Determining Fault in an Autonomous Vehicle Crash: The ChallengesJanuary 18, 2018
By: Chris Moulder, CPCU, ARM – Vice President, Broker, Worldwide Facilities – Atlanta
When fully automated cars become commercially available, it is likely that the majority of blame will shift from drivers to manufacturers in case of an accident. However, when considering the cars on the market today—many of which have some level of partial-automation technology—the picture is less clear.
Advanced Driver Assistance Systems (ADAS) technology is taking up more of the tasks of driving, but it is not designed to be completely independent. Determining fault in an accident often involves unraveling the complex interactions between human drivers, ADAS systems, and software.
Here are a few things that complicate the task of determining fault in an accident—even with cars that are nowhere near completely autonomous.
Today’s semi-autonomous vehicles depend on a complex network of cameras, radar, computers, and software to sense and respond to the environment. The software underpinning the system can be hacked.
This has been demonstrated to dramatic effect a number of times, starting as early as 2013, when researchers hacked existing automated features in a Toyota Prius, a Ford Escape, and a Jeep Cherokee. They found that, for instance, a Jeep traveling 80 miles per hour could be forced to try parking itself, and a Prius’ collision avoidance system could be hacked to slam on the brakes.
This is already a potential problem with autonomous features on the road today. But with increasingly self-driving technology, computers will control more of the vehicle’s functions—putting passengers even more at the mercy of hackers.
Difficult environmental conditions
Today’s partial-automation technology is generally designed for optimal road and weather conditions, and tested in carefully controlled settings. However, the game changes when these cars are let loose on roads that are not as well-maintained as they could be.
Unfortunately, according to the national transportation research organization TRIP, approximately 69% of the urban roads in the United States are considered in “poor” or “mediocre” condition. This means washboard, faded lane markers, buckled asphalt, potholes, and other challenges—many of which autonomous technology has difficulty handling.
Challenges the sensors weren’t designed for
Even though camera systems on today’s partially-autonomous cars often have a 360-degree range of sight, they also have some blind spots. For instance, the cameras on many cars can’t “see” a construction worker directing traffic, sense when a traffic light is red or green under certain conditions, or detect when a truck needs more room to turn. They have trouble seeing painted lines on the road if they’re faded or covered in snow. And software still has trouble predicting and responding to unpredictable elements, such as bicyclists, motorcycles, and pedestrians. Research and Development operations, and OEMS, are continuing to hone the ability to respond to these elements, but work still needs to be done.
Negligence or tampering by the driver
When different software and hardware takes on some of the tasks of driving, it becomes increasingly important to get those systems serviced in a timely manner. If drivers don’t maintain their vehicles appropriately—and we all fall behind on this—can they be charged with negligence if a crash occurs? How demanding can insurers and manufacturers be in expecting drivers to service their vehicles on time? If software updates are wirelessly pushed to vehicles owners, what happens if a critical update is missed for any variety of reasons?
In addition, it’s possible that owners would try to hack the software in their own cars—to remove or change safety features, for instance, to get a different performance out of the car. If this happens, how culpable is an owner in the event of a crash?
No matter how well designed, no autonomous vehicle will be crash-proof. Cars don’t need to be completely self-driving for new technology to change the risk scenario—and insurance agents will need expert help in navigating the changing needs of auto manufacturers, both today and in the future.
For more information, contact Chris Moulder at email@example.com or (678) 736-6723. Read and download Chris’s white paper “Autonomous Vehicle Insurance: At the Intersection of Manufacturer & Individual Risk.”