Calgary Research on Vision-Based Navigation for Drones

Dr. Mozhdeh Shahbazi, professor of geomatics engineering at the University of Calgary's Schulich School of Engineering is working on developing autonomous navigation technology for unmanned aerial vehicles (UAVs.) She anticipates the technology will be used in a next-generation breed of 'intelligent' drones that will aid in everything from search and rescue to aerial evidence collection for law enforcement. Bryan Passifiume/Postmedia Network

Dr. Mozhdeh Shahbazi, professor of geomatics engineering at the University of Calgary’s Schulich School of Engineering, is working on developing autonomous navigation technology for unmanned aerial vehicles (UAVs.) She anticipates the technology will be used in a next-generation breed of ‘intelligent’ drones that will aid in everything from search and rescue to aerial evidence collection for law enforcement.

Dr. Shahbazi has made developing a safe and workable UAV platform — one that operates autonomously without reliance on GPS — the focus of her academic career.

Shahbazi’s ambition is to create flying machines free from the constraints of GPS, which must ‘see’ satellites in the sky to fix their location on Earth.

“When you go to cavernous areas of downtown, for example, with its high-rise buildings, or when you go to forests when you’re surrounded by trees, then you completely lose the GPS signal,” she said.

In other words, when that GPS fix is lost — so goes the craft’s ability to navigate.

Instead, visual sensors and on-board processors allow her drones to not only form a complete awareness of their surroundings, but also interact with their environment intelligently.

“Vision-based navigation is just like in humans,” she explains.

“The way we position and navigate ourselves is that we can see the environment, or even touch it.”

Shahbazi’s system works in the same way, learning from its environment and knowing the difference between an open door and a closed one.

Using optical, laser and infrared sensors, the drone scans its environment from different angles — mimicking how our two eyes give us perception of depth by presenting two overlapping, slightly offset images.

This allows the UAV to navigate completely on its own, completing whatever task it’s assigned without human control or programming.

Shahbazi sees no limits what those tasks may be: three-dimensional mapping, structural evaluations, and even search-and-rescue or evidence collection for law enforcement.

Those are uses that particularly excites Shahbazi and her team, who see their autonomous drones of particular use in search-and-rescue, using thermal imaging cameras to quickly locate lost or stranded people.

Adapting current laser mapping used by forensic analysts could result in police having drones zipping around over scenes, collecting evidence many times faster that current methods — as fast as 20 to 30 minutes.

Because these units are capable, and designed, to fly out of the direct sight of their operator, the only place her team is legally permitted to test them outdoors is the Canadian Centre for Unmanned Vehicle Systems drone range near Foremost, Alta. — more than 300 km southeast of Calgary.

While Transport Canada has rules about autonomous aircraft operating beyond line-of-sight, she said provisions do exist for developers that can prove their technology is bulletproof in terms of obstacle avoidance and crash prevention.

Shahbazi’s work has garnered interest from both industry and government, and she hopes her advances in autonomy will eventually lead to similar advances in intelligence in her flying robots.

“Sooner or later, they will approve this,” she said.

“That’s what we’re looking for.”

Photo: Bryan Passifiume/Postmedia Network

Source: Edson Leader

Leave a Reply

Your email address will not be published. Required fields are marked *