In 2009, Google launched its top secret autonomous vehicle project with the goal of delivering a driverless car by 2020. Over the last decade most major auto manufacturers have successfully incorporated many nifty Advanced Driver-Assistance Systems (ADAS), like parking assist, in their top-selling models.
But how close are we to seeing Google’s 2020 goal realized? Are we finally on the cusp of having that autonomous vehicle revolution we’ve all heard so much about?
Not quite. At this point, driverless cars have a long way to go before they can become the norm on our roadways. Recent challenges for self-driving automobiles point to the need for ongoing research and development. Over the last decade there have been major failures with the ADAS equipment, two fatal accidents, and additional challenges created by inclement weather conditions.
Each of these aspects of autonomous car development will need to be perfected before this potentially $7 Trillion business finally gets off the ground. Make no mistake, a multitude of companies are working day and night to figure out sound resolutions that will put fully autonomous cars into mass production.
In the meantime, the reality is that insurance rates will not be impacted anytime soon by the driverless car market.
Driving Without a Brain
The concept of a self-driving car is intriguing because it eliminates human error, as well as various unpredictable human behaviors like speeding or texting while behind the wheel. However, ADAS programming doesn’t currently have a feature that helps prepare the vehicle to respond when fellow human drivers act erratically on the road.
Human drivers often leave self-driving vehicles confused as to how they should react in traffic when humans go rogue. The fact that ADAS programming does not possesses any common sense has actually caused quite a bit of frustration.
For example, neighbors in the Phoenix, Arizona suburbs have expressed their concerns with Waymo’s driverless test vehicles. In fact, some citizens have said they “hate” the autonomous test cars and have even illegally driven around the vehicles.
Since 2017 Waymo (a subsidiary of Google’s parent company, Alphabet Inc.) has logged thousands of driverless miles throughout the country, including Arizona. Unfortunately, these test runs have not always gone smoothly. It’s been reported that self-driving autos stop abruptly in traffic when seeking an opportunity to merge or turn. Waymo self-driving cars also struggle while proceeding through busy intersections and frequently come to a complete stop as a result.
The below video is a perfect example of the Waymo vehicle’s struggle to understand and keep-up with human drivers.
Studies Confirm Driverless Cars Aren’t Yet 100% Safe
ADAS technologies, such as back-up cameras and radar, are used in self-driving vehicles, same as they are in “regular” models.
However, fully autonomous vehicles rely even more on ADAS since they can’t pass control to a human driver or signal a human driver to take the lead. Essentially the ADAS capabilities in an autonomous model are driving the car.
So if the ADAS in a driverless car is insufficient then the vehicle is in a lot of trouble. Based on the latest road tests, ADAS may have a long way to go before self-driving automobiles are 100% road ready.
Case in point: recent studies conducted by The Insurance Institute for Highway Safety (IIHS) and AAA both detail the clear failings of autonomous car applications.
The IIHS study focused on Adaptive Cruise Control (ACC) and active lane keeping sensors. The ACC is meant to maintain a consistent speed while active lane keeping ensures the vehicle stays in its current lane.
On the Society of Automotive Engineers (SAE) International Scale both features are categorized as only a Level 2 (partial automation) rather than the maximum Level 5 (full automation).
The IIHS results were quite surprising:
- A 2018 Tesla Model 3 slowed down without cause 12 times, including seven times when trees cast shadows along the road
- A 2017 Mercedes-Benz E Class did not reduce speed even though a stationary vehicle was directly in its path, until a human driver put-on the brakes
- During the active lane keeping test, three different vehicle models were unable to stay in their lane when taking on curves
- The same active lane keeping feature was tested on hills too. The BMW did not stay in its lane during all 14 test drives. The 2018 Volvo S9 stayed in the lane 9 out of 16 test drives. The Tesla Model S swerved back and forth, trying to find the correct lane.
AAA also tested four different vehicles equipped with ADAS applications and came up with similar results as IIHS.
The test vehicles used by AAA labored when encountering even moderate traffic, curved streets, and busy intersections. Almost 90% of driver intervention was needed because the car simply could not stay in its own lane. Plus, AAA’s study detailed multiple times when the vehicles did not adequately brake and abruptly changed speeds.
Driverless Cars Cause Fatal Accidents Too
One of the biggest expectations for the introduction of fully autonomous vehicles is the capability of saving lives. After all, most fatal accidents are a result of human failure. If this factor could be eliminated then, the logic follows, deadly crashes would become obsolete. Unfortunately, two fatal accidents involving self-driving vehicles set the driverless car industry back quite a few steps.
In 2016, a Florida man died while riding in a Tesla set to its Autopilot feature.
Last year an Uber autonomous car, moving at 40mph, struck and killed a pedestrian crossing the road.
Even though the Uber automobile was being driven under the speed limit and a safety driver was onboard, the fatal accident still happened.
The Weather Forecast for Driverless Cars Is Partly Cloudy
Another concern, albeit less pressing than some of the others reported, is how inclement weather negatively impacts the ADAS functions of a self-driving automobile.
Precipitation like snow and rain causes lidar laser lights to splinter. Cameras do not capture clear images when fog or snow occurs. GPS becomes spotty when the weather is poor. Radar isn’t able to clearly define obstacles in the roadway unless the skies are dry and clear.
To resolve these types of equipment issues, industry leaders are working diligently to get vehicle technology that will “think” logically.
The idea is to use special programming called “sensor fusion” which will assist the autonomous car to evaluate all application data and only use the best information to perform properly.
Of course, this bad weather solution may take quite a while since most autonomous car businesses are quite busy solving a long list of other problems.