The Self-Driving Vehicles Thread

[The Spin Doctor]

Phew!

[Horse]

https://www.volvoautonomoussolutions.co ... ts/tara/ar

Volvo autonomous quarry truck

[slowsider]

https://www.volvoautonomoussolutions.co ... ts/tara/ar

Volvo autonomous quarry truck

*Waves*

https://www.constructiontechnology.medi ... 51.article

[Horse]

https://www.constructiontechnology.medi ... 51.article

Someone pointed out that off-road AVs are likely to be different from on-road in two main ways:
- No rules of the road if there are no roads
- On-road is a non-contact sport, off-road is often deliberately impacting the environment

That excavator ticks both!

[wheelnut]

https://www.cnbc.com/2021/08/23/elon-mu ... great.html

Elon Musk says Tesla’s latest beta self-driving software is ‘not great’

I can see Tesla going down in a blaze of litigious glory at some point.

I can appreciate the drive for autonomous cars, but the technology is nowhere near yet, and won’t be for a good few years yet. I think I the car manufacturers are starting to realise that now.

[Horse]

I think I the car manufacturers are starting to realise that now.

"Autonomous vehicles in the smoke and mirrors may be slower than they first appeared"

[Horse]

With 41 fatal collisions (I haven't heard of a single collision with multiple fatalities), you have to wonder how many collisions occur.

Just happened across this. Large vans, with beacons and roof-mounted light arrows. 26 that have been clattered, with 5 in 15 days.

FWIW the US is over 50 fatalities this year.

Being because this on LinkedIn today:


Translated:

This is the 31st van collision on our network since the start of the year !!!!

[Cousin Jack]

Camo innit.

Red and white diagonal stripes - looks like it's moving away from you. And big orange flashing lights, just dazzled him.

[Lutin]

Tesla withdraws self-driving beta over software issues

Tesla has withdrawn its latest "full self-driving" car software update after drivers complained of problems.

Some drivers reported intermittent issues such as safety alerts sounding, despite no danger being present.

The latest version was rolled back on Sunday afternoon, less than a day after it was released.

Tesla chief executive Elon Musk said there were "some issues" with this version, but it was "to be expected with beta software".

A few hours earlier, Mr Musk said the launch would be delayed because the company's quality assurance team had found "regression in some left turns at traffic lights".

[The Spin Doctor]

Would you be happy running a business on a PC running beta software?

Why would you be driving a car running on 'beta' software?

Either would be fun in a non-critical environment, just to see if / where it breaks.

[slowsider]

A dissenting voice:

https://www.bloomberg.com/news/articles ... obal-en-GB

[Horse]

A while ago I posted a video about using real crash information to test and develop AV software. This covers something similar.


https://www.sciencedirect.com/science/a ... 7521004851

Waymo simulated driving behavior in reconstructed fatal crashes within an autonomous vehicle operating domain

Highlights

• A novel alignment methodology was employed for counterfactual simulation of reconstructed collisions.

• The simulated Waymo Driver prevented the initiation of every fatal collision in the dataset without performing urgent evasive maneuvers.

• As a responder, the system was estimated to prevent 82% of fatal collisions and mitigate an additional 10%.

• The majority (63%) of the avoided responder scenarios were prevented without the need for urgent evasive action.


Abstract

Preventing and mitigating high severity collisions is one of the main opportunities for Automated Driving Systems (ADS) to improve road safety. This study evaluated the Waymo Driver’s performance within real-world fatal collision scenarios that occurred in a specific operational design domain (ODD). To address the rare nature of high-severity collisions, this paper describes the addition of novel techniques to established safety impact assessment methodologies. A census of fatal, human-involved collisions was examined for years 2008 through 2017 for Chandler, AZ, which overlaps the current geographic ODD of the Waymo One fully automated ride-hailing service. Crash reconstructions were performed on all available fatal collisions that involved a passenger vehicle as one of the first collision partners and an available map in this ODD to determine the pre-impact kinematics of the vehicles involved in the original crashes. The final dataset consisted of a total of 72 crashes and 91 vehicle actors (52 initiators and 39 responders) for simulations. Next, a novel counterfactual “what-if'' simulation method was developed to synthetically replace human-driven crash participants one at a time with the Waymo Driver. This study focused on the Waymo Driver’s performance when replacing one of the first two collision partners. The results of these simulations showed that the Waymo Driver was successful in avoiding all collisions when replacing the crash initiator, that is, the road user who made the initial, unexpected maneuver leading to a collision. Replacing the driver reacting (the responder) to the actions of the crash initiator with the Waymo Driver resulted in an estimated 82% of simulations where a collision was prevented and an additional 10% of simulations where the collision severity was mitigated (reduction in crash-level serious injury risk). The remaining 8% of simulations with the Waymo Driver in the responder role had a similar outcome to the original collision. All of these “unchanged” collisions involved both the original vehicle and the Waymo Driver being struck in the rear in a front-to-rear configuration. These results demonstrate the potential of fully automated driving systems to improve traffic safety compared to the performance of the humans originally involved in the collisions. The findings also highlight the major importance of driving behaviors that prevent entering a conflict situation (e.g. maintaining safe time gaps and not surprising other road users). However, methodological challenges in performing single instance counterfactual simulations based solely on police report data and uncertainty in ADS performance may result in variable performance, requiring additional analysis and supplemental methodologies. This study’s methods provide insights on rare, severe events that would otherwise only be experienced after operating in extreme real-world driving distances (many billions of driving miles).

[Mussels]

Sounds promising but I'm dubious about how they selected which accidents to reproduce.

[Cousin Jack]

I like the idea of it avoiding most of the crashes and reducing the severity of lots of others, BUT

what is it like in doing everyday, routine, dull driving? Basing any software on the exceptional cases is itself a major risk. Saving 82% of 72 serious crashes is great, but not if it actually causes 000's other crashes that would not have occurred with real drivers.

[Horse]

but I'm dubious about how they selected which accidents to reproduce.

A census of fatal, human-involved collisions was examined for years 2008 through 2017 for Chandler, AZ, which overlaps the current geographic ODD of the Waymo One fully automated ride-hailing service. Crash reconstructions were performed on all available fatal collisions that involved a passenger vehicle as one of the first collision partners and an available map in this ODD to determine the pre-impact kinematics of the vehicles involved in the original crashes.

NB ODD is operational design domain, ie where it is able to operate

[Horse]

but not if it actually causes 000's other crashes that would not have occurred with real drivers.

Spin (possibly, apologies if not) posted something about collision involvement of the Google cars (they publish full details of every crash) and IIRC they had fewer. Plus, IIRC, they were more likely to be the 'innocent party'.

[Horse]

Nearest I can find right now:

https://www.google.com/amp/s/venturebea ... month/amp/

Waymo’s driverless cars have driven 6.1 million autonomous miles in Phoenix, Arizona, including 65,000 miles without a human behind the wheel from 2019 through the first nine months of 2020. That’s according to data from a new internal report Waymo published today that analyzed a portion of collisions involving the robo-taxi service Waymo One, which launched in 2018. In total, Waymo’s vehicles were involved in 18 accidents with a pedestrian, cyclist, driver, or other object and experienced 29 disengagements — times human drivers were forced to take control — that likely would have otherwise resulted in an accident.

[Cousin Jack]

I have never been to Phoenix, Arizona, but I have been to Arizona. Empty is the word that springs too mind.

I have driven in a few big US cities, San Francisco, Las Vegas, Portland, Seattle, and Boston. Boston traffic is reckoned by locals to be almost as bad a New York, and extremely scary. I have also driven in London, and Boston was a pussy cat by comparison. A 'successful' autonomous system in Phoenix may be useless in London (and perhaps vice versa)

[The Spin Doctor]

but I'm dubious about how they selected which accidents to reproduce.

A census of fatal, human-involved collisions was examined for years 2008 through 2017 for Chandler, AZ, which overlaps the current geographic ODD of the Waymo One fully automated ride-hailing service. Crash reconstructions were performed on all available fatal collisions that involved a passenger vehicle as one of the first collision partners and an available map in this ODD to determine the pre-impact kinematics of the vehicles involved in the original crashes.

NB ODD is operational design domain, ie where it is able to operate

What about non-fatal crashes? Just because a crash DIDN'T result in a fatality isn't any reason to exclude it!

[slowsider]

I have never been to Phoenix, Arizona, but I have been to Arizona. Empty is the word that springs too mind.

I have driven in a few big US cities, San Francisco, Las Vegas, Portland, Seattle, and Boston. Boston traffic is reckoned by locals to be almost as bad a New York, and extremely scary. I have also driven in London, and Boston was a pussy cat by comparison. A 'successful' autonomous system in Phoenix may be useless in London (and perhaps vice versa)

Maybe not vice versa
https://wayve.ai/blog/unlocking-markets ... generalise