Testing Autopilot

bwilson4web said:
In ordinary traffic, a crossing car or truck will trigger a hard brake. But why these are detected and the tractor trailer remains invisible is a mystery.
Click to expand...

Because using software in an attempt to recognize objects in video images is not reliable. It might work... and it might not. Even 99.9% reliability isn't sufficient, if that means it fails to "see" 1 out of every 1000 semi trailers. And at best, it's quite slow in terms of computer processing speeds.

Using active sensors -- phased-array radar or lidar -- is much more reliable and much faster. That's why Waymo's self-driving cars use all three systems: Cameras, lidar, and phased-array (so-called "high-res") radar.

Aside from ordinary video cameras, Tesla is only using extremely short-range (within a few feet) radars and extremely short-range ultrasonic scanners, plus a front-mounted low-res Doppler radar.

Here is an attempt to show, visually, what Tesla's low-res Doppler radar "sees":

screen-shot-2018-06-18-at-5-56-16-am.jpg

(I'm linking to the source for the sake of completeness, but I'm very skeptical about a lot of what's said in that article. Looks like a lot of guesswork to me, and I think some of those guesses are wrong.)

Note the size of the radar return (indicated by the size of the circle) shows only distance, not size or shape. Also, note it ignores stationary obstacles such as telephone poles and trees. (A couple of objects, indicated in orange, are labeled "stationary", but my guess is those objects were first detected as moving, and are now identified according to their previously detected position when they stopped. Or perhaps at least one is a false positive, since the larger orange circle on the right appears empty.)

Quite clearly this is entirely unsuitable for 3D mapping. If, for example, there is a tractor-trailer rig pulling onto or across the highway, into the path of your car, then the self-driving system needs to know how big it is... not just how close it is.
 
Last edited:
Even ‘automatic door’ radar should haves detected the truck.

Now I’m curious about the Tesla radar . . . and that could be a bad thing.

Bob Wilson
 
The drive from Huntsville AL to the Memphis TN SuperCharger takes US 72 but Mississippi had some surprises:
MS_cross_010.jpg

There are dashed, white lines on the right side, straight alongside the lane. AutoPilot drives straight across with no problem.

The real surprise was an intersection that did not have the dashed lines on the right-hand side:
MS_cross_020.jpg

Trying to split the difference, the AutoPilot steered a curve to the far right ditch. Once I figured it out, it was easy enough to hold the steering straight and the rest of the trip was less exciting.

Bob Wilson
 
bwilson4web said:
The drive from Huntsville AL to the Memphis TN SuperCharger takes US 72 but Mississippi had some surprises:
MS_cross_010.jpg

There are dashed, white lines on the right side, straight alongside the lane. AutoPilot drives straight across with no problem.

The real surprise was an intersection that did not have the dashed lines on the right-hand side:
MS_cross_020.jpg

Trying to split the difference, the AutoPilot steered a curve to the far right ditch. Once I figured it out, it was easy enough to hold the steering straight and the rest of the trip was less exciting.

Bob Wilson
Click to expand...


Wow, that is certainly a BETA product, and Elon guides for FSD by the end of the year? Ya, Right...
 
I'll OK with today's basic AutoPilot and if it gets better, I'll say THANK YOU. Things I'd like to see:
  • Detect tractor trailers that do not have side skirts.
  • Use road edges to handle missing or worn lane markers.
  • Handle 1 lane into 2 or more lanes without oscillating the steering.
Bob Wilson
 
Looking at the engineering specs for the Continental ARS4-B, the Model 3 radar unit, reveals the beam pattern sweeps side-to-side:
ARS4_010.jpg

There is no elevation scan which begins to explain why there can be ambiguous targets detected.

A better fix would be another radar unit rotated 90 degrees to scan in elevation. Then collect the beams into a matrix and the problem is solved.

Bob Wilson
 
I'd posted this in three other forums only to discover two were interested and one treated it as "horrible!" Regardless, here it is with more background:

Basic Autopilot

In a rush to visit my Mom with my wife and her lap dog, I left my electric razor and Continuous Positive Airway Pressure (CPAP) machine at home. In a world of disposables, the razor was easily fixed (Barbasol still tastes crappy) but a CPAP is not an off the shelf item. Used to treat sleep apnea, it ensures my sleep is not interrupted by short airway blockages. Then my wife spent seven days in the hospital.

Easily confused by strangers, my presence in her room anchors her in a safer reality to avoid more aggressive restraints. But her 24 hour care requires my sleeping in that abomination called a 'reclining chair.' It has two positions, seat parallel to floor or raised ~20 degree back. There is an extendable leg lift but the 90 degree, seat-to-back angle is fixed. Reclined, it is like sleeping in a bucket with the legs and knees above the rim. To mitigate, I filled the seat-to-back hole with spare sheets, blankets, and a pillow filler so I could 'cat nap' and be available to help the staff. My wife was discharged after IV antibiotics reduced the hand swelling and pain. So I took a nap at my Mom's home, packed the car, and we were off to home in Huntsville at 4:00 PM.

'Micro sleep' is a thing that my narcoleptic wife (and late aunt) suffered. Involuntarily, the victim goes to sleep for 2-5 seconds with no warning and the most effective treatment is uninterrupted sleep. After seven days cat napping in a bucket chair and one post release nap, I was primed.

Tesla's Autopilot has radar dynamic cruise control and automatic emergency braking to avoid running into things. The optical and ultrasonic sensor based, lane steering keeps the car from darting into a ditch or into on-coming traffic. It was on the leg to Decatur around 7:00 AM there were at least five micro sleep events that I was aware of and Autopilot kept us safe. I stopped in Decatur for a pee, stretch, and coffee break before driving the last 20 miles home.

Home again, I have my CPAP sleeping aid, my wife has her dogs, and me in familiar settings. Life has returned to what passes for normal ... but it is life thanks to Tesla's Autopilot.

Bob Wilson
 
It was on the leg to Decatur around 7:00 AM there were at least five micro sleep events that I was aware of and Autopilot kept us safe. I stopped in Decatur for a pee, stretch, and coffee break before driving the last 20 miles home.
Click to expand...

I do find it very disturbing that people who quite clearly shouldn't be driving a car, are using Tesla's AutoSteer driver assist feature as if it makes the car a true self-driving car, in the sense of Level 4+ autonomy... which it most definitely is not. We've seen reports of drunk drivers trying to rely on Autopilot + AutoSteer to get them home, too.

I have a friend who was diagnosed a couple of years ago with epilepsy, which she didn't previously know she had. She had to suddenly stop driving. That is the responsible thing to do; this guy suffering from narcolepsy is a danger to himself and others on the road, and absolutely should have his license revoked immediately!

Tesla's Autopilot + AutoSteer remains at Level 2+ autonomy. Tesla has demonstrated Level 3 autonomy in some demo videos, but their production cars aren't up to that level yet. Waymo apparently has Level 3 autonomy in its self-driving taxis.

Nobody has yet achieved Level 4 autonomy. It will be a wonderful breakthrough when that happens; those who cannot drive, or can't drive safely, will be able to have an autonomous car drive them around -- their very own robo-chauffeur!

I seriously doubt Tesla will be the first to achieve Level 4 driving autonomy, but I hope whoever does develop it will license the tech for all auto makers to use.
​
 

I'm not sure what you're suggesting, but I'm pretty sure you don't actually mean a similar radar scanner pointed 90° away from the other. Scanning straight up is unlikely to be of any use, 99.999% of the time!

Of more general use would be a scanner with a broader vertical scan, but shorter distance. That would hopefully prevent such accidents as the Tesla car which, under Tesla Summon, ran its windshield into the load sticking out from the rear end of a flatbed truck. With such an elevated obstacle, it's impossible for a bumper-mounted sensor to detect it.

And it's stupid to put the primary scanners down near the bumper. They should be elevated as far as possible. Mounting a lidar scanner under a dome on the roof may look "ugly" to us, because we're not used to it, but that's the proper placement for a self-driving car's scanners. However, with solid-state scanners, the dome can be eliminated (see photo below).

Women and other short people enjoy the elevated seating position of an SUV or minivan, because it helps them see over other cars on the road. For the exact same reason, the primary scanners in a self-driving car should be as high on the car as possible. If that means having a roof which is a few inches thick, then that is what the future of automotive design should be. Those who find an elevated roof to be "ugly" are simply going to have to get used to it. No doubt back in the day of horseless carriages, people also found it "ugly" when auto makers started putting the engine in a metal box at the front of the car, rather than mounted over the rear axle.

Technical advances affect fashion, and often Form follows Function.

maxresdefault.jpg

Toyota's Platform 3.0 prototype
 
A couple of corrections:
  • micro-sleeps - treated by CPAP, it is not a problem. In this case, I'd forgotten my CPAP but my wife's seven day hospitalization was unexpected. Regardless, a micro-sleep is NOT expected and in my case, I thought driving into the rising sun would be OK. That drive showed I didn't realize what was about to happen. Regardless, I have a diagnosed sleep apnea and treated by a CPAP, I'm more than ready to match those who are not yet diagnosed as having the syndrome. Yes, I pushed the margins but didn't realize it was a problem. Fortunately, AutoPilot kept us safe. Would you prefer an alternate outcome?
  • sleep doctor report - a month before this trip, my sleep doctor gave me a clean bill of health which he also gives to truck drivers and pilots. Do you need to read a copy?
  • we arrive safe and sound - when I got to Decatur, I did the standard fix: caffeine, biology break, and walk-around. This was NOT an option on the road leading to Decatur because it is so sparsely populated.
I don't mind even for a nanosecond if anyone wants to say "YOU SHOULD NOT BE DRIVING" because it is inaccurate. Narcolepsy treatment worked fine for my wife. CPAP therapy works fine for me. Insulin works fine for diabetics. Glasses work fine for those with near-sighted vision. ... What a concept, modern medicine can treat a disorder!

So 'stick it up your *ss' if you think I'll ever call the AutoPilot that saved me, my wife, and her lap dog from five potential accidents driving to Decatur. If you are really up in arms, take it to your local legislator and create a new population of welfare recipients. But until you at least write a letter or do a face-to-face with your state and federal representatives, might as well join our schizophrenic friend "Green".

Bob Wilson
 
Last edited:
bwilson4web said:
sleep doctor report - a month before this trip, my sleep doctor gave me a clean bill of health which he also gives to truck drivers and pilots. Do you need to read a copy?
Click to expand...

Okay... I need to do some back-pedaling here.

1. Bob, I thought you were relating a report from someone else; I didn't realize that you were talking about your own experience.

2. If your doctor gave you a clean bill of health and said it's okay to drive, then he's the expert and I'm an uninformed layman.

3. Two to five seconds of sleep when driving sounds dangerous to me, but again if your doctor okays it, then perhaps I'm just demonstrating my ignorance of the subject.

Bob, no offense was intended, and if I had realized you were relating a personal experience, I probably would have -- wisely -- kept my opinions to myself. Peace, brother?
​
 
Pushmi-Pullyu said:
3. Two to five seconds of sleep when driving sounds dangerous to me, but again if your doctor okays it, then perhaps I'm just demonstrating my ignorance of the subject.
Click to expand...
It is dangerous which is why I use a CPAP machine. Normally associated with excess weight, I had pulled 30 lbs off and not planned to stay longer than 2-4 days. Events led to a longer stay and camping out in my wife's hospital room. I had not suffered a micro-sleep for years once I had started using CPAP therapy.

What is sad are how many overweight people, a Southern tradition, do not realize they have sleep apnea which means they are not even aware of the symptoms.

Bob Wilson
 
Source: https://electrek.co/2019/02/13/tesla-autopilot-crash-rate-nhtsa-contested-report/

. . .
Now the claim [40% reduction in accidents after Autopilot raw] is being contested by Quality Control Systems Corp (QCS), which describes itself as a company providing “forensic statistical services and products”.

QCS filed a FOIA (freedom of information act) lawsuit against NHTSA for them to release everything they had about Tesla Autopilot.

The company eventually obtained the data and after analyzing it, they state that NHTSA couldn’t make that claim and that the data actually suggests that the crash rate went up after Tesla introduced Autopilot:
. . .​

The QCS source page: http://quality-control.us

You can easily download the report in which we find:

NB: The airbag deployment and mileage data on which NHTSA based its findings were apparently collected through July 8, 2016.(1) Tesla announced on October 19, 2016 substantial changes to the hardware on which Autopilot relies in vehicles produced as of that date.(2)
. . .
pp. 2 of 24​

So now we have a problem. This report analyzes data from a version of AutoPilot with stale data that ends in July 8, 2016. But QCS claims AutoPilot was replaced after October 19, 2016 and today is June 3, 2019. Stale data about an early version, it doesn't make sense to waste time on this.

Source: https://www.greencarreports.com/new...-better-than-with-human-drivers#disqus_thread

Bogus Claims​

Source: https://hcai.mit.edu/tesla-autopilot-miles-and-vehicles/

Tesla Vehicle Deliveries and Autopilot Mileage Statistics
Last Updated: April 27, 2019 Originally Posted: June 20, 2018

This page provides estimates of Tesla vehicles delivered and Autopilot miles driven segmented by Autopilot hardware version, based on the following milestones:

Sep-2014: Autopilot Hardware 1.0 installed (Autopilot not enabled).
Oct-2015: Autopilot enabled.
Oct-2016: Autopilot Hardware 2.0 released.​

Bob Wilson
 
Last edited:
Source: https://www.consumerreports.org/autonomous-driving/cadillac-tops-tesla-in-automated-systems-ranking/

. . .
Autopilot and Super Cruise were the clear winners. These systems accelerated and slowed comfortably and were able to reliably keep the vehicle centered in the lane for several miles at a time. The Nissan and Volvo systems had trouble with curvy or hilly roads, and they had frequent lane departures.

Nissan says that its lane-centering is limited intentionally, to make sure that drivers don’t lose engagement. “We’re concerned about driver overtrust in the system,” says Andy Christensen, lead technology expert for technology, planning, and research. “By dialing up the vehicle’s control too high, you may wind up in a situation where the customer doesn’t understand their role in driving.”

Mikael Ljung Aust, a driver behavior specialist at the Volvo Cars Safety Centre in Sweden, agrees. He says that driver engagement remains a key concern for the company and that drivers should think of the system as more like “advanced power steering.”
. . . ​

I recently saw a Nissan commercial about their ProPilot that was shown doing AutoPilot things and wondered what it was about.

Bob Wilson
 
I received my used 2018 model 3 mid range with "full self driving" on 06/14/2019, which replaced my used 2015 i3. I've noticed a huge lack of information from car re-sellers when it comes to the specifications and purchased options on Teslas. My re-seller couldn't even tell me what kind of range it had, but I knew it was either the mid range or long range RWD from pictures. For the price I was hoping it was the long range. I was slightly disappointed when I found out it was the mid range, but through the roof when I found out it has FSD.

I've been thoroughly enjoying my car and using FSD when on the freeway. I'm wondering how others get into accidents using FSD on roads with cross traffic. For me FSD does not engage until I enter a navigation destination and enter on-ramp, FSD then messages me it will disengage when I exit the off-ramp. On regular public roads the system is using advanced cruise control with steering assist, but I don't expect it to act like FSD. I was also happy to find out I could force lane changes by using the turn signal in order to place me in a better lane for my commute.
 
Shadrach777 said:
I received my used 2018 model 3 mid range with "full self driving" on 06/14/2019, which replaced my used 2015 i3. I've noticed a huge lack of information from car re-sellers when it comes to the specifications and purchased options on Teslas. My re-seller couldn't even tell me what kind of range it had, but I knew it was either the mid range or long range RWD from pictures. For the price I was hoping it was the long range. I was slightly disappointed when I found out it was the mid range, but through the roof when I found out it has FSD.

I've been thoroughly enjoying my car and using FSD when on the freeway. I'm wondering how others get into accidents using FSD on roads with cross traffic. For me FSD does not engage until I enter a navigation destination and enter on-ramp, FSD then messages me it will disengage when I exit the off-ramp. On regular public roads the system is using advanced cruise control with steering assist, but I don't expect it to act like FSD. I was also happy to find out I could force lane changes by using the turn signal in order to place me in a better lane for my commute.
Click to expand...
Just curious, how much?

Bob Wilson
 
bwilson4web said:
Just curious, how much?

Bob Wilson
Click to expand...
45.5 + TT&L with 7.7k miles on it, found the original sticker in the glove box that said the car was 53.7 and didn't show FSD on the options. I don't know if FSD was purchased at full price or if it was purchased in March when heavily discounted.

I had actually ordered a blue SR+ back on May 29th, but it was taking forever to get to me. I even called Tesla and told them I would be willing to quicken the process and purchase something that was in local inventory (they gladly switched my order to local inventory that was red with sport wheels). I was scheduled a delivery date for the 21st, but I got skittish and canceled.

I saw this blue used version and knew it had premium interior (which I originally wanted). It was a risk because I didn't know if the person before me had purchased enhanced auto-pilot, let alone FSD.
 
Shadrach777 said:
I saw this blue used version and knew it had premium interior (which I originally wanted). It was a risk because I didn't know if the person before me had purchased enhanced auto-pilot, let alone FSD.
Click to expand...

Looks like Lady Luck granted you her favor.

Congratulations, and I hope you enjoy your "Lemur" Model 3 for many years!
:)
 
I drove nearly 1,500 miles Thursday to Saturday, just under 30 hours, with over 95% using Navigate on AutoPilot. Over 95% was 'hands free' after I 'put a sock in it' to suppress the nagging steering reminder.

Yeap, WAYMO and the LIDAR systems are so way ahead just like fuel cells and nuclear fusion. Way way ahead in the future.

Bob Wilson
 
Last edited:
You must log in or register to reply here.
Back
Top