CHOICE membership

Autonomous Vehicles


#21

A simple challenge here is that Google, Volvo, Uber etc are all trying to make the technology fit the world as we know it today?

There is a wealth of public information or comment on the potential benefits, applications, and risks measured against current knowledge.

Should we also be considering how we change the environment so that the risks arising from autonomous vehicles are reduced through separation or other? This may be a harder debate than simply hoping we can work around any short comings of AI working reliably in the environment of the average Australian roadway.

I’d hoped to support this observation with related research however it is very sparse and indirect except for a Blade Runner inspired future.


#22

Like the “trackless light rail” that can follow a painted line, instead of rails? A sheltered environment might facilitate development, but would be an interim measure.

In one incident, a development vehicle crashed into a concrete barrier because the lane markings had worn faint. Relying on modified environments relies on maintaining the modifications.


#23

… a bit like when a car runs off the road and hits a tree. They never ran off the road when there wasn’t one - of course they weren’t cars back then, but carriage, chariot, whatever, for the sake of the discussion. They probably still hit trees though, occasionally. I guess the point is how modified the environment is, and how adaptable the driver (wet or soft) is to the modified environment not presenting ‘as expected’ - and of course the other case of how well the driver can adapt to a deliberately compromised modified environment, an ‘exploit’ or ‘attack’ if you will. Imagine the ‘fun’ one could have with a line painting machine …


#24

The ABC DECIDES for us?
(Warning - mild humour and facts follow. This post is rated PB for partly believable and scored 3 flying pigs in my review.)

Yes they really are Driverless cars, and not autonomous vehicles if you follow the ABC’s use of language.

@draughtrider previously clarified

Is the test of a successful autonomous vehicle going to be judged by how closely it approximates the behavior of an everyday inattentive driver of today?

The ABC article is very direct in the assessment of Victoria’s Road system as an example.
Infrastructure Victoria has also delivered a report to the Government saying “good quality” roads are required for driverless vehicles to operate effectively.

How much that will cost is unclear, but the RACV’s Dave Jones said there was already a huge road and transport backlog in the state.

The need for legislative change has also been recognised.
A report by law firm Herbert Smith Freehills warned that governments needed to get the changes right, in order “to ensure there is no repeat of having to retro-fit legislation to keep up with technology, as was the case when Uber disrupted the taxi sector”.

Many of us might question the likelihood of all the state and territory Ministers for Transport agreeing on the changes by 2020 or their ability to get it right for the consumer. Right for Google etc perhaps?


#25

I forget which of our members posted it, he had been involved in multi-state issues and the memorable quote was, to paraphrase, ‘When you have 7 people in a room and each already has it perfect, it is hard to get changes…’ I doubt the states could agree on the specs for road lines, locations of traffic signals, signage, or how to respond to emergency lights be it slowing to pass or giving way to emergency vehicles, especially whilst stopped at red lights.

While an autonomous car could be programmed for 7 versions of the rules, the few kms near state borders could become dash cam videos of legend.


#27

Another article regarding autonomous vehicles.


#28

Could be viewed as modifications to the environment, necessitated by inadequacies of the (human driver) technology. Different technologies; different modifications.

On the other hand, our goal should be autonomous vehicles that perform better than human drivers in the (unmodified) real world. Something like the DARPA challenges, the ultimate aim of which was autonomous battlefield vehicles.
https://www.darpa.mil/about-us/timeline/-grand-challenge-for-autonomous-vehicles

For example, many years ago, an empty water delivery truck was driven off a mountain road near my place. It was the end of a long, hot (45°C+) day and the driver was understandably fatigued. The road wound around to the extent that the driver simply missed seeing that it curved away. He drove toward a distant length of road, without noticing the gully between. Fortunately, the lantana proved thick enough to hold the vehicle and prevent injury. Would an autonomous vehicle have done the same? If it had radar, probably not.

Meanwhile, even the footpath isn’t safe:


#29

Sounds great.

I guess at some stage there will be a tipping point where the environment will more closely track the needs of autonomous vehicles to the detriment of riders and drivers.

Ultimately the time might come when you’ll need to take secondary means with you to traverse incompatible places, like a weekend at Boggy Hole, or a Simpson Desert crossing or any number of so-called tracks that follow open plains or creek beds where the only guides are the edges of the creek or a GPS coordinate and knowledge of what points of reference can be used along to way. It’s not uncommon for some of us …

For those living in boundary-rich environments where everything is well defined, the level of control that can be rolled out to preserve everyones safety will be interesting to watch. The needs of the pristine automated environment could have interesting impacts on the perceived freedoms of the population - and like the aforementioned speeding fines/etc it will all be sold under the banner of safety.

I think it will be about as successful as the picture until we are ‘fully migrated’ …
image

… probably make ‘Demolition Man’-s take on the brave new world look conservative. You know what they will automate next !! :wink:


#30

Interesting options @Fred123.

Although in the article I think I lost the thread at the

I appreciate the optimism of some of the press, and wonder at the sensational future ahead! The articles can devalue themselves by being less than complete in their assessments.

Consider what was possibly left out in any of these various linked items?

No more waiting at the intersection to cross the road, the lights are gone. Just step right out and act like an MD on an emergency to deliver a baby while wearing your Nobel Prize and three Olympic Gold Medals. Whose life will AI put at risk? The smart money may be on autonomous humanoids who can wireless interface with the traffic.

Any volunteers for an upgrade? (Insert: Blue Police Box emoji)


#31

Hopefully our optimism here does not extend to

Common sense is that replacing human combatants with non-human is not a solution. Working towards shared goals together is far more effective.

Permitting a machine to operate in a controlled environment reduces risks. Permitting a small group of coders to preconfigure their view of the world and deliver possibly traumatic outcomes in an open environment is not without consequence.

There are numerous working automated manufacturing and operational environments including mining. It is not the machines that are the problem. It is the uncertainty around human interactions in these environments that create risk. To date the best solution remains keeping the two separated by physical barriers (interlocked guarding being one way).

It would be easier to accept that autonomous vehicles should be permitted when the developers accept they cannot eliminate the risk of inadvertent human behavior in an uncontrolled environment. Even in a test situation the first fatality involving an autonomous vehicle demonstrated we have some way still to go?


#32

Your comment piqued my curiosity. Imagine if all technological advancements were viewed similarly. Would we still be awaiting mass deployment of motor vehicles? Of course modern technology has wider repercussions than these examples, but food for thought on how society has historically dealt with such issues. Can and will it continue to do the same in these times of supposed AI, laws, 24x7 information and so on?


#33

Your reality is poignant!

I more correctly wonder if spending billions developing autonomous vehicles is the wisest use of funds?

Funds that corporations have acquired through profiting from consumers.

How would you spend as a philanthropist $100B to improve our lot?
Eg by buying and closing coal fired power stations or developing autonomous vehicles or something else?

Or if you chose to invest that amount?
Eg by developing a better battery or developing autonomous vehicles or something else?

It is possible to see the benefits of the change towards autonomous vehicles, that there is a need for compromise in doing so with some pragmatism and caution on the way. Is this a good enough outcome for the consumer?

or
Is it really just a way for Alphabet and others to become more dominant in our lives and take ever more of our consumer dollar while leaving the consequences to others?


#34

I love the idea especially for people who can’t drive at all.No doubt when those kinks are ironed out.It’s going to be a great thing.Having it being a electric vehicle would also be a huge plus


#35

Not quite sure what you think you’re saying there @mark_m. At present, we have unpredictable humans in the environment and behind the wheel. We can’t do much about those in the environment. All the robot has to do is perform better than the one behind the wheel. Given that (IIRC) 90% of accidents are attributed to human error, that’s a pretty low bar.

We’ve come a long way:

The first competition of the DARPA Grand Challenge was held on March 13, 2004 … None of the robot vehicles finished the route.

The second competition of the DARPA Grand Challenge began at 6:40am on October 8, 2005. All but one of the 23 finalists in the 2005 race surpassed the 11.78 km (7.32 mi) distance completed by the best vehicle in the 2004 race. Five vehicles successfully completed the 212 km (132 mi) course

There are always other uses. Which is “wisest” is a matter of opinion.


#36

And it is a simple mater of opinion that research into bridging the gap between fundamental discoveries and military use is not an efficient use of capital. Without argument it is likely to be highly profitable?

How would anyone given the freedom choose to spend $100B if not for their own benefit? There are many other options. Why did Alphabet go down the path of autonomous vehicles when It could have chosen education in remote communities or zero carbon energy generation or?
Maybe even a gun that will not fire if pointed at another person - that would be a good use of AI would it not?

The market has made a decision there are profits in autonomous operation of vehicles. And consumers can be sold on it for the greater good of those who can afford to mortgage all to the winner of the automation race.

Note that it is probable for success there will likely be several key solutions that all need to co-operate, just as an all Mac or all Android future. That is the prize. There is no room for two winners as there is little chance more than one can coexist? Not Beta and VHS all over again.

Compared with the much higher rates of loss of life due to disease, drugs, domestic violence and mental health in more advanced economies the impacts of losses through road trauma appear insignificant. Compared to the losses in less developed consumer societies is the need for consumers to invest heavily in autonomous operation so critical to our success, or something else?


#37

The DARPA Challenges are just an example of how quickly we can advance. Treating them as the be-all and end-all is disingenuous.

Again, not quite sure what you think you might be saying @mark_m. It’s quite possible to do two things at once (the old walk & chew gum thing). Nobody is suggesting that all available resources be committed to this one task. :thinking: Or are you?


#38

This is either cool or spooky:


Where’s the fun in a bike that you can’t prang?

BMW is not alone:

computer vision and face recognition system allows the user to summon the bike with the wave of a hand.

It’s inevitable, apparently:


Of course, it began with the DARPA challenges:


#39

It was never about doing two things at once.

IE. If you can create components within AI or autonomous operation that are fundamental and common to all vehicles, and patent them you carry an advantage. Similar to the internet, there is only one system. And for wifi despite variations the CSIRO finally gained recognition it holds a core technology patent.

My error in not being a little more direct.

It is not necessarily essential to have the best autonomous vehicle design to be successful. It may be just as valuable to hold one or more of the core patents that enable all vehicles. It seems unlikely two totally different systems for operation of autonomous vehicles will be able to intermingle freely. It would be like trying to view a VHS tape in a Betamax player.

You had me briefly with the benign benefits of the DARPA challenges.

The autonomous motor bike links leave me lost for a reply.

My bicycle when in my hands has a mind of its own already. It does not need another! :crazy_face:


#40

Like a great many other markets. You do realise that patents expire, don’t you?

On that point, the purposes of patents include encouraging innovation and discouraging trade secrecy. The patent gives the holder exclusive rights, for a limited period, to reward their innovation. It also exposes details of the innovation, so that others can build on them.

In the field of autonomous vehicles, those challenges are significant. Does any other adjective that you choose to apply change that?


#41

Yes.

Between all of us it is unlikely we could reliably predict who will in the near future make the most profit out of autonomous vehicle development.

Whether by patent or more typical of modern technology IP imbedded in firmware and operating systems (methods that have much longer tenure than a patent) the original comment was chosen to illustrate the reasoning.

Autonomous vehicles will arrive at some time in some form as everyday transport. It is interesting to observe progress technically.

Perhaps some informed comment on the views of the insurance under-writing industry and relevant state government legislative working committees would be useful here now?