Hundred-Tonne Robots That Help Keep New Zealand Running [video] | Hacker News Reader

Hundred-Tonne Robots That Help Keep New Zealand Running [video]

youtube.com

155 pointsprotomyth39 comments

British mining company Rio Tinto has 73 416-ton autonomous trucks transporting iron ore 24 hours a day in Australia. The human team overseeing the robots work 750 miles away.

https://www.technologyreview.com/s/603170/mining-24-hours-a-... (2016)

Rio Tinto is a dual headquartered British-Australian company

They've got thieving from the Australian public down to a fine art, gotta hand it to em.

It is insane how little royalties we impose on the big resources companies. But at least Rio Tinto and BHP pay a decent amount of company tax.

It’s the offshore owned ones that pay almost no tax (Glencore, Chevron, British Gas, Exxon, Energy Australia, Peabody, Chevron, etc.).

Investigative journalist Michael West has a good rundown of the top 40 companies that manage to make huge amounts of income but manage to reduce their taxable income to crazily low levels to avoid paying tax - https://www.michaelwest.com.au/tax-dodgers/ - a lot of resource/energy companies there!

Just going further and further off-topic here, but it's so depressing contrasting that list against what Australian politicians actually treat as important issues in regards to the Australian economy.

But then, what would one expect when the Australian Prime Minister (at the time) was one of the names in the Panama Papers listing investors making use of complex offshore shell companies registered in tax havens for the purposes of avoiding local tax law.

Also, Foxtel! $0 in tax on $6 billion total income. I guess that covers all the Australian pirates' share of GoT downloads.

Australian politics is among the most duplicitous in the world. The Australian people have a long history of letting their ruling classes get away with murder - pillage of the land is nothing in a country founded on genocide.

I used to think it was an annoying joke that, when confronted with the fact of my Australian heritage, my American and British colleagues would .. 'jokingly' .. refer to me as a criminal from the prison down under. Now, in my more mature years, I see "Australia, land of criminals" more as a cynical truth, quite as a matter of fact. Australians don't care that their land is being mistreated by massive robots run by multi-national, quasi-sovereign entities - as long as there are avocados to smash ..

I believe the Government(s) of Australia have facilitated the thievery rather than preventing it.

It's a good business decision to save billions with a few millions. I question whether any democracy system is able to handle this.

Pretty much, except for Gough Whitlam and he got removed in suspicious circumstances (involvement from the British Establishment including the Queen and rumoured involvement by the CIA).

The CIA's involvement in the Whitlam fiasco is no rumour - it has been broadly acknowledged, in fact.

https://www.theguardian.com/commentisfree/2014/oct/23/gough-...

"Our man, Kerr...": no nation has fallen so hard on such little breath.

https://paleofuture.gizmodo.com/cia-declassifies-report-on-t...

Chinese Port Goes Full Robot With Autonomous Trucks and Cranes (May 2018)

https://singularityhub.com/2018/05/17/chinese-port-goes-full...

“By the end of 2018, something will be very different about the harbor area in the northern Chinese city of Caofeidian. If you were to visit, the whirring cranes and tractors driving containers to and fro would be the only things in sight.”

“Caofeidian is set to become the world’s first fully autonomous harbor by the end of the year. The US-Chinese startup TuSimple, a specialist in developing self-driving trucks, will replace human-driven terminal tractor-trucks with 20 self-driving models. A separate company handles crane automation, and a central control system will coordinate the movements of both.”

Really like how they minimize wear to the pavement by just shifting the location of where the robot operates each time through a cycle.

This is basically pavement wear-leveling. Wonder how much more life they can get from this.

I would also imagine the next generation could include using thinner pavement where they know that the robots would never go through, as this should drastically save on material.

I'd say probably a lot depending on just how much they move the containers around. For an off the cuff calculation considering just the wear from containers we can take the support pads as being 6 inches (larger than real life) there's ~78 pad lengths between it and the far end pad that are untouched in an unmoving stack and if we move it sideways as well there's ~14 so that's just shy of 1100 spots each pad could rest that aren't used under a static container stack. This does ignore the wear from the tires of the movers but it's a quick illustration of all the space underneath the containers that can be wasted.

This planning has hidden costs for other operations though. At my university they where not allowed to move certain equipment whole or even at all through some corridors, as that sort of thinking on the new building had rendered it basically unsafe for usage.

If in doubt over engineer. When handling use cases be in doubt.

I was lucky enough to tour the equivalent facility in Australia. They learned pavement wearing the “hard way” realizing the robots were much more consistent than humans. As such they would actually shift the entire facility a little bit each day (as movement allowed).

One thing we have barely started uncovering are these “2nd order” problems that come with automation.

I first heard about this with plane automatic landing systems, where the planes would always touch down at the same spot on the tarmac. Randomization needed to be added to keep the tarmac from cracking for longer.

yes, I spent some time on google trying to find the article but I guess my googlefoo is failing me today. However I do remember it was the autoland system developed by McDonnell Douglas on the MD-11. They had to add some small randomization to keep if from always touching down on the same part of the pavement all the time.

That'll be a useful technique for autonomous trucks and roads, because we're going to have the same issue writ large. As long as they don't get too close to other traffic because they're focused on wear leveling.

The headline's referent in non-video form: https://en.wikipedia.org/wiki/Straddle_carrier

But note that this video is talking about "A strads" that is, automated straddle carriers, all the "boring shuffling" is done by automated carriers that can find a container in the yard, and do the tedious work of taking all the containers on top of it off one at a time, then taking it where it needs to go and finally if necessary putting the others back. But meanwhile humans are bringing new containers into the yard with conventional carriers, at least for now.

It's interesting how localization is still among the biggest challenges, despite some great new technology like UWB (for instance, https://www.decawave.com/product/dwm1000-module/). Visual SLAM is obviously difficult in an environment where most of the visual mass (containers) moves.

From what I understand, in robotics SLAM is still one of those areas of development that still has a lot of room for innovation, if you have the chops for it.

They are smart to use specialized sensors here (not sure which spectrum) to triangulate. You can get a consumer version of this in some floor cleaners now! [1]

[1] https://www.amazon.com/d/Robotic-Vacuums/Mint-Automatic-Floo...

After some digging, it appears the particular tech they are using is Locata[1], which uses a time of flight based positioning system.

To deal with the terrible multipath problems in a site full of metal boxes, a phased array approach is used to so the receiver antenna is directional and electronically steerable. Locata have a clever way of doing this called vRay [2]. This uses time-slicing rather than many simultaneous receivers to cut costs, and you can see this ball of smaller antennas on top in several shots.

In contrast, the ultrawideband (UWB) tech you have linked uses the frequency specific nature of multipath to average out the error, but as UWB is over a range of frequencies used for other purposes, it is very limited in legal TX power, and so I suspect it would have range issues in this application.

[1] http://www.locata.com/technology/locata-tech-explained/locat... [2] http://www.locata.com/article/vray-antenna/

In the port of Rotterdam there has been a highly automated container terminal since 1993: the ECT Delta Terminal. See https://myservices.ect.nl/Terminals/rotterdamterminals/delta...

That's a pretty cool "physical-world" use of the best kind of automation: automating the mind-numbing repetitive bits so you can get more people on the fiddlier stuff.

That's a lucky coincidence for now, not the reason for doing it. Historically, automation did the opposite and put human assembly line workers on mind-numbing repetitive tasks. Robots these days often do more skilled tasks like welding, painting, and machining while humans do simple assembly. The interviewee in the video suggests that the balance could shift in the future too, presumably to eliminate all the human crane and straddle-carrier operators.

I was wondering why they used EM instead of HF audio for positioning. Sub-nano must require expensive clocks and very tight tolerances. They need centimeter level precision not sub-mm. Maybe the speed of sound varies over 100m ranges?

Why not do embedded magnets or wires in the road?

I don’t mean to judge their decision, just wondering how they settled on EM for positioning.

Sound localization is directly affected by wind, because the propagation speed is relative to the moving air mass. A wind moving at 0.05 Mach (35 mph, not unlikely in an exposed port) will scale the distance measurements by 5 meters over 100 meters.

Could they do the positional shifting to prevent wear and tear of the asphalt surface with embedded magnets or wires?

Seems like no problem, if they merely designed it so that instead of using the embedded positioning devices as cargo placement markers, they used them provide accurate carrier locations, then tell the carriers to put the cargo in different positions as needed.

I’d say having landmarks (like literally spray marks on the land) and wheel encoders should be much more robust. Perhaps also with some ToF system as a secondary source of positioning.

Come to think of it, could the ”we’re only semi automated” be a way around the unions?

> I’d say having landmarks (like literally spray marks on the land)

Such marks could wear away, or be obscured by spills, debris, or rain.

Ideally, they should be included, along with other sensor systems for positioning, just to give another set of data points for the localization algorithm to work with.

Practically, though, they probably aren't, because it raises costs both for development and maintenance of the systems.

Any system has failure modes. The advantage of a visual system is that it is also useful for human beings. Especially given how the original article talks about semi-automated settings, this type of collaborative approach could prove fruitful. It sounded like ToF did not work very well for them across the entire environment due to lots and lots of little steel cages (i.e. cargo crates) too.

Phase information is already used by precise GPS receivers. The electronic components and software are probably much cheaper than wires in the ground everywhere.

Also, the guy did say they have centimeter precision, not sub-mm. cm is the wavelength of microwaves. The nano referred to nanosecond timing, which is common in every computer these days.

These kinds of suggestions of using a simple mechanical solution instead of a sophisticated one often come up when high tech machines are discussed. Driverless cars should use wires in the road, or communicate with each other using radio instead of vision, SpaceX booster landings should use towers and wires to guide it the final few meters, walking robots should use wheels, Boston dynamics' Handle robot should be a conventional 4-wheeled vehicle with a robot arm on it, aerodynamically unstable aircraft that depend on computer control should not be made, etc. I think people have a bias that says "Tasks I could imagine doing myself are cheap and tasks where I'd be out of my depth even knowing if they're possible are expensive.". But really, hardware is expensive and software and electronics are cheap, at least by unit costs.

Wondering why they chose straddle carriers over automated cranes and trucks which seem to be more common in other automated container terminals.

They were already using straddle carriers, and a mix of cranes and trucks would probably require a larger footprint.

The robot in question is Konecranes A-STRAD

Reply on news.ycombinator.com