What's the lowest score that a machine could get in golf?

[Rain Man]

I was at a restaurant tonight, and they had some golf match with Mark Rypien, Annika Sorenstam, and some E-List celebrities. I was watching a putt go wide, and thought, "I bet you could design a machine to putt for you."

It then made me start thinking about using a machine for an entire golf game. Could you design a machine that could account for wind, elevation, temperature, length, etc., and score a perfect 18? If so, could a machine do it consistently?

My initial reaction is that, if we can put a rover on Mars we can probably design a machine that could score an 18 in golf. But then again, you have wind gusts and temperature microfluctuations and tiny little twigs blowing on the ground that probably can't be predicted.

I don't think you could design a machine that could hit a hole in one every time. But I bet NASA or DARPA could design a machine that could do something like this consistently on 18 holes: 2 holes in one, 14 2's (one drive, one putt), and 2 3's (one drive, two putts). That would be a 36 over 18 holes.

But I'm not a golfer, so what do I know? What do you think?

Poll to follow after I spend ten minutes studying the lay of the land and making practice polls.

[Rain Man]

Originally Posted by cdcox

I don't know anything about robots and haven't golfed since I was 16, but this sounds like a fun project.

There has to be a market demand for a golfing robot. If people watch golf with humans, they'll certainly watch it with robots. If we had to, we could probably even make the robots overweight and alcoholic to up the entertainment factor to the John Daly level.

[kccrow]

There are a couple of difficult judgements that humans can make that a robot, to my knowledge, cannot. One is crosswinds down course. A sniper, a golfer, etc, all judge crosswinds down course because they will affect trajectory. A robot, I believe, would only be able to account for the wind around it, not changes in wind down course. At 500 yards, the winds can definitely change.

The other is slope an angles, GPS is good within sometimes inches, but usually a foot or two, and within a few degrees, but I have yet to see GPS capable of calculating exact slope and angle 500 yards downfield. You could pre-program your robot with the course layout via GPS within a certain accuracy, say 98%, but it'll never be perfect.

If he robot does miss, you now have to account for terrain changes. Proximity switches and the like could monitor immediate terrain height, such as long grass versus putting green, but it wouldn't be able to distinguish sand, or water, from putting green. Perhaps infrared imaging would be helpful in making a distinction between hazards, but your robot may be 500 yards from the intended area you want to analyze. If you want this to be a self-contained golfer, then good luck. You're robot would have to traverse the entire course using gps tracking of its location and plot a thermal map before golfing, or something to that effect.

You'd have to put some kind of homing device in the cup for it to recognize location of the cup.

Regardless of the above scenarios, putting, still, would be a nightmare. Pro courses have subtle changes in grade that GPS is incapable of projecting, at least to my knowledge. Ground penetrating radar wouldn't help you here. Radar would never provide you with the accuracy you'd need. You basically need airborne laser scanning of some kind, which is difficult given your self-contained golfer.

I don't know, what to say, I'd be hard pressed to think a robot could break par. I'm sure there is technology out there I know little about, but I can't think of anything that would make it realistic to expect the robot to break par, even with pre-programming of the course.

I'd venture to say if you had a par course of 72, your robot could get within 80%, or a 90 with as many factors taken into account as possible.

[cdcox]

Originally Posted by kccrow

The other is slope an angles, GPS is good within sometimes inches, but usually a foot or two, and within a few degrees, but I have yet to see GPS capable of calculating exact slope and angle 500 yards downfield. You could pre-program your robot with the course layout via GPS within a certain accuracy, say 98%, but it'll never be perfect.

My robot has lidar.

[Squalor]

Originally Posted by Rain Man

To help clarify assumptions, let's set the following ground rules:

1. The machine uses current technology.
2. The machine swings a club at the ball.
3. The machine is mobile and can ambulate to the ball at any point on the course.
4. All equipment is current golf technology. Current clubs, current ball, current tee. The machine also sports a natty pair of plaid pants.
5. The machine has no human assistance once turned on. However, it can include any sensors that are currently available - optics, radar, weather, etc.

Gotta give you a Main Ran on this, I don't believe under these criteria that your golfer is no more than a device. It cannot play par.

[cdcox]

Those suggesting that an optimized robot would struggle to make par need to see this, which hasn't really been optimized or equipped with machine learning or anything advanced.

[Rain Man]

Originally Posted by cdcox

Those suggesting that an optimized robot would struggle to make par need to see this, which hasn't really been optimized or equipped with machine learning or anything advanced.

That's impressive by both parties.

I concede that it's a hard problem for a machine, but at the same time we have some wicked good technology in today's world. I stand by my assertion that a machine could shoot in the mid-30s, albeit maybe it would require a non-gusty day to get that score. As long as it can identify the hole, all the other stuff is just physics with a bit of luck.

[cdcox]

Originally Posted by Rain Man

That's impressive by both parties.

I concede that it's a hard problem for a machine, but at the same time we have some wicked good technology in today's world. I stand by my assertion that a machine could shoot in the mid-30s, albeit maybe it would require a non-gusty day to get that score. As long as it can identify the hole, all the other stuff is just physics with a bit of luck.

I think a computer with machine learning is going to eventually be able to assess the waving of the branches far better than any human to judge wind effects. Finding the hole isn't even an issue. AI discovered the concept of a cat by just viewing youtube videos without having any instruction of what to look for. It just scanned random youtube videos and found this repeating image of a cat that was obviously significant. I don't think programming it to find a non-moving white circle in the middle of a green field, or a flag, is going to be much of a challenge.

[cdcox]

Originally Posted by Rain Man

That's impressive by both parties.

I concede that it's a hard problem for a machine, but at the same time we have some wicked good technology in today's world. I stand by my assertion that a machine could shoot in the mid-30s, albeit maybe it would require a non-gusty day to get that score. As long as it can identify the hole, all the other stuff is just physics with a bit of luck.

Not to mention that the robot was holding its own against a pro champ within this aspect of the game.

[Deberg_1990]

Ill bet it could beat Tiger Woods nowdays.

[Stewie]

How does a machine setup and hit a shot like this?

[cdcox]

Originally Posted by Stewie

How does a machine setup and hit a shot like this?

I think the winning strategy for the machine will be to play the odds to avoid the bunker altogether. It is going be much more precise than a human and will have a much more quantitative estimate of the odds of where the ball will end up. If you ever have played chess against a good computer program, it just doesn't open itself up to situations where it is going to put itself in a bind.

[tk13]

I wouldn't be surprised at all if you could build something that would score in the 30s. Completely talking out of my behind, but this is ChiefsPlanet.

I would think wind gusts after the ball is struck would one of the biggest variables. Depends on the course and pin locations though. There are a lot of courses you can't just roll it up there, or stop it on a dime depending because of the green speed or where the pin is located. Would you be able to stop a ball hit 550 yards on a small green?

[cdcox]

Originally Posted by tk13

I wouldn't be surprised at all if you could build something that would score in the 30s. Completely talking out of my behind, but this is ChiefsPlanet.

I would think wind gusts after the ball is struck would one of the biggest variables. Depends on the course and pin locations though. There are a lot of courses you can't just roll it up there, or stop it on a dime depending because of the green speed or where the pin is located. Would you be able to stop a ball hit 550 yards on a small green?

So a tough course might have one or two of those holes. Maybe in scores in the low 40s on a course like that.

[Hootie]

on Tiger Woods 2005 on XBox live one of my Xbox live friends and I built a custom par 3 course, played it online, used a grease pen, and marked our screens on every hole to know where to aim the ball and we both knew how many clicks of spin we needed to hit a hole in one. I shot an 18 numerous times!

!!!

[Hootie]

***CORRECTION***

We knew how many clicks of power we needed for each club, and then when the ball was in the error, we also knew how many clicks of topspin or backspin we needed.

We had lives.