"How Can Autonomous Vehicles Learn to Navigate Snowy Conditions? Ottawa Company Shares Their Expertise"

Fahed Hassanat and his team at Ottawa-based Sensor Cortek are tackling one of Canada's significant challenges for autonomous vehicle adoption: snow. They have developed a solution by outfitting cars with sensors covering them from bumper to rooftop. This innovative approach aims to enable autonomous vehicles to navigate safely even in snowy conditions. By addressing this obstacle, Sensor Cortek is contributing to the advancement of autonomous vehicle technology, potentially revolutionizing transportation in snow-prone regions like Canada.

When a major snowstorm hits Ottawa, most residents retreat indoors, grumbling about the weather and the impending task of shoveling snow. However, for Fahed Hassanat and his team at Sensor Cortek, a significant snowfall is a reason for excitement.

"The worse the conditions, the better," remarked the Ottawa-based software company's chief operating officer and head of engineering.

"Nowadays, heavy snowfalls are less frequent than before, so whenever they occur, we spring into action."

Drivers for Sensor Cortek take to the roads in cars equipped with sensors covering them from bumper to rooftop. Their mission: to tackle one of Canada's biggest hurdles to autonomous vehicle adoption—snow.

Snow presents a challenge for sensors, often obscuring and confusing them in adverse weather conditions, which complicates the training of self-driving vehicle software and algorithms. But it's not just snow that poses difficulties.

"You have snow, rain, fog, dust," Hassanat noted.

To prepare vehicles to handle whatever weather conditions arise, Sensor Cortek equips them with a suite of technology, including laser imaging, detection and ranging (Lidar) sensors, radar, cameras, and advanced GPS systems.

Lidar sensors emit laser beams and capture reflections from the environment to create 3D point clouds, offering information about the location of objects. Radar, on the other hand, utilizes electromagnetic waves to gather data about surroundings.

"Radar is a very complex sensor...but its advantage is its ability to function even when covered in mud," Hassanat explained. "It can operate in any weather condition."

Lidar and cameras rely on line of sight, making them susceptible to obstructions between the sensor and the object it needs to detect. On the other hand, radar doesn't require direct visibility to detect objects, making it more versatile in challenging weather conditions.

The amount of data generated by Sensor Cortek's sensors is substantial, with approximately 10 gigabytes of data produced every minute during operation. This testing often occurs at Area X.O, a private 1,850-acre site in Ottawa featuring a 16-kilometer track, where various equipment, including farming machinery, military tanks, and emergency vehicles, can be evaluated.

The effort to prepare driverless cars to handle any weather is crucial, as even simple driving maneuvers can become complex and hazardous in adverse conditions, according to William Melek, director of the University of Waterloo's RoboHub.

Sensor Cortek's driving experiences vary greatly, encountering different numbers of pedestrians, varied movement patterns, fluctuating speeds, and changing weather conditions with each drive. Fahed Hassanat, from Sensor Cortek, has encountered situations where sensors return with snow accumulation or have been affected by winds, blowing snow, or debris, impacting sight lines.

Following each drive, Sensor Cortek utilizes the collected data to train artificial neural network models. These models are then combined with the sensors to detect road users in all weather conditions, empowering vehicles to make safer decisions. This iterative process is essential for enhancing the reliability and effectiveness of autonomous driving systems.

The ultimate goal for Sensor Cortek is to develop deep neural network and AI-based perception systems that significantly enhance the ability of autonomous vehicles, as well as other sensor-dependent technologies, to perceive their surroundings effectively in all visibility conditions, thereby ensuring safe operation.

The challenges posed by extreme weather conditions for autonomous vehicles have been acknowledged by major auto and tech companies. For instance, Alphabet Inc., the parent company of Google, faced difficulties with snow as far back as 2015, as revealed by Chris Urmson, the director of the company's self-driving car project at the time.

Since then, several driverless car projects have been abandoned for various reasons, including cost-benefit analyses and safety concerns. For example, Ford Motor Co. and Volkswagen AG decided to abandon their autonomous vehicle venture, Argo AI, citing doubts about its profitability. Uber Technologies Inc. also sold its self-driving car division to the startup Aurora in 2020 following a fatal incident involving one of its test vehicles in Arizona.

William Melek acknowledges that in the short term, companies persevering in this field may achieve some level of self-driving capabilities with human intervention. However, he believes that it will likely take many more years before fully autonomous vehicles are ready for widespread deployment, and it may also require significant efforts to build trust among Canadians and other potential users.

Melek's personal assessment suggests that fully autonomous vehicles could still be 15 to 20 years away from becoming a reality, reflecting the complexity of the technological, regulatory, and societal challenges involved in their development and adoption.