Google (NASDAQ:GOOG) (NASDAQ:GOOGL) recently reported that its driverless cars were involved in 11 "minor" accidents over 1.7 million miles of testing during the past six years.The company added that "not once was the self-driving car the cause of the accident."
Three of those accidents occurred since California started an official test program for autonomous vehicles last September. A fourth accident, involving a vehicle using Delphi Automotive's driverless technology, occurred last October. Google and Delphi claim that all those accidents were caused by human error, and not by problems with the underlying technology. Will these reports delay Google's efforts to put autonomous vehicles on roads?
Are driverless cars really safer?
Eleven minor accidents over six years doesn't sound statistically significant. But based on the national rate of "property damage only" crashes of 0.3 per 100,000 miles driven, as reported by The Associated Press, Google's rate of three crashes per 140,000 miles driven seems relatively high.
Chris Urmson, the director of Google's driverless program, stated in a blog post that "we’ve been hit from behind seven times, mainly at traffic lights but also on the freeway. We’ve also been side-swiped a couple of times and hit by a car rolling through a stop sign." The majority of those accidents happened on city streets instead of highways. Nonetheless, California is requiring all companies testing driverless cars to post a $5 million bond against the possibility of their vehicles causing injuries, deaths, and property damage.
Even the smallest accidents must be reported, regardless of whether the autonomous vehicle was being driven by a human or computer. Google and Volkswagen (NASDAQOTH:VWAGY) have proposed that the rule be changed to only include accidents that were directly caused by the computer. Google also points out that many minor accidents with regular cars are never reported to the police and thus are not included in official statistics.
Is Google getting ahead of itself?
Google, Daimler's Mercedes-Benz, Hyundai, General Motors and other companies have all claimed that driverless cars could hit roads by 2020. However, these companies each have slightly different expectations of just how autonomous vehicles will be.
Major automakers like Mercedes, BMW, and Toyota's Lexus already offer radar-guided cruise control systems that keep its cars at safe trailing distances or brake to a complete stop to avoid accidents. Ninety percent of leading automakers obtain these "crash avoidance systems" from Israeli company Mobileye (NASDAQOTH:MBBYF). Mobileye is also developing a semi-autonomous system that keeps a vehicle in a single lane.
Mobileye's crash avoidance system only costs about $1,000, since it uses a combination of older cameras and radars. By comparison, Google's cars -- which use rotating LIDARs to create 3D models of their surroundings -- are loaded with up to $250,000 in equipment. Google maps out roads in high detail to assist the 3D rendering process, but its prototype cars still have a max speed of 25 mph.
Over time, costs will inevitably drop and max speeds will rise, but it seems like Mobileye's semi-autonomous approach is currently the more practical one for automakers, drivers, and governments.
Google's long-term goal isn't to sell cars. Speaking at the recent Code Conference, Chief Business Officer Omid Kordestani admitted that Google would probably never sell its own cars. Instead, it will likely license out its driverless technology to automakers through partnerships.
With Android Auto, its mobile mirroring solution for car dashboards, Google already gathers information about a user's travel habits. With the ability to enable companies to ferry passengers between areas without drivers in the future, Google could turn roads into a smart transportation grid. This would help Google Maps extend its lead over rivals like Apple Maps and Nokia's HERE Maps.
By doing so, Google could eventually reach beyond automakers and start touching other industries. For example, taxi and bus companies could license Google's technology to replace human drivers or demote them into "machine monitors." Retailers and parcel services could similarly license the technology for delivery vehicles. Payments for those transportation or delivery services could also be collected via Android Pay.
The key takeaways
In my opinion, Google's accidents with driverless cars clearly highlight human carelessness instead of flaws with computers. However, fully autonomous cars would likely only be ideal in a world where everyone rides in one. Unless that happens, Google's driverless car project will remain a "moon shot." It's unlikely that fully autonomous cars will be everywhere by 2020, but semi-autonomous solutions could make driving a lot safer over the next few years.