Advancing Table Tennis Robots: PongFox Tests Service & Return Feature - Thoughts?

[NextLevel]

Ok how about this case, the pad can be used when doing drills to track your consistency, it can show you how often you hit a target at a particular speed and frequence, or while doing drills at known places and randomized drills. Now as a player wouldn't it be helpful to realise where your consistency start declining and will help you to start working on that with something to measure with?

Not to the level of player you are currently pitching this to. That's what we are evaluating here, the limitations of the setup in simulating realistic play. If you want to measure seeve target practice, there are many ways of doing that, and yours is reasonable even if a bit large. Maybe a smaller target pad as well to increase the difficulty?

Being able reduce the time for the ball to come back or increase and ary the spin on the return in response to the incoming ball is also critical as the player gets better. Like I said, this might work maybe for hardbat. For table tennis it has major limitations for higher level practice and I don't mean world class either, I just mean decently strong tournament amateur like USATT 1400.

 

[UpSideDownCarl]

Since this is really the same thread as Palguay's, does anyone mind if I merge the two threads? Then, if the people Tony is asking come to the thread and comment, they are going to the original source.

 

[cytivrat]

just slow down.
i'm busy learning this feature (lol)

@Tinykin
@pingpongpaddy
@VictorMoraga
@cytivrat

Hi gents, that has been tagged - please see OP above.

Carl, I notice a couple of these members hasn't been around of late... continuing our "hijack thread".... one of the sad parts is loosing quality, contributing active posting members.

Well, I'm not a coach, so it is hard for me to say how it influence learning process, but since you tagged me:
1. I've never see such thing before, it is something new for me. From this points of view, it is a good thing that robot can be taught a new trick, it is one step ahead from a long way to make robot close to human coach and it is better to make that step rather than not, it is progress.
2. Can it be useful? Yes, it should be good for beginners, for those who have problems getting in ready position after serving (from my point of view, mostly for adult learners). But in this case it should be done under coach supervision to get properly serve type/robot configuration (and then it might be better just to use coach) or for some simple serves (like short underspin, getting underspin return. But the better players is, the more likely to get harm not getting real human return.
3. Am I interested to use it on my own? Maybe, but most likely no. There are too many conditions I should fulfil to make it close to real life: like place landing spot and robot close to each other, to get realistic angle, program first robot shot depening on serve I'm going to practice, maybe robot distance to net should depend on services type, length, etc. to get more realistic trajectory, timing and other parameters of return ball. So I would be limited in serving one serve, then spend some time to adjust everything to use another serve, serve it for some time, then again. Probably, if I had it at home then maybe I would try to use it (definitely, it is better to have this practice, than not to have), on the other hand maybe (it is even better to have a ball on a stick placed somewhere on a table and do serve, hit that ball, serve, hit that ball and so on). But since I don't have table at home, rather limited in time I can spend on tt and have to go somewhere to play, it is definitely easier and much more effective to pay a coach to return my serves if I wanted to practice it, or practice it a little bit before/after training session on my own and then go to a tournament and improve my skills there, or to play some kind of drills with a partner to get into 3rd ball attack situations

As a conclusion and for those who don't want to read all those words above: might be good for beginners (maybe for adult learners, maybe for recreational players), but not good for people being properly taught.

 

[Tony's Table Tennis]

Since this is really the same thread as Palguay's, does anyone mind if I merge the two threads? Then, if the people Tony is asking come to the thread and comment, they are going to the original source.

well, my goal was to really just speak to the 13 coaches in the video thread, since there was history of "why robot is not good for learning"
wasn't meant to be a full on dialogue, haha

any ways, I think the owner of the robot has gotten his feedback needed, so yeah, I don't mind the merge.

 

[Tony's Table Tennis]

Not to the level of player you are currently pitching this to. That's what we are evaluating here, the limitations of the setup in simulating realistic play. If you want to measure seeve target practice, there are many ways of doing that, and yours is reasonable even if a bit large. Maybe a smaller target pad as well to increase the difficulty?

Being able reduce the time for the ball to come back or increase and ary the spin on the return in response to the incoming ball is also critical as the player gets better. Like I said, this might work maybe for hardbat. For table tennis it has major limitations for higher level practice and I don't mean world class either, I just mean decently strong tournament amateur like USATT 1400.

check this pad out (posted on the original thread by another user)

imo, it is very difficult for robot without sensors (actually seeing the ball) to be any form of realistic.

The AI study one of the Taiwanese academic team is doing is to read the body action and to train the AI to read spin and what not.

That is the Li brothers I posted a while back.
These are proper smart people, who will likely spend years to come up with something, that is still very not applicable for the masses. maybe 10 years later, something good can come up for lower level players.

The Taiwanese project is to use AI to tell you how you can play better (ie fixing your strokes)

 

[Tony's Table Tennis]

Well, I'm not a coach, so it is hard for me to say how it influence learning process, but since you tagged me:

oh, I see your name added to the list of people who can comment on the video thread

 

[Der_Echte]

Depends on the playing level. I could argue that a basic robot has value for training all levels of movement. But this tool could cause serious problems for a player higher rated than even US 1500 IMHO. I would think even US 1000, but it depends I guess.

I will not argue against that.

A robot can ruin even a 2000 player if not used effectively... but can help in some ways too. They are not sold for decades in these quantities if they are totally worthless.

 

[Tony's Table Tennis]

I will not argue against that.

A robot can ruin even a 2000 player if not used effectively... but can help in some ways too. They are not sold for decades in these quantities if they are totally worthless.

I agree

we have 1 robot in the school (or I think 3, haha)
only 2 people i know of used it.... over the past maybe 5 years,
the head coach (when no one is around and he wants to exercise) and a top level coach, who is a chopper (national women team level and had wins over World top 50 player).
the chopper coach set it on full blast top spin to train before national trials. She had been coaching for 2 years, and very rusty and use the robot to just get some feeling back.
obviously, once she started getting things going, she got male pros to top spin for her.
She is maybe the highest level that I know that used robot, but it was a special case scenario and she is a chopper, so top spin feed is very welcomed.

 

[brokenball]

Thanks for the link. We did think of the accelerometer but in terms of getting something out in a short time without making significant changes to hardware this was a simple for us to deliver, hopefully that will be something to look into over a longer time once we can gather enough data about the usefulness of the tool

I was not suggesting exactly that device. There may be better or cheaper devices available.
Also, I think an accelerometer on the net to adjust the throw timing is more important than the pad.
Do you use a microcontroller in your robot?

As for the robot being bad, how can a robot be bad? like everything it depends on how you use it. They are good for exercise. I have a Newgy 2050. It is good for practicing returning back spins because it can provide very strong back spin. My third coach was rated about 2500 but he couldn't chop. I sometimes move my 2050 back from the table. The robot is too limiting being attached only to the end and center of the table.
A Newgy 2050 does have a problem where there is only one throw wheel so the spin is proportional to the speed. It can't shoot slow spinny loops or short back spins. A more advanced robot should have two throw wheels so the speed and direction can be changed to throw slow spinny loops or short back spins.

 

[UpSideDownCarl]

well, my goal was to really just speak to the 13 coaches in the video thread, since there was history of "why robot is not good for learning"
wasn't meant to be a full on dialogue, haha

any ways, I think the owner of the robot has gotten his feedback needed, so yeah, I don't mind the merge.

oh, I see your name added to the list of people who can comment on the video thread

Many of those people are just good players whose commentary is worthwhile rather than them being coaches.

But the important issue is, this version of discussion didn't really belong in a thread that is intended to make it so people can post video footage and be comfortable that the people commenting on their technique are not going to be the kind of people who don't know what they are talking about and will say things that are harmful or insulting.

This subject just does not really fit into that thread's purpose.

 

[pingpongpaddy]

just slow down.
i'm busy learning this feature (lol)

@Tinykin
@pingpongpaddy
@VictorMoraga
@cytivrat

Hi gents, that has been tagged - please see OP above.

Carl, I notice a couple of these members hasn't been around of late... continuing our "hijack thread".... one of the sad parts is loosing quality, contributing active posting members.

HI guys
I am still lurking on here and mytt
I do have an amicus prime robot at home but most of my playing, training, coaching is done 80 km away where I play. At age 74 I don't have the same motivation to work with it as I would have even 10 years ago.
It's nice to see some attempts to add value to the robot training process but I think its quite challenging to add auditory tools as typical training halls are quite noisy environments. Maybe the sensory catch net could be exciting...
When I trialed my amicus at a club it created some interest but it tended to require too much input from the expert user when people wanted to try different scenarios. On the other hand it was great for helping say advanced cadets to work say to work on difficult shots like perhaps looping a heavily chopped ball from 3 positions. You can have 6 cadets queuing up to take their shots and they handle everything themselves including picking up the balls to maintain continuity. Meanwhile coaches can be doing something else.
Still I think in a big busy hall that little girl doing the feeding so well reminds us that human multi ball feeders are varied, fun and and don't break down easily.
Using my robot at home I have found that just serving practice is the thing I do most. The other thing I found useful is to set up a quite a long set of different challenging placements and just practice reacting and moving to the right position by watching the robot head. Its not the same as reacting to opponents bat but it does have some benefit.
I have to be honest though a good human multi ball trainer would be my choice to work with rather than a robot if I wanted to develop my game
I would use human trainer if I wanted to do a really high quality session The robot just performs at a basic level

 

[NextLevel]

I will not argue against that.

A robot can ruin even a 2000 player if not used effectively... but can help in some ways too. They are not sold for decades in these quantities if they are totally worthless.

That's not my point though. I have seen world class players use robots. My point is that given what is being trained, is the robot a net positive or negative?

Handling spin you know and moving to the ball, the robot is usually a net positive. We know those contexts. We know how to get robots to simulate them reasonably well

The particulars of serve and third ball as laid out in this design? Not so much.

A world class player can use a robot to train footwork or even hit basic strokes. I have seen this nuff times.

Can a world class player use this for serve and attack practice? That's my point.

 

[Tony's Table Tennis]

That's not my point though. I have seen world class players use robots. My point is that given what is being trained, is the robot a net positive or negative?

Handling spin you know and moving to the ball, the robot is usually a net positive. We know those contexts. We know how to get robots to simulate them reasonably well

The particulars of serve and third ball as laid out in this design? Not so much.

A world class player can use a robot to train footwork or even hit basic strokes. I have seen this nuff times.

Can a world class player use this for serve and attack practice? That's my point.

I feel amateurs tend to misunderstand pros a lot.
I mean, if you see a pro use a robot, it doesn't mean, the pro got there because of the robot.
Pros may use it to stay fit or work on certain drills to "maintain" themselves.
They are not using it to improve or develop new skills or techniques.

Serves and 3rd ball. If they use this.... the robot reaction time is way too slow on a long fast serve.
The pro could possible serve 2 long balls, and the robot "return" still hasn't arrived yet and I'm not joking about 2 serves for 1 return. I truly believe there is enough time and very possible.

The AI training in Taiwan academia project, is using AI to read you and correct you.
So until a robot can teach you a thing or too or read you and process the reading data for good purposes. It is just a unsmart feeder that is programed and the user need to match the program and not the other way around.

Good human feeders can match the human player. That to me is the real smartness required for growth.
So, robot/ai in table tennis is still years behind the true meaning of "smart", especially for any player say about usatt 1300 or 1500 and up. For beginners, it can be useful. But still requires a coach imo.