Augmented Reality Welding System | Hacker News Reader

Augmented Reality Welding System

millerwelds.com

110 points1970-01-0195 comments

This is so extremely underwhelming that just a "meh" really doesn't cut it.

What would really help welders would be actual augmented reality welding - like you have a monitor in your helmet which shows you the environment in normal daylight with high contrast and also shows you the puddle under the arc in normal daylight with high contrast - but doesn't show you the blinding bright arc itself (instead it shows a convenient indicater for where the arc hits the surface and whether you are too near or too far from the surface with the gun).

Basically welding as if you had eyes capable of filtering out the bright arc without any reduction of contrast or brightness anywhere else. For extra satisfaction add some microstabilization to the gun, so it can help you keep the tungsten needle at the optimal distance to the surface despite your hands becoming more jittery as you age.

I've wished for this product for years.

Steve Mann created a prototype yeeaarrss ago that looks incredible, but no one has taken it to the commercial stage yet: http://wearcam.org/mannventions-password-stefanosmannaz13/ma...

The link mentions that the topic is mined with a patent. Sadly makes a lot of sense for no one even daring to think about innovating in that field then.

edit: deleted

I am not a welder.

That being said, I'd imagine the active circuitry required to capture and process the images in realtime (or 30fps, in this case) is expensive and bulkier than what's currently used. Processing stereoscopic 2048x2048 HDR streams is no easy task, you'd basically be looking at a VR helmet once you consider where to put battery and compute unit.

It's a cool invention, but probably one of those "analog computer" ideas that's harder to scale/apply than the cheaper alternatives.

welding helmets can exceed $1k already. and at a working minimum of $80/hr, if you can improve quality and reduce time (rework is the worst thing here - it never comes out right, takes forever, and so its often more cost effective to start the work from the beginning), the economics might easily work out..especially at the higher end

Welding helmets are bulky, VR headsets are similar. Lets parts in a helmet but the helmet needs to protect your face and neck from sunburn and potentially bits of metal (depending on your position).

And any processing bulk could be belt worn or even sitting on the ground connected by a cable: you're already tethered to the welder and buffer gas tank after all!

Obviously the high dynamic range is a challenge, but besides that: car reverse camera's cost a few bucks, I don't think this would need to be Way more expensive. It also doesn't have to be stereoscopic and 2048x2048 I'd say... - it would quickly be a step up from the current solution which only allows you to see the welding arc

A car's reverse camera doesn't have the dynamic range to filter headlights at night, much less an arc that's hot enough to give you a sunburn.

You could just use two cameras, or even four, or eight cameras! Each with a different level of filter. Camera sensors are cheap. Degradation of the sensors on the "bright" cameras would be an issue, however.

I mean augmented reality has been extremely hard even for normal uses without the extreme environment of welding involved. Even Apple still hasn’t nailed it yet according to reports. Hard to see how a welding company could succeed where Apple and Meta struggle.

AR is used today in automotive manufacture and repair (Bosch), and surgery (Mount Sinai, John Hopkins). There's also military applications, or construction. Meta is not, by a long shot, on the forefront of actual uses of the technology. Their hardware, and ease of use is top notch, no question about it.

But there are real, and pratical use cases out there done by smaller companies in specialized fields. The issue lies in creating something with mainstream appeal, which I do agree with you, that's a whole other story.

It looks like the Bosch solution is an AR iPad app. I was really thinking about the challenge of developing a full headset for welding. https://www.boschaftermarket.com/gb/en/diagnostics/solutions...

It wasn't totally clear, but it looks like Mount Sinai might be using Microsoft hololens.

It does seem like John Hopkins has something interesting.

I am not saying it is impossible, but I certainly would not say I was "surprised" Miller welding has not developed full AR glasses for welding. AR is hard and this is early days.

I'd imagine a thermal camera could be useful too.

I'd be fascinated to hear about attempts. It seems like that the welding spot would be bright enough to overwhelm the sensor making it difficult to get anything out from it other than white

This is definitely done in the research world. The flash/brightness from the arc isn't an issue because it's a different wavelength (UV vs. IR).

certainly in the immediate vicinity of the arc. the secret would be to accommodate for that in the exposure. you're also going to throw out the most energetic parts of the spectrum with a filter.

idk if thermal would be that useful for steel, especially if you can clearly resolve the boundary of the melt pool - but for high conductivity metals like copper and aluminum, it might be a real godsend. the problem with these is that since they wick heat away, you need to build up the temperature of the whole piece until its close to melting, and then just push it over with the arc. its pretty common to take it too far and have a big section of work turn into a slaggy mess.

It's going to have mount a headset on a already bulky face shield and do a better job than the existing solutions. I'm just thinking about how hard it can be to get a vr headset aligned- with two free hands. Many people set up and then nod forward to get the shield down.

I could see a solution that provided off bore data/feedback but that could probably be managed with a mini led projector or just leds since it's so dark already.

Agreed. For training purposes this software could be integrated in to the AR you describe so the user could get immediate feedback on the quality of their work without an instructor.

To me the holy grail of AR would be expert level overlays for construction or repair jobs. If I could wear AR glasses and see part removal order, torque specs, and reassembly order I could do any mechanical repair job, provided I have the tools.

I saw some video where someone made welding mask with smartphone instead of dark window and using just camera preview, surprisingly great results as it was able to darken image enough that you could clearly see through arc. probably doing some software magic could help with part where you also could see environment around

I agree - I imagine that welding is generating all KINDS of crazy UV emissions that get through the filters and eventually kill your retina. Add the fact that welders are probably "second-hand smoking" while welding and I think there's a real opportunity to help society.

Nah, with good filters, the UV radiation doesn't reach your retina. And modern masks have filtered ventilation too (in addition to workplace fume extraction if working indoors). But you need to focus on the workpiece, the gun, the puddle... the whole time. The contrast you get through that filters is shit - can't really see a lot apart from the arc and the puddle (especially if you are older). Being a welder guarantees a lot of eye strain. It isn't like they actually get damaged, but it can be very unpleasant nonetheless. And all that just because of that ultra-bright arc. With augmented reality, you could just see all the things like if there where no filter and the arc wouldn't be there.

I wonder what could manage that... maybe some kind of sensor fusion with ultrasound combined with dead reckoning and positioning and attitude sensor in the gun?

Nah, it's plenty enough to use two (or more) cameras where one has a hefty ND filter, then you use an HDR algorithm to fuse the images.

I suck at welding. I have a tig welder, and I’m not good at it. I’ve taken a short course, but it’s not enough.

As someone who needs more education and practice, I don’t understand how this can help. Like, isn’t the only way to learn how to weld actually welding? Can a simulation be good enough? I have my doubts.

Sure this can simulate holding stuff and what would happen in some conditions.

I doubt this makes a big difference because the things that screw up welding isn’t just the motion. It’s the prep and cleanliness of the material, the shape and cleanliness of the tungsten rod, the position of the rod in the cup, the downdraft, and all kinds of crazy details that I don’t think can be simulated and these things need to be hard won by real practice.

I'd imagine this tool is meant for welding schools who want to be able to do something like:

1) Pull intro classes out of their shop to increase capacity 2) Reduce costs associated with power/gas/metal/etc. 3) Give people a taste of what welding is like without having to do safety training

Only someone who hasnt welded would think this is something amyone wants.

Welding is a very "real world" activity. Cleaning things properly takes a non insigificant amount of time and you really need to get a feel for the materials.

I'm pretty sure this thing is exclusively for people who haven't welded before...

Yea and it wouldn't get them any closer to actually welding. Just have them... Actually weld? It was pretty easy for me to pick up multiple techniques in high school, no augmented reality system needed

I think getting them to learn how to do the repetitive motion properly without having to actually weld or get input from an instructor to know what they're doing wrong (angle/speed/distance/motion) would speed the process of teaching up a lot.

Technology can be used to augment and speed up human cognition and learning. Take it as a guide, not as the end all be all tool.

I think that if you made people "get good" on the simulator before they touched the real thing they'd get better at the real thing much faster/cheaper.

I think you overestimate the expense of scrap metal for laying practice beads. This simulator is waaaay more expensive

Wire and gas costs aren't the issue. It's the cost of the welding machine because only one person can use it at a time. You can buy multiples of these for the cost of one industrial machine.

I'd imagine that space requirements are pretty significant too. I've never taken a welding class so I don't know how it's normally done, but it seems like having thirty people in a room, even a big one, would be pretty tough to do while maintaining appropriate ventilation and without people arc-flashing each other. Whereas I could see having thirty people in a normal classroom using this augmented reality system. Or using it while they wait for their turn at a real welding station.

Being able to practice at-home seems like a win too. You don't have to booking time at a busy workshop if you rented the VR system as part of your class. You need to finish X hours in the sim and Y hours in person in order to graduate, but if you can do the X hours at home, at your leisure then it becomes that much easier to finish on your schedule, boosting completion rates and skill level.

Does a welding school not have multiple welding machines already? Are they buying new machines every year?

Is it actually less expensive than a MIG welder from the same company?

https://www.millerwelds.com/equipment/welders/mig-gmaw/mille...

That's why I specified an industrial machine. No fab shop is going to use a 20% duty cycle machine. Granted, you can weld just as well with a cheap machine.

Why would someone learn to MIG weld on an expensive industrial machine? You can get a cheap one at Home Depot for $200.

A trade school is not an industrial fab shop

I've worked with the sub-$1k 200A knockoffs and they are quite usable

Hell, Harbor Freight sells really shitty welders for less than $150, so if someone was looking to practice at home, for cheap, with a slightly different system, it seems you could get a lot of mileage out of that for cheaper than a $3,000 VR system, or having to fight for limited time on an industrial grade system. Obviously it's not the same thing, but practice practice practice.

If you're teaching people, you give them crappy little stick welders. $200 each, maybe.

Ah this must be why I never really figured out how to weld in high school.

That's how I learned, and oxy acetylene

The consumables are basically free. The cost is all in kitting out and compliance of running the hot work facility. Lots of nickels and dimes to add up.

If I'm paying for welding classes it's damn well going to be on actual gear.

If you need (1), you're doing well enough to rent a new space. If you need (2) and want to offer (3), you probably never welded before and people should stay the hell away from your course.

You learn welding by actually welding, and getting it spectacularly wrong for a while. Sometimes quite a while. Sometimes forever.

Reduce costs by charging another 3-4k for new equipment? Brilliant

in the most constructive way possible - you're just being lazy

it just takes practice as you say, and its the kind of practice where feedback is good and nearly instantaneous. it also goes pretty fast. by the 40th hour of welding you should be doing a passibly good job.

yes, you will have to struggle some with issues like bridging, undercutting, porosity, and under-penetration. all those are tractable and pretty easy to debug.

and yeah, if your major problem is cleaning the work or your tungsten, then maybe welding isn't for you...but I kinda doubt it

finishing is where you can really distinguish between the professionals and the weekend warriors.

yea, obviously learning is doing. but this looks to me like a product targeted at trade schools, to be used in the very very first stages of an introductory class. instead of using workshop space and materials to do intro concepts it can be done in a normal room, freeing up the workshop during that time slot for other uses.

schools are a continuous pipeline of new students to graduates.

IDK I’d say high school. I took a semester welding class at community college and we only spent 3 sessions on MIG (what this appears to be) before moving on to other processes. Most of the learning was around dealing with PPC, listening for the sizzle, identifying a good bead and common mistakes. This seems expensive for a one-class tool, especially since they still need the regular welders to complete the training.

Hi, as a fellow amateur tig-er, I'd highly recommend checking out Weldingtipsandtricks on YouTube, the guy is a great teacher, and there are plenty of troubleshooting topics for new welders. He helped drastically improve my aluminum welding

Hell yea Youtube U.! That's how I learned to weld as well, saw an open box arc welder at harbor freight and started watching and sticking rods.

Meltin' Metal Anthony has a good channel, though he's gotten super crass the past year (maybe i didn't notice it before?) but still relevant.

WeldTube and their videos are good too, those guys are great at teaching and are good for a laugh sometimes too

All those above can be seen on youtube. I don't tig but have watched a lot of videos on it and there's so much to learn from what's out there

It looks like this AR welding product from Miller is specific to MIG. A TIG simulator would be interesting, but it would probably be a lot more work to pull off convincingly.

(For the non-welders reading this: with TIG welding you hold a "torch" in one hand that has a tungsten electrode and it blows an inert gas such as argon at the work area, and in the other hand you have welding rod that you periodically dip into the pool. MIG, on the other hand, uses a torch where the filler rod is the electrode, and there's a motorized feeder that's constantly extending the filler rod as you weld.)

edit: It looks like Miller do also have a virtual TIG setup, someone else posted a link to it. It looks pretty good, actually.

My father was the director of a vocational school and for the beginner those system seems to help.

I myself took some lesson for tig - as other have said, a big part of it was cleaning up and prepping. Prepping the tungstène every minutes made you motivated to improve fast :D

With practice, I can weld in some complicated position. The inconfortable way of working is the biggest issue (under a tube with a mirror for example, with my left hand holding the electrode as a right handed man… ). Nothing can teach you that beside doing it. And it will happen, only more often in a boiler room :D

My first thought watching this video was that its primary use is probably going to be weeding out people who would never be good at it e.g. because they lack a steady hand or lack hand eye coordination.

In my experience (learning aluminum TIG at a hobbyist level) having a "steady hand" is all about setting things up from the start so your torch hand always has something stable to rest against.

There will, of course, be people on either extreme who either have a stable enough hand that they don't need to rest on anything, or who have hands that are so shaky that propping against something doesn't help, but I think for the ordinary able-bodied person it just takes some practice and careful attention to workspace setup.

I've never done MIG, but I imagine it's a bit less of an issue since you aren't constantly trying to get the electrode as close as possible without ever touching. I could see this product being pretty good at telling students "hey, you're holding the torch at the wrong angle" or "you're going to fast" or whatever. Basically, teaching students a few good habits from the beginning.

That seems... great? If I can buy a Groupon and play for a couple hours on a sim for $100 and figure out if I have the aptitude for it, and if I enjoy doing it, before dropping a couple grand on a class I won't finish and will refuse to pay for, then that seems like a win for all parties.

Yes, it seems like a good idea.

These aren't used as teaching tools so much as "demo" tools. They are the rough welding equivalent to running a flight simulator on a laptop with a single USB joystick. They are good for bringing out to trade shows, job fairs, and other marketing venues to sort-of simulate what it's like to weld something, with all the dangers removed.

The only advantage i see for this system is that it eliminates the liability of students working with real welding equipment.

I learned to weld in high school. The first priority is safety education. While this is neat, that is a very massive missing element that students must become familiar with when learning to weld.

I'm an amateur welder too. I suspect this is helpful for bringing down the cost for initial training. You'll still need hands-on, but this will get you the first 40% to 60%.

Beyond the welder and material for training, facilities also need to have robust electrical supplies, welding tables, and inert gas supplies. This cuts all of that out.

Is this a $3000 app with some repurposed plastic pieces and a booklet of stickers?

If so... I love it.

I'd be curious if anyone has experience with the system (and with welding) and could describe the feedback that it's able to provide. I rely so heavily on the sound of the bead being laid, and I can't imagine this can even semi-reliably convey the same vast amount of information as you can glean from your ears.

I learned welding in high-school so I'm far from a professional. But I agree you get a lot of feedback from the tactical sensation of the stick / filament and the sound, which would be hard to reproduce in an app.

I've heard from people who have used this VR welding system that there are some key differences between it and real welding, notably for the stick welder version. To simulate the stick being consumed, a servo motor pulls the stick through the back of the handheld part. This results in the balance of the handheld part changing noticeably as you weld.

Haha, man. Just imagine how much an iPhone would cost if Apple made a few thousand of them per year instead of however many they're cranking out now

Pretty neat, looks like it would let you teach in a regular classroom, and would cut down on materials costs by a huge amount.

Would be cool to see what a tig teaching system would look like

>Would be cool to see what a tig teaching system would look like

The TIG setup: https://youtu.be/xCbmFFPSF7o?t=77

Ok this changed my mind about this tech. Tig welding really takes thousands of hours to learn, you do get limited by how much material you can burn through

Oooo... That'd be fun. I could see there being a bit more benefits with a tig system like this, insofar as being able to spend more time on the technique and getting feedback from the system on your timing and consistency.

Materials costs? Honestly? It would take decades, what kind of materials do you think starting welders are practicing on? Grade2 titanium?

i mean, even steel isn't _that_ cheap. not to mention Al.

I'm skeptical of the product myself but there's a lot of materials/equipment for welding: grinders, PPE, physical space to do stuff, cutoff saws (I want an Evolution cold cut saw but it's not worth the price for me right now), consumables, gasses etc.

Edit: but yeah all those are likely less than the "affordable" MSRP of 3150 clams - i've always been curious who Miller's target clients are, in my experience they have great tech and products but are super expensive. YesWelder is a better alternative in my opinion fwiw

Schools already have the equipment, scrap steel isn't expensive and most trade schools with metal shops have a bunch lying around. Sure blades and grinding wheels need to be changed out sometimes but for an established program this product seems rather silly to me. The student gets the absolute minimal exposure with basically no real practice

Anyone practicing on aluminum has already been welding for a while so again this product would be useless for them. I don't think any reasonable program would start out new students on aluminum

I agree and am still trying to determine the target use case here. Maybe Miller is so flush they developed it and don’t have much to lose, if people buy into the ecosystem good for Miller.

IMHO manual welding is interesting but ultimately robot arms are coming down so far in price in the near future it's soon going to be more useful to just straight up train people how to specify work for the arms.

Often really substantial experience is required to approach welding problems, including interpreting order intent (quantity and material related concerns such as jig and fixture design, adjacent process speed and cycle time affinities, likelihood of follow-on orders, etc.), interpreting drawings (often made by non-welders and potentially requiring feedback) plus understanding finished workpiece intent and loading, intermediate work holding and handling, welding process selection, shield gas and flow rate selection, manual dexterity and precision, bead material selection, optimal power levels, temperature monitoring, feed rates, environmental safety and pollution concerns, questions of finish including scratch prevention, grinding and polishing, cleaning, and potentially subsequent surface treatment, inspection and testing, etc.

If anyone wants to work on automation for these sorts of fabrication related problems, we will be hiring in a few months.

Doesn't that really only apply to mass production, a LOT of welding is one-off (or few-off) work used to connect parts (which may have been mass produced).

If you mean to actually weld one off parts-- is the sensing technology in place to accomplish that? The speed and feed you go at depends on the ambient conditions and part geometry. Even if so, you have the problem that much welding is and must be done in-situ because the assembly is not transportable. In some cases you could bring in a robot arm, but even where you can that seems like lot of effort vs just sending a welder in.

Traditional manufacturing is described as "low mix, high volume" (ie. same part in great quantities). Modern manufacturing tends towards "high mix, low volume" (ie. different part each time).

This is partly because, ultimately, it is more cost efficient to produce multiple parts using similar processes on the same equipment than to produce all parts on dedicated equipment using a broader mix of processes. It is also partly because outsourcing fabrication makes sense for many people, which means the fabricator themselves is receiving a range of orders that has to be executed in rapid succession on a given equipment line.

While this transition presents a challenge for everyone (design, management, HR, etc.) it is also a huge opportunity, particularly for we software people.

While it saddens me to hear what you're saying, I literally left my career as a blacksmith for embedded because even the interesting bespoke projects are all going to CNC. I'd be interested in your project, sounds like it's a rare intersection of my domains of interest

Awesome, please send me an email. As well as fab, we do a lot of embedded, too.

Like other comments, I agree that the technology leaves something to be desired. However, I learned welding from a community college in high school. I wasted A LOT of welding rod/wire and metal just getting the pacing and angles correct. I could imagine a tool like this helping someone grease the groove without so much wasted resources.

Looks like a training helmet/system.

It seems like a lot of weld quality is evident through visual inspection. It would be interesting if you could use computer vision to rate the weld in near real time to help identify if the setting might need to be optimized before needing to redo the entire seam.

It does provide real time feedback, apparently. The product video shows this and scoring capabilities in the software.

But for a make believe weld, not a real one.

Sure but if the system can get your arc length, work angle, travel angle and travel speed dialed and consistent before you even plug in a real welder that's a decent increase in the throughput of a vocational facility with a given number of welding stations because now a fraction of the training can be performed in the classroom.

It'll never match the muscle memory. Practicing welding isn't that expensive either...

After the safety courses you're never in the classroom. You should always be behind the stick with an instructor.

It really only takes a summer in the barn to learn how to weld.

Just put in 100 hours or so and you're good. 10000 and you're a pro, just like any trade, including programming.

It doesn't need to match muscle memory. It needs to lop off the first 10-20hr of actual welding where the instructor runs from booth to booth telling people to correct their arc length, angles and speed thereby increasing the efficiency by which welders can be trained.

This is cool from a tech perspective but the rest of me is cringing because there's simply nothing left untouched by tech. I love welding because it's so dang "real".

Makes sense from a teaching perspective (I guess) - certainly nice to cut down on material and consumable costs, and cutting and grinding materials.

What about pipe though? It seems this product only offers flat stock welding which seems like a Major limitation for real world welding practice.

Truly hope the "feature" of rod sticking and undercut are built into the software - those are major things to learn when starting out

Edit: guess the product is more MIG/wire feed focused so maybe rod stick is not relevant here

There is nothing like the smell of hot, ground metal to remind you of the heights of career achievement possible in the industrial Midwest of America. A simulator will never give you that holes-burnt-into-your-shirt feeling.

If you can weld, and you can code, the world is your **ing oyster. One of the two will pay your bills.

I got to play with this welding simulator at a Toronto Transit Commission (TTC) open house last year. Seemed cool.

Welding is very physical, this simply won't work. In stick welding (the only kind I can do), if you don't pay attention to the feel of the rod, you'll get stuck. This is playing a welding video game. That could be useful for certain extremely limited circumstances, but this is not it.

I wish it would simulate the heavy cabling/tubing that is attached to the welder. Can have someone simulate that by randomly tugging on the back of the welder to simulate the cable getting caught on things.

Unnecessary.. What would be nice though is a helmet with a camera to replace regular auto-darkening hoods. Maybe with bluetooth so you could rock out on the job, maybe some cams in the back for fire detection.

It strikes me that no one pointed out to their product managers the liability (kernel sources, packages, updates, et al) that is shipping an ~~Android~~ mobile device as a part of their product.

I thought this said augmented reality ‘wedding’ service and began wondering how that would work out…

Reply on news.ycombinator.com