Torc’s autonomous trucks, like self-driving cars, are powered by a combination of algorithms, sensors, artificial intelligence, and so much more. Each individual piece plays a key role in operating our self-driving systems, but none are greater than the sum of their parts. These elements work in harmony to perform successful driving behaviors, especially where inclement weather is concerned.
How does weather affect self-driving vehicles?
Let’s think about our sensor suite as a “panel of experts” in their fields. Whether it’s research or execution, each expert is highly trained in their specialization. In any given scenario, they can problem solve and offer solutions with their years of knowledge behind them.
On this panel of autonomous trucking “experts”, we’d have a few different specialists: LiDAR, cameras, radar, and more. In an inclement weather situation, each would chime in with their unique perspective, but the decision would ultimately be made by the collective.
Let’s say our autonomous 18-wheeler is heading down a stretch of highway when a cloud of fog rolls in. As the fog overtakes the truck, the panel of self-driving sensors confers to figure out the vehicle’s next steps.
With high levels of fog, the visibility is low as the visual contrast isn’t significant enough for cameras to make a clear distinction between obstacle and air. Therefore, if the cameras on a self-driving vehicle were acting alone, they may decide that the best course of action is to slow down or pull over if the weather is significant enough.
At the same time, LiDAR, which works via bouncing laser beams off surrounding objects, might be sensing a wall of gray rather than a weather occurrence. This technology works so well at detecting its surroundings that it is thought of as a key piece in future lunar landings, but as with all technology, it can’t overcome physics and may draw some false positives and negatives from its surroundings. Acting alone, LiDAR may suggest slowing down so that it has an extended amount of time to deduce what’s happening around it.
However, radar has a different outlook altogether. Radar technology transmits radio waves, which aren’t affected by weather. Therefore, thick fog isn’t as much of a challenge. Radar’s drawback is that it only captures part of the image of the road around a self-driving 18-wheeler. As an expert panelist, radar may suggest that the truck is good to move forward slowly and with caution.
Before making a decision on what driving behaviors the truck should perform, the sensors work together to offer data on a given scenario. In our heavy fog scenario, each sensor said that visibility was low; radar tells us that it can confirm the fragments of images that LiDAR and cameras are able to provide, therefore giving us a complete picture. Each component plays off the other’s strengths and weaknesses, allowing the software to be the tie-breaker of our board of experts. When navigating significant weather, teamwork amongst these technologies is what moves self-driving vehicles from one stretch of highway to the next.