Steven Waslander looks forward to his drives becoming a lot easier.
“Driving on highways and in cities is dull, time-consuming, and can be very stressful,” says Waslander, director of the Waterloo Autonomous Vehicles Laboratory (WAVELab). “I would be just as happy gazing out the window while the car does the dirty work for me.”
The push toward driverless vehicles is happening in corporations, too. Here in Ontario, University of Waterloo researchers are collaborating with BlackBerry’s QNX division and companies including Renesas Electronics and Denso to advance autonomous driving research. That includes testing an autonomous vehicle nicknamed Autonomoose — with a safety driver behind the wheel, of course.
Waslander is not expecting entirely driverless vehicles soon. “The failure rates are still too high,” especially in bad weather. But it is possible to reduce accident rates by combining the advances of an automated system with the abilities of a human to understand complex situations, he says.
His team is making autonomous vehicles more reliable and safer, bringing cars that can do the driving for you closer to reality.
Machines that can mimic the cognitive abilities of humans are arriving.
Smartphones with personal digital assistants, customer service chatbots and online ads use what is known as “artificial intelligence.” Microsoft recently bought Maluuba, founded by two University of Waterloo alumni. It is an artificial intelligence company working on natural language processing.
A major approach to artificial intelligence is deep neural networks, powerful software algorithms that are composed of the equivalent of “neurons,” connected by “synapses,” layered so that at every level, more abstract information is being modelled, says Alexander Wong, a Canada Research Chair in Medical Imaging Systems at Waterloo, and one of the co-inventors of evolutionary deep intelligence for enabling operational artificial intelligence under everyday scenarios.
Deep neural networks allow the machine to observe and learn from a wealth of data, rather than just following instructions. These machines become proficient at tasks such as knowing where a photo was taken, or being able to recognize different languages.
“You can say the software writes itself based on the data and task at hand, which makes it ideal for doing things where even humans do not know the exact rules or steps to perform the task,” Wong says.
The power is remarkable. Wong and his team have introduced discovery radiomics, where the artificial intelligence can discover biomarkers from medical imaging data and help spot tumours faster. One study showed it could accurately detect lung cancer, 91 per cent of the time. Their evolutionary deep intelligence yielded deep neural networks up to 100 times smaller for operational performance.
In the next 10 years, Wong expects that many devices and appliances might also be imbued with artificial intelligence. An “Internet of Intelligent Things” will augment human intelligence and capabilities like never before, Wong says.
“The recent advances in AI, in my opinion, are nothing short of revolutionary.”
The Internet of Things
The dawn of the Internet of Things (IoT) era is lighting the way to smarter cities.
Be it the visual or sensor data generated at a traffic intersection, a pumping station or a bus, Internet of Things (IoT) technology “wakes all of that data up,” says Kurtis McBride, of Miovision, a company that he and two of his University of Waterloo classmates co-founded to manage traffic intersections using cellular networks and cloud computing.
“It will help us solve all sorts of problems, whether it is congestion, energy consumption or how we distribute water through our cities.”
Miovision will soon move into Catalyst137, a former tire plant on Glasgow Street in the heart of Kitchener. Developed with Frank Voisin from Voisin Capital, it will be a catalyst for IoT and device-driven software companies. The building and streetscape itself will also have hundreds of sensors to make it hackable and a showcase for the region’s technical ingenuity.
Strides are also being made in connected wearable devices.
The most obvious applications are in gaming, using such devices as virtual reality glasses. But John Zelek, a systems design engineering researcher at the University, explores how wearable devices might help elderly or disabled people overcome their challenges. He works on navigation devices for people who are blind and people with Alzheimer’s disease.
“From my experience, technology will be adopted if it is easy to use, invisible, inexpensive and works well. But that is easier said than done,” Zelek says.
Watching 3D printers print objects layer by layer is a glimpse into the future. After all, if we can build plastic toys or airplane parts that way, why not knee joints or organs?
Mihaela Vlasea, a University of Waterloo researcher in mechanical and mechatronics engineering, is helping to make that future a reality.
During her postdoctoral work, Vlasea helped to develop 3D printer technology that creates a bone-like substitute for joint implants that could replace metal and polymers that tend to wear down in the fluid environment of the body. That work, a collaboration of the University of Waterloo, University of Toronto and Mount Sinai Hospital, “is now quite mature in terms of clinical trial readiness,” Vlasea says.
She and her team are working on making additive manufacturing machines better at using patient’s CT scan data to interpret and execute manufacturing and produce quality parts.
Factories of the future will also likely employ additive manufacturing alongside conventional manufacturing. Together with powerful computing algorithms, additive technologies will open the door to designing and manufacturing products that could not otherwise be made, Vlasea says.
Anxiety about the robotic future is fueled by dystopian science fiction such as Westworld or Terminator.
But University of Waterloo robotics researcher Dana Kulic isn’t worried just yet.
“So far, the most amazing thing we have seen is when a human designer poses a specific problem or question, there are algorithms that can answer that question really well,” she says. But in knowing what questions to ask, humans are still at the helm.
Nevertheless, new human-robot interactions are happening. A good example is the social and customer service robot, Pepper, already being commercialized by SoftBank Robotics.
Better sensors, actuation mechanisms like grippers, access to cloud-based computing and more powerful decision-making and planning algorithms are rapidly advancing the robotics field. But challenges remain, especially in abilities like walking or making decisions in rapidly changing, complex environments.
Artificial intelligence will be a key to enabling future robotics, Kulic says. “The next steps involve exploiting more of those algorithms on the action side, and in decision-making and control processes.”
Commercially, such companies as Clearpath Robotics, the brainchild of four University of Waterloo mechatronics engineering graduates, are already making robots for a wider range of industrial environments.
While a Rise of the Machines takeover is unlikely, Kulic says we can all expect more robot interaction in the future.
Quantum computers and cybersecurity
Quantum computers exploit the strange quantum mechanical properties of subatomic particles to do computing in a more powerful way. Instead of “bits” of zero or one, quantum bits (qubits) can be in mysterious combinations of both zero and one.
A fault-tolerant quantum computer that, in principle, is scalable could be at the stage of being experimentally demonstrated within four years, says Michele Mosca, co-founder of the Institute for Quantum Computing and a professor of mathematics at the University of Waterloo. That will lead to intense effort and investment into scaling it up.
“Once we have close to 100 logical quantum bits, we will be able to do interesting calculations that were previously not possible. Once we have a few thousand qubits, we will be able to break currently deployed public key cryptography,” he adds.
Mosca sees a significant chance of reaching the 1,000-qubit mark in 15 years. “The challenges are daunting, but brilliant minds have made tremendous advances already, and even more people and investments will continue to push the frontier quickly,” he says.
Opportunities and risks loom large. Quantum computers could be used to simulate and design new quantum materials that might be the energy solutions of tomorrow, or simulate how new drugs will interact in the human body. But the ability to break public key cryptography creates an unprecedented risk for governments, corporations, banks and public infrastructure.
It will take years to fully deploy and test the quantum safe tools, so “it is absolutely critical that we start preparing for this transition immediately,” Mosca says. That’s why he co-founded evolutionQ, a company dedicated to helping organizations manage their quantum risk and to deploy cyber tools designed to be safe against quantum computers.
Big data and medicine
Mathematical modelling of huge data sets, or big data, is already starting to reshape medicine and biological sciences.
The biomedical research group in Waterloo’s applied mathematics department, in particular researchers such as Siv Sivaloganathan, are using mathematics to better understand how diseases start and progress, so doctors can make better clinical decisions.
Applied mathematics can be used to gain better estimates of the radiotherapy dose needed to eradicate a tumour while inflicting minimal damage on normal cells. Mathematics can optimize the delivery of a drug through the skin or create more detailed models of the human brain.
Meanwhile, on the commercial side, Bioinformatics Solutions Inc. is a Waterloo spin-off company that is advancing drug discovery through professional proteomics software. Digital tools are also being commercialized for public health and medicine in companies such as Aterica and Intellijoint Surgical, both started by University of Waterloo alumni.
But in the future, with artificial intelligence, mathematical modelling will be even more powerful, delivering brand-new insights into how complex biological systems interact and evolve.