Hacker News new | comments | show | ask | jobs | submit login
Tesla is holding a hackathon to fix two robot bottlenecks in Model 3 production (electrek.co)
85 points by evo_9 7 months ago | hide | past | web | favorite | 48 comments

I love electric vehicles, however Electrek is a very pro-Tesla "news" source. The owner is a massive fan and finds the most positive spin for Tesla on essentially every story. It's sort of sad that it's one of the main electric vehicle news sources on the web now, since it's basically Fred's I Love Tesla blog.

Aside from making it easier to get clicks, I think that's just the enthusiasm and the fact that Tesla is the only game in town right now. From what I've seen, when Electrek criticizes other automakers it's because they're not doing enough for electrification. There are many different reasons why automakers themselves are not eager for EVs, but that's what Electrek is judging them on. It's no surprise that Tesla is the standout.

> Tesla is the only game in town right now

Well, if you get your news from sources like Electrek, you'd think so.

Yet here we are with both Jaguar and very soon Audi selling fully electric SUVs that are competitive with the Model X on both features and price.

And in the second month after launch (March 2018), Nissan sold 11 400 units of the new (150 mile EPA range) Leaf; next year they have a 230 mile range model coming out. It may not be a Model 3 killer based on range, but they sold 300 000 of the old model, so it's no compliance car.

Nissan also currently has a monopoly on fully electric MPVs (e-NV200).

These may not all be sexy cars with ridiculous modes, but if they help save our planet, I'm all for it.

I think what you'll see with a lot of competing EVs is the same thing that happened with GM's Bolt. Decent enough car. But that's about it. The problem is that it's not enough to just make an EV. Wait until lower-priced variants of the Model 3 are being produced to see how outclassed the Bolt is, and wait until Jaguar and Audi's SUVs are out to see how they really compare. Chances are it won't be favorable for Jaguar or Audi, but I'm open to being surprised.

The first genuinely serious EV from another automaker in my opinion will be Porsche's Mission E, but take note of how well it can match the characteristics of a Model S, a family sedan, you can already buy today. Still, just the idea of an all-electric Porsche is badass.

> These may not all be sexy cars with ridiculous modes, but if they help save our planet, I'm all for it.

We're in agreement that any measure of EVing is good. Many of the challenges facing automakers are systemic, so they are in a tough spot even if they genuinely believe EVs are the future.

Tesla is the only one that has "burned their boats" and fully aligned their corporate future with electrification.

On that spectrum there's Tesla, a bunch of Chinese firms like BYD then a big gap before firms like Jaguar, Audi, Volkswagen all of whom are selling products that Tesla can rightly attack as being the past, not the future of transport.

I think every bit helps, including efficiency, smaller cars, public transport, city cars and non plugin hybrids like Nissan's eNote range but Tesla is setting the pace to a large degree at the moment.

Tesla is absolutely not the only game in town, they're just the coolest. They're the market leader in the US, but they aren't shipping substantial numbers of vehicles in Europe or China, which are much bigger EV markets. The Chevrolet Bolt and 2018 Nissan Leaf are serious competitors to the Model 3.

Yeah, just look at those shipment volumes. It's not hard to see how a reasonable journalist might not be that excited about the Bolt or new Leaf. Or vice versa.

I actually read most of their stories and have not yet seen a story that has turned out to be false. Mean while almost every story / comment posted on Hacker News will receive up votes if it has a negative spin and down vote if it has a positive spin, no matter the content. Yes Elekrek is filled with Tesla fanboys but HN is also filled with Tesla haters. See my comment for example of down voted positive comments with no counter argument offered.

Just because something isn’t false doesn’t mean it’s not highly biased.

Biased and true is better than biased and false.

Unbiased doesn't exist so I'm ignoring it.

"Perfect doesn't exist so I'm going to ignore better."

> Unbiased doesn't exist so I'm ignoring it.


> Perfect doesn't exist so I'm going to ignore better.

are not analogous. Now if the line were instead something like

> Unbiased doesn't exist so I'm ignoring less-biased (or efforts to reduce bias).

then you'd be onto something, but as it stands, what you've got there is a false equivalence.

You can do a search for "tesla" submissions ranked by popularity for all time: https://hn.algolia.com/?query=tesla&sort=byPopularity&prefix...

You'll see that Tesla's press events have gotten plenty of extremely positive discussion/upvotes on HN. "All Tesla Cars Being Produced Now Have Full Self-Driving Hardware" has 1,493 upvotes:


I should clarify that I meant the current state of HN. (I thought that was obvious)

No, it wasn't obvious.

But it is obvious you didn't do the search I recommended. Or else you would've seen how recent the most popular Tesla posts are. The second most popular Tesla story was posted just 2 months ago.

I'm willing bet many people (like me!) only downvoted you because you're complaining about downvotes, in violation of the HN community guidelines.

No I am explaining a phenomenon and that was the evidence I provided. Be my guest and downvote to your hearts content I could not care less.

Any news on whether the hackathon is internal or external? If it's come-all-takers, it looks like Tesla is airing the unspoken point of hackathons (to provide the company with unpaid innovation). A bold move...

edit: is, it's going on as we speak

Doesn't the fact that the hackathon is already happening, but we (even in this community) are only now learning about it pretty much answer this question?

If it's external.. wouldn't it attract a lot of competitors? "Hello I'm Hans from Aud... Augsburg!" "Ah you're German too? I mean, my grandmother is German, oh look, see my pen? She gave me a stationery set that says Mercedes Benz on it!".

Even simple things like having two robot arms for the same task with one idle but ready in case the other fails; or being able to electronically label some assemblies as rejects and no-op everything on the down the line until they hit an eject point. Ability to re-insert them again later. On-off ramps with even 20 minutes of buffer capacity would be extremely helpful in building redundancy into the system.

20 minutes of buffer in an auto plant? That's a building the size of a large supermarket.

Buffering space is a big cost. Here's a Coca-Cola bottling plant.[1] This one has an extremely low employee count. It's mostly buffering space. That gives employees time to move to the trouble spot and deal with problems. Since Coca-Cola has been at this a while, they must need all that buffering.

In an auto final assembly plant, you don't have the luxury of big buffers. On a rigid line, there's no buffering at all - if any station stops, the whole line must stop. That leads to quality problems - if someone drops a bolt, that bolt never gets inserted. A few seconds of buffering between stations helps. That's what Toyota's kanban system introduced. If you watch videos of Toyota production lines, they have a big timer showing the cycle time of the last cycle. It's not always the same. Sometimes a cycle takes a few seconds longer. If something goes wrong, like a tool breaking or a bad part, the stations upstream use up their buffering capacity and stop one after another. But the whole line doesn't have to stop for a one minute tool change.

[1] https://www.youtube.com/watch?v=7V6JO0Rfgas

Disclaimer: I work for GM, any opinions are solely my own.

This is enormously context-sensitive, but it is not at all odd to have a buffer with several hours worth of vehicles in auto assembly, for example between shops (ie between body and paint and between paint and final assembly).

As you get closer to the end of the line, the buffers get much smaller, but you may have a 5 minute buffer between the last few lines in Final Assembly.

One other thing to point out here; The two main styles of conveyance in vehicle assembly are fixed position stop and stop stations (also known by many other names).

- Fixed position stop is what you call rigid line, and it is used where humans work on cars. The conveyor runs at a constant, slow speed through constant size work areas called footprints, such that each worker gets about a minute or so to do a set of assembly tasks. Humans can attach parts as the vehicle moves down the line. With fixed position stop, there is no buffer as the vehicle carriers are physically connected to a conveyor chain.

- Stop stations are used where robots work on vehicles. It is very difficult to get a robot to work on a moving vehicle (but possible). Generally these conveyors move about twice as fast as the rigid stations, but all work is done while the vehicle is stopped. Also, each vehicle moves independently between stop stations as fast as the conveyor can move, and there is some buffering between each station.

> 20 minutes of buffer in an auto plant? That's a building the size of a large supermarket.

Yes, but it's in a mezzanine above the work area.

This doesn't inspire too much confidence, Tesla at one point asked for volunteers to ramp up Model 3 production and now a hackathon to solve a bottleneck. We get they burn cash like there's no tomorrow ( pun?) but this reeks of desperation. He should at least pay the winners.

Seems like a better way to solve the problem than doing nothing, or continuing on business as usual.

If rewarded accordingly I agree. But the fact he could get a HUGE payout if certain milestones are met and has the nerve to not only ask for volunteers but now for programmers to fix his issues for free is ridiculous. Source: https://www.cnbc.com/2018/03/21/elon-musk-could-earn-55-bill...

How is this different from any other situation where an employee contributes to the company and the CEO benefits from their results?

The pay disparity between workers and executives is important and worth discussing, but what makes this particular example of it worth pointing out?

Well one, the employee is getting paid to do the duties their job asks for. They aren't working for free.

Tesla/Musk appears to be following the model of massive greedy billion dollar corporation that would rather take free work than appropriately compensate workers/consultants.

There is a massive difference between BEING PAID and NOT BEING PAID when asked to do work. Seeing how Musk overworks and underpays his workers I'm not surprised by this move.

Where do you get that these people aren’t being paid?

From the User you were replying to: lechiffre10

Also if only the winners are being paid, what about everyone else? They've invested just as much time and their contributions were still valued by Tesla. This isn't Glengarry Glen Ross, these are real humans.

Well where did they get it from?

As far as I can tell, this “hackathon” is being done by employees on company time, so I’m extremely confused.

Is the hackathon not during work hours?

When you've taken $20B in funding that's a false dichotomy.

The is no reason to believe that the volunteers and hackathon are open to anybody but paid employees/contractors.

The volunteers email went out in an internal employee list.

"Work free overtime! Win prizes! First prize is a $50 gift certificate at Chili's. Second prize is a steak knife. Third prize is you're fired."

From the article:

> A hackathon is a sprint event where programmers are invited to compete in fixing a problem or creating a product in a short period of time with prize money for the highest performing participants.

“Hackathon” - oh weee I’d love to work 100+ hours to “win” the prize. This isn’t worker exploitation or overworking at all. /s

The headline is evocative of the introduction to the book "The Goal" by Eli Goldratt. That book was transformative for me.

Said it once, will say it again. Elon needs to provide stochastic simulations of his production line with realistic failure rates of each component. Understand where those high failures are and build in the same redundancy he does to the Model 3's frame.

Is it just failure rates, though? He's said in the past a lot of it that they have 30,000 parts and a massive number of suppliers to rely on. For instance, with the model X doors, there were substantial delays because their contractor said it could be done hydraulically, but it wasn't completed, so then Tesla did it itself with electromechanical systems. To me, that massive delay wouldn't be captured by your production line simulation.

Managing suppliers/supply chains, testing their designs, integrating their designs into yours is the major challenge for a manufacturer, IMO. It goes well beyond the challenge of testing and verifying elements of your own design.

some youtube video about the latest conf call said they redesigned the parts to reduce and streamline everything.. but anyway your poind holds

>the same redundancy he does to the Model 3's frame.

If Ray Charles tried to shoot the broad side of the barn he's probably hit the pinch welds on a Tesla in the next town over. Superfluous part numbers (like their multi-piece wheel wells) and material excess is the kind of redundancy the Model 3 has the most of.

Tesla doesn't make efficiently designed cars any more than Juicero made efficient juicers. Tesla has carved out a market niche that they won't easily be unseated from but efficient design is not that niche.

You can simulate and model all you want. If you don't have the expertise in house you won't be able to get much value out of it.

I agree, but for context, remember that the other car companies have been fine-tuning their supply chain processes for the last 100 years. Telsa is by no means a GM or Ford in and as far as supply chain refinements. I imagine GM or Ford wish they had the popularity and customer desire that Telsa has, and we've been witnessing the start of their forte into the mass production and no matter how many experts you throw at a difficult problem, you still need to refine and mature your processes.

I assume Elon meant an internal hackathon here I doubt anyone with the skills needed is going to work for peanuts

Okay, I don’t know what else to say at this point: HN needs a filter function so users can screen out content.

Said it yesterday, say it again today: I’m fucking sick of seeing every tiny little irrelevant thing involving Tesla in the news. It’s STUPID. I bet google’s holding a hackathon to fix something too but I don’t see 45 upvotes on that....in fact nobody’s even writing about it. This is just clickbait bullshit.

HN is on the web, which until the copyright vultures get their way, it's still a platform which the user can mold quite freely. There's no need to beg the site owners for that feature.

Ex: https://news.ycombinator.com/item?id=13467611

Having a filter would improve the feed a lot for me. Excluding Google/golang spam would make this site a lot more enjoyable.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact