Short-Term TSLA Price Movements - 2016

[Haddock]

Very sad. I've always wondered why there are no side underride guards in the US, but apparently they are not very effective for cars. Might have helped a little in this case though.

Relevant:

"The U.S. government doesn't require tractor-trailers to have front or side underride guards. In Europe, front underride guards have been required on large trucks since 1994 to protect passenger vehicle occupants in crashes with combined speeds of about 35 mph. Europe also requires side guards to protect pedestrians and bicyclists but not people in passenger vehicles. The Institute has found that front and side underride guards have the potential to reduce injury risk.

In a 2012 IIHS study of fatal crashes between large trucks and passenger vehicles, an estimated 63 percent involved the front of a truck, 22 percent involved the side and 15 percent the rear. Analyses of smaller samples of fatal crashes found that 88 percent involving the side of the large truck and 82 percent involving the rear produced underride."

From NHTSA to weigh new underride guard rule

 

[MrRandyB]

It sounds like the tractor trailer guy shouldn't have went perpendicular and that the driver wasn't paying attention.

Reading it again more closely makes me think the tractor trailer was maybe stopped or going slowly perpendicular to the road while and the S just slammed right into the side. Unfortunately, its sounds like the sensors do not seem to work that high up and the bottom of the tractor trailer impacted the windshield of the Model S.

Does sounds like a freak accident and not an AP problem. This will probably blow over and once the investigation clears should help the SP.

 

[Haddock]

‘Cars’ front-end structures are designed to manage a tremendous amount of crash energy in a way that minimizes injuries for their occupants,’ says Adrian Lund, Institute president. ‘Hitting the back of a large truck is a game changer. You might be riding in a vehicle that earns top marks in frontal crash tests, but if the truck’s underride guard fails— or isn’t there at all — your chances of walking away from even a relatively low-speed crash aren’t good.‘” Underride guards fall short

 

[Yggdrasill]

Yeah - what is AP supposed to do when a tractor trailer drives across the highway perpendicular to the Model S?

Preferably stop. ASAP.

But it's well known that the autopilot doesn't work perfectly when it comes to stationary objects. It doesn't have the required sensors to distiguish very well between objects at the side of the road like road signs and light posts, and a stationary car. Something moving across the field of vison with no change in distance would likely be interpreted as stationary. (The autopilot is intended for highway use, where there is no cross-traffic.)

This is something that will likely be resolved with autopilot 2.0. Maybe even the cars already delivered with dual cameras would handle the situation better.

 

[aznt1217]

You guys do realize that the car went under the gap of the trailer. Think of the fast and furous civic scene in the first movie, but instead of the car being parallel, it is perpendicular and instead of a low rider civic, it was a Model S. RIP to the driver, but you kind of see where I am getting at.

 

[VanE]

I'm sorry but : "when a tractor trailer drove across the highway perpendicular to the Model S" this is just a freak accident. CNBC plastering the headlines such BS.

Oh yes! Over te next few days the news will be covered with this story. It is absolutely a tragedy, but the news will love to blow it out of proportions with headlines such as "Tesla's Autopilot caused death" and " Why you should never trust autonomous cars".

Big chance that it will overshadow the Q2 numbers next week. The only way to make sure that the media won't take this story to it's extremes is to have Musk go out and explain to the world why it happened and what they're going to do to make sure that it won't happen again (continuously updating the AP software).

 

[Johan]

Preferably stop. ASAP.

But it's well known that the autopilot doesn't work perfectly when it comes to stationary objects. It doesn't have the required sensors to distiguish very well between objects at the side of the road like road signs and light posts, and a stationary car. Something moving across the field of vison with no change in distance would likely be interpreted as stationary. (The autopilot is intended for highway use, where there is no cross-traffic.)

This is something that will likely be resolved with autopilot 2.0. Maybe even the cars already delivered with dual cameras would handle the situation better.

Wouldn't also the addition of other sensors using other technology than visual analysis/cameras, for example Lider, Radar or Ultrasound go a long way to alleviating these kinds of "blind spots" in the current system? Such sensors don't care about light conditions or colors, they measure the absolute distance to solid objects and world be able to pick up and let the system react to a suddenly decreasing distance to another object (i.e. an impeding crash)?

 

[esk8mw]

You guys do realize that the car went under the gap of the trailer. Think of the fast and furous civic scene in the first movie, but instead of the car being parallel, it is perpendicular and instead of a low rider civic, it was a Model S. RIP to the driver, but you kind of see where I am getting at.

That's the thing. I bet if the car slammed into a brick wall (front end absorbing the damage) as opposed to the windshield impacting the lip of the trailer, the person would have survived. This was a just a perfect storm of bad conditions.

 

[MrRandyB]

Elon just tweeted with a link to the blog

 

[Johan]

I wonder about the timing of this. News released on June 30th (last day of Q2) shortly after market close. Hmm... Peculiar timing.

 

[X Yes?]

I hope when Tesla announces deliveries this weekend with markets closed that will be the equivalent of seeing the train coming but unable to get off the tracks.

Unfortunate analogy in the wake of the Autopilot death.

 

[Stoneymonster]

Wouldn't also the addition of other sensors using other technology than visual analysis/cameras, for example Lider, Radar or Ultrasound go a long way to alleviating these kinds of "blind spots" in the current system? Such sensors don't care about light conditions or colors, they measure the absolute distance to solid objects and world be able to pick up and let the system react to a suddenly decreasing distance to another object (i.e. an impeding crash)?

I'd be interested to know why the radar didn't see this. I understand the arguments about stationary objects (avoiding for example false positives with parked cars along the side of the road), but it seems like if something is dead ahead, and the closing speed is high, that's a problem that can be detected.

 

[MrRandyB]

Friday

Today is Thursday

 

[esk8mw]

Oh yes! Over te next few days the news will be covered with this story. It is absolutely a tragedy, but the news will love to blow it out of proportions with headlines such as "Tesla's Autopilot caused death" and " Why you should never trust autonomous cars".

Big chance that it will overshadow the Q2 numbers next week. The only way to make sure that the media won't take this story to it's extremes is to have Musk go out and explain to the world why it happened and what they're going to do to make sure that it won't happen again (continuously updating the AP software).

The market is truly irrational if this incident overshadows Q2 numbers (unless there's a ho-hum result of meeting guidance exactly or something). A single death, tragic as it is, is not going to have a material long-term impact on Tesla's business prospects.

 

[Johan]

Today is Thursday

Oooops. (My summer holiday starts tomorrow so I have a Friday feeling). Thx for the correction!

 

[MrRandyB]

I'd be interested to know why the radar didn't see this. I understand the arguments about stationary objects (avoiding for example false positives with parked cars along the side of the road), but it seems like if something is dead ahead, and the closing speed is high, that's a problem that can be detected.

I am thinking this is a similar situation to when the guy crashed his car with summon into the trailer. I think the sensors just don't look up that high.

Edit - link Tesla Model S driver claims his car crashed into a trailer on its own, Tesla says ‘Summon’ was activated

 

[Yggdrasill]

I'd be interested to know why the radar didn't see this. I understand the arguments about stationary objects (avoiding for example false positives with parked cars along the side of the road), but it seems like if something is dead ahead, and the closing speed is high, that's a problem that can be detected.

The autopilot likely doesn't see the lane markings far enough ahead to determine whether the stationary object is outside a curve, or whether it is in the lane. This can really only be resolved by adding more sensors and better SW.

 

[VanE]

I wonder about the timing of this. News released on June 30th (last day of Q2) shortly after market close on a Friday. Hmm... Peculiar timing.

I thought about that too for a second. But I don't think that the NHTSA gives a crap about the valuation of a company to do something like that. (but I have almost no knowledge about the NHTSA so I could be dead wrong)

 

[Stoneymonster]

The autopilot likely doesn't see the lane markings far enough ahead to determine whether the stationary object is outside a curve, or whether it is in the lane. This can really only be resolved by adding more sensors and better SW.

Yeah I realized that shortly after posting. Though the radar should give the object extent, i.e.: wow that's a 40ft long wall (semi), not a car.

 

[Haddock]

We know the AP software isn't great at spotting stationary things. But I also wonder if it "looked under" the truck.

These are artificial trucks from a game, but there are lots of pictures of US trucks with this huge gap under them.