I guess you missed my /s in my post. The point was that the post I replied to could have phrased their statement differently. I simply made an in jest agreeing response.
Thanks for playing though. /s
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
AP just sent me under a truck
- Thread starter superbaka
- Start date
I guess you missed my /s in my post. The point was that the post I replied to could have phrased their statement differently. I simply made an in jest agreeing response.
Thanks for playing though. /s
I have one more update to my OP :
I no longer believe this was a problem with semi's. In the last week, there were 3 similar occasions where NOA shared space with another vehicle (all kinds) in a merge lane and the system refused to acknowledge the car was besides me in the exact same manner as my video.
Today, I was trying to exit a freeway with incoming traffic and I was sharing space in the merge lane with an incoming car. We were already coming dangerously close to each other, when the system initiated a hard turn to exit the freeway directly into the car. I aborted at the last second and it was truly an unacceptable decision for AP that would have resulted in a collision. Not by coincidence, but by steering me into the car.
I will now stop using NOA during merges/shared lane space until the system is fixed. I did not see this behavior over the past few months so something went downhill.
I no longer believe this was a problem with semi's. In the last week, there were 3 similar occasions where NOA shared space with another vehicle (all kinds) in a merge lane and the system refused to acknowledge the car was besides me in the exact same manner as my video.
Today, I was trying to exit a freeway with incoming traffic and I was sharing space in the merge lane with an incoming car. We were already coming dangerously close to each other, when the system initiated a hard turn to exit the freeway directly into the car. I aborted at the last second and it was truly an unacceptable decision for AP that would have resulted in a collision. Not by coincidence, but by steering me into the car.
I will now stop using NOA during merges/shared lane space until the system is fixed. I did not see this behavior over the past few months so something went downhill.
Last edited:
NRG4All
Member
I have tried the AP trial twice and have gone back to just the plain Cruise Control. There are a couple of reasons. First, the AP wants the car to stay in the middle of its lane. When passing a semi I want the car to give more room on the right and less on the left. Sure enough when we were passing a semi we got to the driver's door, and fortunately there was a turn signal on his door, but it was blinking. The massive truck started moving into my lane. I recognized that before the car did anything. I suspect because the truck was still in his lane. Then I had to physically wrestle the steering wheel to move the car over and at the same time mash the accelerator. Second, when I pass a semi I will speed up four or five miles per hour to get by the truck quickly. The AP keeps the car at the same speed which may only be one mph or less faster than the truck. One look at the tire treads at the side of the road is a good example of what could hit your car when passing a semi. It is good to get by quickly. Also, I listen for the sound that semi tires are making. Before the retread separates, it will make a sound. If I hear that then I pass as fast as I can. Unless I'm mistaken I believe that semi trucks only need to have new tires on the front wheels, the rest can be recaps. I'm sure AP doesn't listen for a potential tread separation. Third, I use the 12 second rule from motorcycling that has you scanning 12 seconds ahead and predicting potential trouble. When I see anything that looks possibly amiss I put my thumb over the horn button and my left hand ready to flash high beams to wake up some driver. I don't believe that the AP uses the horn or lights for warning other drivers. I've driven over 300.000 miles on motorcycles and am alive with all my faculties intact. When the self driving function can get to the level that can handle all of these situations, I may consider purchasing it.
Kant.Ing
Member
Exactly what I am doing. I have also trained myself to continue to track the vehicles fall behind me without even looking.
My requirement for an autonomous car is a car with a sensory system (6th sense welcome) and reaction time for making correct decisions better than the best human drivers. In the age of AI, this requirement shouldn't be unreasonable. If the computer is still guessing around or the manufacturers are still talking about statistics, I'd rather drive myself.
Sometimes ‘Genius’ Is Just Great Management
"To improve his decision speed, Senna spent hours honing his concentration and visualizing races. He would slow-jog tracks, sometimes imaging his body was a car and the soles of his shoes were tires. Before the 1984 Montreal Grand Prix, a researcher gave a group of Formula One drivers a test to see how quickly they could make correct decisions. Their average was about 270 milliseconds, or 10% faster than a typical human. Senna's personal average of 240 milliseconds was off the charts. At top racing speed, Senna's exhaustive mental practice gave him a time advantage over his rivals equivalent to five car lengths."
My requirement for an autonomous car is a car with a sensory system (6th sense welcome) and reaction time for making correct decisions better than the best human drivers. In the age of AI, this requirement shouldn't be unreasonable. If the computer is still guessing around or the manufacturers are still talking about statistics, I'd rather drive myself.
Sometimes ‘Genius’ Is Just Great Management
"To improve his decision speed, Senna spent hours honing his concentration and visualizing races. He would slow-jog tracks, sometimes imaging his body was a car and the soles of his shoes were tires. Before the 1984 Montreal Grand Prix, a researcher gave a group of Formula One drivers a test to see how quickly they could make correct decisions. Their average was about 270 milliseconds, or 10% faster than a typical human. Senna's personal average of 240 milliseconds was off the charts. At top racing speed, Senna's exhaustive mental practice gave him a time advantage over his rivals equivalent to five car lengths."
You sound like a good driver. I am similar, with motorcycle experience commuting in rush hour across los angeles, where lane splitting is legal. I favor owning small cars that are responsive (model 3 is a bit large for me but it is pretty responsive).
It is worth noting that, in theory, the neural net method can improve vastly. Elon already mentioned the idea of making better decisions around semis. Since they know where the car is, you can build assumptions like "pass quickly" or stay to one side. That part is not hard at all.. I have a programming background...its very possible... and I think they will get there. AP seems to be, "Stay alive long enough to see it get amazing".
Until then, I mainly use it for traffic. (lifesaver!) I also use the turn signal method to move AP out of the way of semi's ahead of time...I drive a hybrid of manual and AP.
jyalpert
Member
Hmm. I'm confused. You're describing adding explicit rules to the system, but Tesla keeps talking about using a neural network, which is not rules-based. It's not clear to me when they say "training the neural net" what's actually meant there. There would seem to be two tasks - space detection and control.
With the cameras, there is clearly a CNN-type algorithm running to perform the surrounding space detection task. You can see from those pictures where the computer is identifying lane lines, other cars and their properties, etc. This is obviously an important task, and you can see from the CAPTCHAs that are like "click on the boxes with crosswalks" that lots of people are working on it. But usually you use humans to tag those pictures - unclear to me how individual drivers are helping "train the neural network" for this task. (I believe This is what LIDAR really helps with, too - identifying objects in 3-D space).
The next step is the control task. You've identified objects, now you have to make decisions. This is what good drivers are great at - look 12 seconds ahead, anticipate the actions of other cars, always leave an out, etc. I'd assume this is the really hard part. I assume Tesla would want to do this with some sort of NN-based architecture, probably using reinforcement learning. In other words, they'd probably want to avoid programming in explicit rules unless absolutely necessary (in static systems, rules can work OK, but in massive spaces where tons of decisions have to be made using probabilities, rules-based systems are difficult to scale). A really good RL-based system would essentially learn that passing semi trucks is a dangerous endeavor and should be done quickly, rather than requiring Tesla to program in that explicit rule. But everyone in the industry knows that RL systems are really, really, really, really finicky and difficult to make work well even for simple tasks.
Right now AP look pretty rules-based - I noticed for instance that the middle-of-the-lane seeking in autosteer seems a lot like simple PID controller behavior (the explicit rule is - stay directly in the middle of the two lines), or in the NoA lane-change behavior people describe - something like "wait until there is X distance between the two cars in the adjacent lane before changing."
I guess what I'm saying is, it's not clear to me what the strategy is for AP. They seem to be picking off individual tasks, which can be done with point solutions (maybe rules-based or using something more complex), but I'm not sure how those tasks relate to the overall goal of real self-driving.
(To be clear, I am not a data scientist working on autonomous driving - likely obvious to anyone technical reading this. I manage data scientists, but work in a completely different space. If there are any engineers here working on this task, I would love to know what is actually going on! Help please!)
Last edited:
Similar threads
