3 ms·

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 pin

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 https://youtu.be/Iod6uwUGM2E?si=rTGYSd6CGHp94etz

scj 3y ago
The mechanism for candlepin bowling seems much simpler: https://en.wikipedia.org/wiki/File:Candlepin_pinsetter.webm https://en.wikipedia.org/wiki/File:Candlepin_pinsetter.webm
cameronh90 3y ago
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.
- AndrewOMartin 3y ago
  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.
- adrianmonk 3y ago
  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.
  - duderific 3y ago
    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.
    - adrianmonk 3y ago
      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.

scj 3y ago

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

cameronh90 3y ago

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.

AndrewOMartin 3y ago
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.
adrianmonk 3y ago
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.
- duderific 3y ago
  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.
  - adrianmonk 3y ago
    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.