Putting Robot Critique

Putting Robot Critique

Dixon Golfer
Dixon Golfer

August 30th, 2007, 4:02 pm #1

Good morning. Last fall you gave my son, Matt a lesson and at one point during the lesson you voiced a critique of the folks who have so far attempted to construct a “putting robot”. I believe your issue was that the designs that you have seen don’t match the way the body actually moves. Can you elaborate on this? I would like to have a better understanding of how the shoulders should move during the putting stroke.
Incidentally, Matt’s putting has steadily improved since your lesson. Two weeks ago he shot a 64 with 23 putts. We definitely want to get another lesson this fall sometime. Dixon Golfer


Joined: January 1st, 1970, 12:00 am

September 4th, 2007, 4:18 pm #2

Dear Dixon,

Let's start with the brutally obvious:

1. robots are not alive.
2. robots do not have a brain.
3. robots do not move.
4. robots are built instead of evolved.
5. robots are all metal parts and some grease.

If someone makes a robot sort of look like a human, that's cute, but functionally unnecessary. What matters functionally is how the parts are constructed and put together.

For a robot made to look like a golfer putting, the usual design has only two parts that matter -- the neck and the shoulder-arms assembly. The shoulder-arms assembly has a hole in the center that fits onto the "pipe" of the neck, so the shoulder-arms assembly can teeter-totter or see-saw about this neck pivot. Other than this functional aspect, the ONLY thing that matters for the sort of stroke a robot makes is the angle the neck pipe makes with the floor. If the pipe parallels the floor, the shoulder-arms assembly makes a straight-back and straight-thru stroke. If the pipe is angled / tilted up a bit, the shoulder-arms assembly makes a stroke that "gates" or "arcs"

1. inside-and-opening-going-back,
2. outside-and-closing-returning-to-impact,
3. square-at-only-one-precise-moment-which-needs-to-be-at-impact, then
4. inside-closing-past-impact-into-the-followthru.

(A "frowny face" as seen by the golfer.)

If the pipe tilts down off level with the floor, the robot stroke goes:

1. outside-and-closing-going-back,
2. inside-and-opening-returning-to-impact,
3. square-at-only-one-precise-moment-which-needs-to-be-at-impact, then
4. outside-and-opening-past-impact-into-the-followthru.

(A "smiley face" as seen by the golfer.) Whatever the orientation of the "pipe" in space, the shoulder-arms assembly MUST and CAN ONLY move in a plane that is perpendicular to the axis of the pipe. Tilting the pipe tilts the plane of motion of the shoulder-arms assembly, and this tilts the stroke of the putter head.

Examining this a bit, we see that the shoulder-arms assembly moves ONLY AS ALLOWED BY THE PIPE. That's because the shoulder-arms assembly is all metal in one unchangeable shape, as is the neck pipe, and because the "fitting" of the shoulder-arms assembly hole onto the pipe can only be done in one way, so the relationship between the neck and the shoulder-arms assembly is fixed and unchangeable. And once the pipe angle is screwed down tight, that is also fixed and unchageable.

Dave Pelz's "Perfy", the perfect putting robot who will teach you how to "putt like the pros" or near enough, wearing a golf cap (scratch golfer!, too!).

Note especially that Perfy does NOT have his hands directly beneath his shoulders, as Pelz claims is required (more on this below). Pelz appears oblivious to this rather obvious conflict with his teaching, and unknown to Pelz, Perfy actually "proves" that Pelz's claim about hand position determining stroke path is false by making straight-back, straight-thru strokes with the hands outside / beyond the line beneath the shoulder sockets.

Harold Swash's "Rail" -- no pretense that it's human, but the "robotic ideal" is still inspired by flawed notions of robotic vs. human biomechanics (see the level setting of the pipe extending out from the vertical base):

The Putting Arc's "Iron Archie" -- named after "Iron Byron" to personalize the robot, but it's the same deal: set the pipe angle to change the shape of the stroke, in this case, 12 degrees off vertical in order to match the shape of the Putting Arc on the ground -- any slight misadjustment of the pipe and the robot cannot trace the shape of the Putting Arc.

Kramski putting robot, another engineer makes a golfer out of metal and bolts, but I don't think Kramksi uses this robot for teaching and instead restricts the use of the robot to testing variables by controlling and eliminating other variables except the one under study (e.g., putter face loft, MOI, etc., uninfluenced by stroke path irregularities):

A human is "slightly" different. The main thing is that there is NOT REALLY a fixed action between the neck angle and the way the shoulders can move. The shoulder frame and indeed the whole upper torso can move with a degree of freedom in a human that has nothing whatsoever to do with the angle the neck makes with the level surface. The acid test for all doubters is that anyone on earth can make a straight-back and straight-thru stroke along the baseboard of a wall WITH ANY NECK ANGLE DESIRED. (If you haven't personally tried this, then don't argue about whether the point is valid or not -- you're not allowed a voice in the debate.) What happens is that the golfer simply moves the shoulder frame in a vertical plane of motion regardless of the angle of the neck. Do that (and only that -- no independent use of arms or hands) and the putter head runs straight back and straight thru; do anything else and it won't.

So teaching a "rule" that the "cervical spine" (neck) needs to parallel the surface is a flawed "robotics" misconceptualiztion of human biomechanics.

The robot shoulder-arms assembly COULD have degrees of freedom introduced at each "joint" if the robot is designed to "look like a human". But functionally the robot doesn't need any "joint" other than the pipe-and-hole-socket where the shoulder-arms assembly fits onto the pipe of the neck. But a human has joints at the shoulders, at the elbows, at the wrists, and at the fingers (not to mention at the hips, knees, ankles and toes). In a well-trained "shoulder stroke" the human does not change any of these joints during the stroke. In an armsy stroke, the human changes the shoulder joint and armpit, as the upper arms change relationship with the shoulder frame and chest during the stroke.

In general, a body part "moves" by changing the joint that is higher up or closer in to the center of the body (more "proximal") than the ("distal") moving body part. The wrist joint moves the hand. The elbow joint moves the forearm and the wrists and the hand. The shoulder joint moves the upper arm and elbow and forearm and wrist and hand. So what moves the shoulder frame (the body part, not the joint)? Answer: the gut muscles, where the upper skeletal frame is "jointed" to the lower skeletal frame by the spine sticking into the pelvis and sheets of muscle and tendon wrap from the rib cage down onto the pelvis in order to make the spine at the "joint" move around in certain ways. This "joint" moves the shoulder frame and the whole upper torso in relation to the lower body. The head and neck can be isolated and withheld from this motion due to the inner ear's connection to the brain, so you can move the shoulder frame without necessarily moving the head and neck too. The spine is willowy and actually bends sideways left-right or forwards-backwards like a young planted tree. That's a funny sort of joint compared to the hingeing and swiveling action of other joints, but it's what happens in the human body when you move the shoulder frame as a whole using the gut muscles.

IF a golfer makes a true "shoulder stroke", he moves the gut joint and this moves the shoulder frame and this moves everything downstream from the shoulders, all as a single unit without independent action at the other joints. The robot does the same because the shoulder-arms assembly CANNOT have independent joint changes other than the see-sawing at the neck.

It's possible to build a robot with all the degrees of freedom, just like a human (see MIT). But in golf no one tries this because they can't use it to show things and making it move is too complicated a job of coordinating vectors. A "putting robot" is moved simply by tipping the shoulder frame down and letting go. The shoulder frame then "rocks" like the pendulum on a grandfather clock and continues for a while as its motion slowly dies out.

So when people use a robot to "illustrate" how a human makes a stroke, they are kidding you and themselves. What they are showing is how the robot makes a stroke, which is defined solely by the fixed character of the parts and the neck angle being the only determining factor.

One of the dumber "rules" supposedly proved by a robot (Perfy, given a name to emphasize how "like" a human he is supposed to be in design and functional properties, which he ain't) is this:

"Unless the hands hang directly beneath the shoulder sockets, the stroke cannot run on a straight-back, straight-thru path."

Two follow-on "rules" are:

"If the hands hang outside the line directly beneath the shoulder sockets, the stroke will necessarily make a "frowny-face shaped" arc inside-square-inside" and

"If the hands hang inside the line directly beneath the shoulder sockets, the stroke will necessarily make a "smiley-face shaped" arc outside-square-outside."

Not one of these three so-called "rules" is true or applies to either robots or humans, which is why they are dumb. Using these rules to teach golfers how to putt is worse than telling a pole vaulting champion he can't clear the bar unless he's chewing Juicy Fruit gum at the time, because it affirmatively messes golfers up trying to comply with a nonsense rule at the expense of learning what really matters for movement and how to do it simply.

Only the pipe angle matters on the robot, not the position of the hands in relation to the shoulder sockets. That is because the shoulder-arms assembly is fixed in shape, and this means that WHATEVER PLANE OF MOTION results from the pipe angle, the shoulder-arms fxied-shape assmebly will transmit the same plane of motion all the way thru to the putter head, regardless of the relationship of the hands and shoulder sockets. So long as the relationship between the hands and shoulder sockets doesn't alter during the stroke, regardless of the relationship's exact form, the putter head will move in a plane that parallels the shoulder-arms assembly. If the pipe is level with the floor, the shoulder-arems assembly can move ONLY in a vertical plane perpendicular to the axis of the pipe, and so does the putterhead, no matter whether the hands hang beneath the shoulder sockets, inside the line, or outside the line. It doesn't matter one tiny bit.

For the human, first of all, the hands DON'T NATURALLY HANG DIRECTLY BENEATH THE SHOULDER SOCKETS. Just look at ANY HUMAN standing still with decent posture -- the hands always are at the end of forearms that angle out of the line directly down from the shoulder sockets, so the hands actually hang directly above the toes. This is due to the fact that muscle development in normal growing up and living creates a steady tension on either side of the elbows that slants the forearm out of this vertical line forward a bit -- usually about 10-20 degrees out of vertical. So getting the hands directly beneath the shoulder sockets at address in putting requires "putting and keeping" them there with tension in the upper arms and pecs to make a certain angle of the upper arm with the shoulder socket OTHER THAN the natural, tension free hanging of the upper arm in the same vertical line beneath the shoulder sockets. The ELBOWS naturally hang directly beneath the shoulder sockets, not the hands. Just look at any relaxed, standing adult human!

Second, the human can move the shoulder frame in a vertical plane regardless of where the hands are positioned in relation to the shoulder sockets. There is no bone connecting the shoulder bones to the spine. There is no bone connecting the shoulder bones to the neck. Check any anatomy book. The shoulder sockets themselves (as opposed to the shoulder frame as a unit) can usually be "hunched" up about 5 inches without bothering the neck at all. And the shoulder frame as a whole can be rocked sideways up and down without moving the head any at all or very, very little. (Incidentally, a really BIG putting stroke only requires one shoulder socket to move down from its starting position and then up above its starting position about 3-4 inches, as the radial arm of the limbs and the putter shaft translate and amplify this small central motion into a big sweep of the putter head back and thru, so you don't really need a lot of independent range of motion of the shoulder frame to putt over a very wide range of distances without involving other body parts to an undesireable degree.)

So these "Hand position determines stroke path" rules are especially dumb, since they don't apply to robots OR humans, and they don't correspond to humans at all anyway.

Robots only illustrate the relationship between the pipe pivot and the shoulder-arms fixed-shape assembly, and even this is not really understood or taught by the people using the robots to teach with. Instead, these good folks confuse matters by suggesting falsely that if you could make yourself into a robot's shape and then mimic the motion the robot makes (and repeat this "perfect practice" 20,000 times), all your problems will be cured. Too bad these folks aren't very clear thinkers or teachers!

There are other differences, such as the "torque" and "coupling" forces in the way the robot "grips" the putter handle and reacts to impact, but these are secondary to the ignorant illusion that the robot illustrates what the human should do. In fact, teachers using these robots never seem to make the point about moving the shoulders in certain planes of motion in order to make related shapes in ther stroke path of the putter head, so apparently they don't know about this. But it's just fifth-grade geometry.


Geoff Mangum
Putting Coach and Theorist
Golf's most advanced and comprehensive putting instruction.

Visit the new PuttingZone Blog for podcasts of putting tips:
Site PuttingZone Blog
RSS XML Subscription

Todd Dugan
Todd Dugan

August 23rd, 2016, 7:43 am #3

Jeff, reading this post, its clear you have a good understanding of putting bio-mechanics. So then, why the criticism of a robot like Iron Archie? The robot is showing you how you CAN swing, one possible way. Rotate your shoulders within a plane perpendicular to the club at address, with joints "fused", and you'll swing the putter within the "address plane" with face constantly square to that plane. What's so bad about that?

I have found that while you CAN rotate your shoulders within any plane angle you like, regardless of the neck angle, you can NOT keep the WHOLE spine fixed unless you rotate the shoulders perpendicular to a section of the thoracic spine, below the shoulder joints (the neck is way too high for the natural axis of shoulder rotation). You COULD use a "hunching" of the shoulder joints, as you mentioned. That's totally un-natural, and not a rotation (circular) of the shoulders at all.

Do you actually recommend swinging the putter within a vertical plane? By keeping the shoulders within a vertical plane? I find this incredibly un-natural...the most natural rotation of the shoulders being perpendicular to the thoracic spine, thus maintaining a wholly fixed spine, head included. Of course, you could incline the thoracic spine to horizontal, like Michelle Wie. Also very un-natural, and somewhat painful, after a while! And yes, I realize that the arms must swing from the shoulder joints to swing the club in-plane this way, assuming the thoracic spine is not inclined all the way to perpendicular to the spine (Seve did that). Tiger is the perhaps best example of this.

What say you?

Todd Dugan, PGA

Joined: January 1st, 1970, 12:00 am

March 23rd, 2017, 4:06 am #4

Dear Todd,

Your description of the tilted-plane stroke used by Iron Archie to trace the Putting Arc "curve" should be: a straight-back and straight-thru stroke on a tilted plane of 14 degrees off vertical, with no face angle change from square at address. This is the true motion that "traces the curve" of the Putting Arc. It's not on a tilt that matches the putter lie and shaft angle, which is almost always 19 degrees.

So what's wrong with that? Well, first, a 19 degree tilt is a worse "curve"than the Putting Arc. If made by running the heel of a putter along a plank tilted back towards the golfer by 19 degrees off vertical, the putter would rise up the plank and also come closer to the golfer. If the backstroke is 2 feet long, and the golfer's body pretty normal, with a distance from the base of the neck to the bottom of the putter about 54", the heel of the putter rises up the plank 5.63" and comes to the inside closer to the golfer by 1.94". The angle of the backstroke off vertical at address is 26.4 degrees. If the putter face loft is 3 degrees, this backstroke delofts the putter face to 23.4 degrees. Because this putter face is tilted out of vertical by 19 degrees, the AIM of the putter face at the top of the backstroke points out past the intersection of the bottom of the plane and the green, the same way a putter face tilted "heel up" by 19 degrees at the top of the backstroke will aim the face out past the intersection line.. This intersection is the AIM line at address and the direction the ball needs to go at impact. The only time this tilted plane actually AIMS the putter face along the AIM line at address is exactly at the back of the ball.

ANY error in the timing or form of the stroke, or in the setup, or in the ball position, results in a line direction error. This is not true when the stroke moves the putter head in a vertical plane. In that case. the putter face always AIMS only at the AIM line at all times, so ball position, stroke timing and form, and setup are not critical factors challenging sending the ball where aimed.

In addition, the tilted-plane stroke has the putter face moving outward coming to the back of the ball (hook side spin) and then inward on the other side of the exact bottom of the stroke (slice spin). None of this is true with a vertical-plane stroke.

The worse thing that happens with a vertical-plane stroke (at least vertical before impact in the forward stroke) is that the putter sole is a bit high or low compared to the height off the ground that was set at address at the back of the ball. If the ball is too far back, the impact is higher off the ground, slightly delighted, and moving down thru the top back quadrant of the ball. Even if the ball is 3-4 inches back from the bottoming -out point in the stroke, this is not especially bothersome to sending the ball where aimed. The ball MAY hop a bit but is impacted with a square face moving along the aim line, and the height of the putter face compared to the bottoming-out height is on the order of the thickness of a quarter -- which is insignificant, something under 0.01. If the stroke bottoms out before the ball position and meets the ball 3-4" forward of this, the putter face will have a little extra loft but will meet the back of the ball on the lower quadrant a little higher on the putter face than normal -- about the thickness of a quarter higher on the face than usual. neither of these impacts lose enough energy to challenge good distance control in the vast majority of situations, and the impact dynamics still sends the ball on the aim line.

In addition, once the arms are hanging naturally off the shoulders, swinging the arms back and thru takes advantage of the Newtonian physics and the top "swing bar"of the shoulder frame that organizes the two "chains" of the arms to PREFER in physics swinging straight beneath the shoulder frame in a vertical plane. In this action, the forward motion of the putter will MOVE the front side of the body only vertically, as there is no physics apart from some ill-advised movement of the golfer that sends the arms and putter anywhere other than straight along in a vertical rising directly above the aim line.

In contrast, the golfer has to "pick" a tilted then learn how to swing along that plane ONLY, both back and thru, with ball position, setup, and stroke being unnecessarily critical. Even so, the arms alone will not swing on a tilt, and the arms in a fixed shape will have to be moved by the shoulders "rocking" in the correct plane. After years of careful observation, it is my conclusion that the shoulder "rock" sucks at doing a single plane on a tilt. What actually happens is that the upper torso invariably rotates, with a perpendicular line out of the chest swinging to the rear in the backstroke (indicating torso rotation) and then swinging back and to the rear in the thru-stroke. If you laid an aiming stick across the chest from shoulder to shoulder and then made a typical shoulder rock, even one intended to move in a vertical plane, the aiming stick at the top of the backstroke will aim to the outside of the aim line and in the thru-stroke will aim to the inside of the aim line. Performing a perfectly straight-back and straight-thru shoulder stroke -- whether vertical or on some intended tilt plane -- is a very seldom thing hard to keep up with, hard to learn, hard to practice, hard to perform, hard to avoid error, and hard to detect and know the cause of error.

In contrast, when the arms alone are moved, the arms flap back and forth fairly straight according to nothing-special muscle action and also stay fairly straight due to the swing-bar plus two-chains structure of the body for swinging the arms. In this case, a perpendicular line out of the chest STAYS straight, without swinging to and fro as it does in a shoulder rock.

This is better appreciated by identifying the muscles that moves the shoulders (without any change in the arm pits) versus the muscles that move the arms apart from the chest (with the rear arm pit opening in the backstroke and then reciprocally the front arm pit opening in the thru stroke). The shoulder stroke is moved into the backstroke by the inner oblique muscle on the front side that tugs the entire rib cage towards the pelvis when it contracts. The obliques are made more for rotating the torso over the pelvis than a "side crunch" motion. The spine is not great with a side-crunch and is much better suited to only a slight lateral flexing and a rotational twisting. In contrast, the front pectoral muscles tugs the lead arm across the middle of the body for a backstroke. Once the arms are suspended by a torso bend at address that clears the rear of the upper arm out over the chest wall. it is pretty darned difficult to contract the front pectoral in a way that directs the arms to the rear, as the chest is in the way and the pectoral muscle can't contract in that direction. Consequently, the pectoral muscle sends the lead arm fairly straight back without any attention by the golfer to make it so. And when the arm pits opening the backstroke and again in the forward stroke, the CHEST is totally uninvolved in the stroke. This leaves the "swing bar" of the shoulders unaffected by the motion, and this of course helps the stroke of the arms goes straight down the aim line without any special care and attention by the golfer.

There are other, more minor reasons for not liking a tilted-plane stroke, but the above are the big ones.


Geoff Mangum
Last edited by aceputt on March 23rd, 2017, 4:12 am, edited 1 time in total.