I'm not sure if their case is entirely without merit. I haven't found the time yet to read their submission but it could be argued that the death was caused actively by the autopilot system. On his own the dozing driver would have hit the accident site at a rather sedate 19 km/h. It was the autopilot that accelerated the car, therefore it might be argued that it should have had better control features to check if it was safe to do so.
I think there are some problems with the case.
The first problem is it tries to argue that the driver monitoring was faulty because the steering wheel torque sensor isn't very good at this. That's an okay argument to make for a 2020 vehicle, but not for a 2016 model vehicle. There is also the issue that even the latest/greatest driver monitoring systems don't always detect the second someone is falling asleep or briefly not paying attention. Having one isn't a guarantee that it would have prevented the accident.
The second issue is it tries to argue about the cut-ins/outs being common, and while that's true it's not true that most L2 systems handle them well. The radar systems are often programmed to ignore stopped objects. Wired has a really good article on it. This means that sometimes these systems are blind to stopped vehicles.
The third issue with it is that it goes on and on about AP being beta technology. That's definitely true of AP2, but this was an AP1 vehicle as it was a 2016 Model. AP1 was built using MobileEye technology which was really proven out. Especially things like TACC which wasn't even really beta at the time.
It also mentions other accidents like the fatality accident involving a Model X in California. But, we now know the driver was playing a video game on his phone during at least part of the drive.
I think its okay to blame AP when it directly contributes to an accident in suddenly making some movement was so immediate, and abrupt that an average human driver wouldn't react in time. But, I'm not okay with the induced argument when it comes to liability in an accident.
To me this feel like it shifts the blame from the driver.
When I had a 2015 Model S with AP1 I really did feel like AP was causing me to lose situational awareness. I wasn't getting drowsy, but I noticed a lagged response to things like debris in the road. I also noticed that I wasn't as clued into vehicles around me. I largely stopped using AP because of this, and truck lust (AP1 liked to snuggle with semi-trailers). Having a better driver monitoring system wouldn't have changed anything since they can't measure situational awareness. It was 100% my responsibility not to allow AP to erode my driving, and the only way I felt like I could do that was to stop using it in most situations.
With my 2018 Model 3 its almost the complete opposite where my driving is sharpened by NoA/AP because I'm always trying to prevent embarrassment. Every trip it's like "nope, you're not going to false break here" or "nope, not going to lane change here".