Costs Could Be Driving Factor With Driverless Adoption

For all the brouhaha surrounding the dangers of Level 3 and Level 4 autonomous driving, few have questioned the dangers of Level 5.

It is assumed that once we get there – once the car is capable of handling itself in a multitude of situations – that it will be the safest automobile ever made. In reality, there may be just as many risks in eliminating steering wheels as there are in allowing them to remain. “There are a few scenarios in cases where you might have to hand over to a driver immediately,” said Niall Berkery, AImotive’s senior vice-president of North America. “One of my contentions is, when you encounter an emergency situation, who do you hand it over to if the car has no one to hand it over to?”

It is that very issue that could ultimately delay the release of fully autonomous vehicles. “For example, with GM’s Super Cruise today, there’s a lot of improvements to the user experience that could be made there,” said Berkery. “They’ve pre-mapped the road and it can drive on many of the highways but at certain times, for whatever reason, there are issues. Maybe the road markings are the issue. Or typically, as you come up to an interchange where one highway might cross over to another highway, the system will disengage. It doesn’t give you a heads-up, even though it has a map on-board and it knows when it’s going to disengage.”

Berkery added that autonomous vehicles should provide a proper warning system that is similar to that of a navigation device. In this case he suggested the vehicle should say, “In one mile from now, your system will disengage.” He said this is an “obvious user experience improvement” that can and will be made at some point in the future but it hasn’t happened yet. Despite the progress that has been made, autonomous vehicles are still in their infancy.

“Most of the scenarios will fall in the bucket of either it’s known – something that is known in advance of it occurring – or weather,” Berkery continued. “Knowing that the weather will deteriorate, that there’s incoming snow, potentially whiteout conditions. A lot of the sensors on the vehicle won’t work optimally.”

Human drivers are more adventurous than driverless cars, however. They may realize that the snow-covered highways are dangerous and still feel the urge to hit the road. AVs are being designed to do the opposite and will either (A) hand the controls back to the human or (B) the vehicle won’t move at all.

Berkery presumes the transition will be handled gracefully in most scenarios but he said there is concern about driver awareness and alertness. If the person behind the wheel is heavily distracted or falls asleep, the vehicle may be forced to pull over. “A lot more work needs to be done on that but I don’t think that’s a roadblock to advancing down Level 3,” said Berkery. “We think the benefit to the consumer is going to be so much greater, having sophisticated highway driving and parking.”

Massive undertaking

Whether simulated or achieved with real cars on real roads, AVs will require a significant amount of miles to match the capabilities of human drivers. Achieving that might be much harder than the hype indicates, however. Berkery explained that while companies like Waymo started their AV projects many years ago, their progress has been extremely limited to a small number of geographic locations.

“Nine years later and they’re still really focused on a couple of small areas,” he said, referring to Waymo’s tests in Mountain View, California and Chandler, Arizona. “You’re talking about 10 million miles but these are maybe not the two largest cities. The big question is the ability to scale – how do you do it? Globally we have 200 countries and, depending on how you slice it, 4,000+ cities in the world.”

There has been an influx of start-ups, particularly those building different types of LiDAR, that are hoping to propel AVs toward deployment but this has led to a number of copycats that may not actually provide the technological advancements they claim to be striving for. “LiDAR, specifically, there’s a lot of innovation taking place right now but a lot of people are reinventing the same wheel,” said Berkery. “The solid state LiDAR, that’s the Holy Grail. Then it’s how many LiDAR are actually needed on a vehicle for absolute redundancy? Some of the cars we’ve seen have anywhere from in excess of seven to 12 LiDAR. Realistically, do you really need a LiDAR in your rearview? People are going to err on the side of redundancy and safety, so you’re going to have a lot more costs into the system.”

Those costs are currently unavoidable, which is why a $25,000 sedan can skyrocket to $200,000 when equipped with driverless technology. Despite these costs and hurdles, Waymo, automakers and others continue to say that deployment is imminent. “Dates have been set,” said Berkery. “Careers are on the line.”


Leave a comment

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