’s cutting-edge research that expands on current unmanned vehicles, but all the technology already exists.

The unmanned aircrafts and ground vehicles already exist and work longer than a human can, but the software is what’s holding back full automation.

“The structure and architecture that brings all this together is what’s lacking and what we can bring together with this research,” he says.

Li says the research will also help day-to-day life easier for people sometime in the next 30 to 50 years.

“It’s very tedious to drive,” he says. “All our cars will drive by themselves and they will not collide with each other either.”

Thibault says doing the research is his “dream job.”

“I wanted to be really innovative and go out and do things that haven’t been done before,” he says.

Since 2003, the Foundation and PETL have awarded $6 million and leverage $21.6 million more to support 691 research assistantships in New Brunswick.