Hacker News Reader

Comment by blehn | Hacker News Reader

Surely there are ways to improve and simplify the pinsetting machine without resorting to strings? I suppose it's not a glamorous engineering challenge but this feels like a pretty lazy "solution"

View on HN

For such an old problem that's clearly had a lot of incentive to reduce cost, you can be sure people have tried really hard to improve it. The fact that they're resorting to a lower tech solution to save maintenance costs shows how hard engineering for reliability is.

When I was working on trains, I was surprised how old fashioned everything was. It hadn't really changed in 100 years. But once you see what an absolute beating they take continuously in normal operation, and the high cost of taking a wagon out of service for any reason - let alone the cost of performing maintenance itself, you realize the value in big dumb machines that just keep on working even if they underperform by some less financially important metrics. There are high-tech trains of course, with modern wonders like electricity and even computers, but they're expensive and don't make sense in many applications.

A lot of "surely they could improve that"s end up meaning "surely they could spend more money on that".

> you can be sure people have tried really hard to improve it

I am certain they have not. The reason I'm certain of it is because the injury rate can be reduced to zero, without impacting the game, by just turning the equipment off before anyone goes near it.

That's surely a matter of the operator's health and safety policy requiring lock-out/tag-out before working on a machine, just like happens in every factory.

I see what you're saying, and I'm sure mechanical engineers have been involved who know what they're doing. It might not be as simple as it seems.

On the other hand, sometimes a product really is behind the times. This is probably a relatively small, niche market. Most of your sales probably happen only when a new bowling alley is built.

Outsiders might not want to invest in trying to break into the market. It's expensive to create a new design and set up a factory. And then you have to convince buyers to buy from you, the upstart, instead of companies that have been selling pinsetters since the 1950s. And if you succeed, the reward might not be that large.

The established companies might have a perverse incentive against innovating. The machines seem to do the job but require a good amount of maintenance, so the manufacturers might get a nice, steady stream of income from replacement parts and/or service contracts. Even if they are interested in updating the design, it's probably a safer bet financially to make incremental improvements instead of a radical redesign. (Why rock the boat if you're making money?)

None of this means necessarily means that the traditional pinsetter isn't as good as it can get. I just think it could either way.

Now that I think of it, if a better/safer/cheaper (non-string) pinsetter is possible, ironically maybe the threat of string pinsetters could be what brings it about. If the traditional pinsetter manufacturers see their business threatened (for the first time in decades), maybe it will force their hands and they'll take a risk on a radical redesign.

I think they've been improving them incrementally over the years, but it's a monstrously difficult engineering problem to do the following:

- Set up all 10 pins reliably and in the same spots

- Lift remaining pins after the first throw and sweep away the ones knocked down, if any

- Load extra pins into a holding area to make re-racks faster

- Other things I'm not thinking of

I encourage you to view this 3d breakdown of how a pinsetter works on YouTube: https://youtu.be/Iod6uwUGM2E?si=rTGYSd6CGHp94etz

The mechanism for candlepin bowling seems much simpler: https://en.wikipedia.org/wiki/File:Candlepin_pinsetter.webm

I fear this is going to be a very naive HN comment but: could a robotic arm and some computer vision do it? It seems like a very constrained environment that would be well suited to CV.

I guess it could but you're running into a problem analogous to CPU vs FPGA. A more universal machine will always struggle against the cost and efficiency of a specialised one-job machine.

That seems plausible, but speed is probably an important requirement. Bowling alleys often charge by the hour, so who wants to go to a bowling alley with slow machines?

The traditional pinsetter machine can sweep away pins in one operation, then set up all the right pins in one more operation. A robot arm can presumably only grab one pin at a time. You probably need multiple robot arms, which adds to the cost.

Yep, not to mention the space requirements. It's hard to imagine fitting several robot arms plus any associated machinery into the relatively limited space at the end of a bowling lane.

Hmm, maybe a hybrid approach would work. The traditional pinsetter has 10 pins ready to go in a holder that lowers them all down from above. That part seems simple enough. The complex part is creating order out of chaos and loading pins into that holder.

Suppose you keep that holder, and then do this:

(1) After the ball comes through, take note of which pins are standing, then immediately sweep them ALL away into an area back behind.

(2) The holder, which already has 10 pins in it, lowers and releases 1-10 pins as appropriate.

(3) While the bowler is taking their next shot, the robot arm works on picking up pins from the back area and refilling the holder.

This way, the robot arm can work in parallel with the bowler taking their next shot. If the bowler is really fast, the robot might still not be able to keep up, but at least it doesn't have to do all its work super quick in a narrow window of time.

Reply on news.ycombinator.com