A San Francisco, California motorcyclist is suing General Motors, accusing one of their self driving cars of negligent driving. According to the suit, the motorcyclist says he was riding behind a self driving vehicle that changed lanes, merging left and leaving an open space ahead of the rider. The rider moved ahead into the space when the vehicle aborted the lane change and swerved back into the spot it had just left, hitting the rider and knocking him down causing neck and shoulder injuries and causing him to take disability leave from work.
GM's account of the crash was somewhat different:
As the Cruise AV was re-centering itself in the lane, a motorcycle that had just lane-split between two vehicles in the center and right lanes moved into the center lane, glanced the side of the Cruise AV, wobbled and fell over,” the report claims, noting that the Cruise AV was traveling with the flow of traffic at 12 mph and the motorcycle was traveling at approximately 17 mph.
Two rather different accounts, but, thankfully, a low speed collision, though still resulting in injuries to the biker. This is still very early in the transition to self driving cars and we're already seeing more accidents. As we've discussed before, the determination of who is at fault is going to be a legal bonanza for attorneys nationwide.
A driver was in the Cruise AV, but his hands were not on the wheel and before he could react, the accident happened.
Why do I get the feeling this is going to be the first of an endless stream of similar cases? One thing for sure, whenever you ride, assume you are invisible and always be ready for other vehicles to do the unexpected. No matter how a lawsuit turns out, when a motorcycle and car collide, the motorcycle always loses. Be careful out there.
Chuck Lantz says
This accident happened on Oak Street in San Francisco, a very busy and very hilly one-way street, with numerous stop-light controlled intersections. I’ve driven on Oak thousands of times. Using a self-driving vehicle on this street is simply insane.
And on the subject of all self-driving vehicles, one of my primary concerns is that there is obviously no human driver to be watched by others on the road. There’s no way to gauge what the car is about to do by checking what the driver is looking at, or better yet, NOT looking at. It’s like all tinted windows. Scary at best.
Paul Crowe says
Well, if they’re going to go on the street, eventually, they’ll have to be able to drive on all of the streets. At this point, they may not be ready.
Rod says
Presumably, the robocar has video or sensor data supporting it’s case. If not, then the motorcycle rider should prevail IMHO.
Paul Crowe says
You would think GM would have all sorts of data to support their case, but if the car saw the bike in its sensors or cameras, why did the collision happen? With only the two accounts to go on it’s impossible to form much of a conclusion other than the obvious one that acknowledges robo cars can have accidents. They may be very safe in traffic consisting of all self driving cars in constant communication with one another, but mixing robo cars and human driven vehicles could be a problem and that will be the inevitable mix for a long time to come.
Giolli Joker says
Yep, surely the car has all the recorded evidence, but if they are right yet the evidence shows that the technology is not where it is marketed to be, they may be willing to keep the data for themselves and prefer to just pay the damages to the motorcyclist. Chances are that the rider and his lawyer are very aware of this.
BC says
i suspect it’s all about ‘latency’ – as in ‘how fast is the reaction/motion initiated -vs- surroundings?’ The bike was prolly close… maybe a little too close… And then the GM car aborted _NOW_ and did a functional swerve, while it’s own self-scanning cameras expected slower surrounding data. A person would have _eased_ back into the lane
Paul Crowe says
Latency likely plays a part, but this might also shine light on all of the decisions the software makes about what to do in a given situation. Computers don’t decide anything, they follow the decision tree priorities the programmer wrote into the code. Avoid this, hit that, and who made those decisions before it was coded, probably not the programmer, he coded what was decided by others. Who made the decision? Save the child, hit the tree, sorts of decisions, but what about motorcycle or car versus car 1 or car 2? So many variables.
Josh says
Those cars are packed with cameras and sensors. The telemetry data should be pretty clear on what happened.
Paul Crowe says
A lot more drivers are installing dashcams to defend themselves in accidents and it’s beginning to look like handlebarcams might be a good idea, too.
Giolli Joker says
I just installed a front/rear dashcam in my car and I’m really happy about it. It is not waterproof though, not suitable for the bike. A look at what the market offers in this direction may be an interesting article.
todd says
Chances are the car “saw” the motorcycle but it was programmed to avoid the other car first and risk hitting the motorcycle. Therefore GM will settle and not release info suggesting that the car is programmed to hit motorcycles.