OOAK Table Tennis Forum
https://ooakforum.com/

Randomising robots
https://ooakforum.com/viewtopic.php?f=67&t=37999		 Page 1 of 1
Author:  GaryAblett [ 16 Apr 2021, 10:41 ]
Post subject:  Randomising robots
So how do some of the randomising robots work ?
Do they have 2 or more heads ?
Are they truly randomising ? For instance, can I get a first ball fed to be right topspin to extreme deep right on my side of table , second ball fed to be extreme left backspin to close to net on extreme left of table, third ball to be dead ball to the center of table , fourth ball to extreme deep right with medium left sidespin topspin etc etc ?

Please don't laugh at me because I have never owned a randomising robot. Most variation I have seen is feeding same spins to left or right.
Author:  haggisv [ 16 Apr 2021, 20:33 ]
Post subject:  Re: Randomising robots
Some robots and randomise the position and speed ball, but I don't think any will randomise spin, because they're no way you can read the spin, so there is no point.
Author:  Hubeer [ 17 Apr 2021, 02:38 ]
Post subject:  Re: Randomising robots
haggisv wrote:
Some robots and randomise the position and speed ball, but I don't think any will randomise spin, because they're no way you can read the spin, so there is no point.


Why can't you read spin ? You read spin in a match don't you ?

If you can train yourself to read spin (as much as possible) without looking at your opponent's racket movement and rubber color, it is far more beneficial. If you condition yourself to be proactive on you own looking at only the ball and not just "react" to your opponent's racket movement & rubber color (in fact most players even at higher levels don't even know or can watch the opponent's rubber color during a rally, as the sport was too fast even in 1983) that would be a better approach as that somewhat eliminates some variables that may negatively impact your ability to perform on your own. yes it is easier said than done but that is why I want a training robot to help me as much as possible.

I also strongly disagree with your statement "there is no way you can read the spin".
Reading sidespin is very easy. The ball will break to either left or right. (I am not saying learning to handle the sidespin & amount of it, is easy . I am only saying reading the left or right ball movement and therefore concluding it is sidespin is easy)
Reading whether topspin & backspin & no spin is "generally' not that hard because topspin usually shoots out or down(due to the spin pulling the ball additively in direction of ball travel) , backspin usually shoots up (or seems to stop, due to backspin pulling the ball back against ball travel direction) and no spin usually somewhat wobbles (somewhat like a perfect basketball free throw having "sort of" zero energy & no forward velocity, when it reaches the basket).
Yes estimating the amount of spin is the hardest but if one can train themselves as much as possible, what is the harm ? (And yes I am also aware of the fact that either top-spin or backspin is not just that and can be combined with left or right side-spin and that complicates reading of the spin quickly)
Yes non-randomizing robots are useful for many drills but there is no reason why spin cannot be randomized if speed & position can be randomized. Not sure how much of a challenge that may be but it may be possible. Solar cells had horrible efficiency in the beginning & who would have imagined cell phones in 1970's . In the evolving current focus on artificial intelligence (AI) I am not sure this thought is so farfetched.

Finally I do not know for sure but there may exist some robots that do randomize spin as well .
Author:  TroubleMaker [ 17 Apr 2021, 05:45 ]
Post subject:  Re: Randomising robots
With the Butterfly Amicus Expert it is possible to set 7 completely different balls and randomise them. The strokes can vary in position, height, speed, length and spin. But I find it very difficult to read the spin when the ball comes out of the robot. That is why I have designed most of the exercises in such a way that only the strength of a certain spin varies. However, it is possible to combine slow balls with backspin and fast balls with topspin. In this way, I can recognise the different spin by the speed to play different strokes in one exercise.
Author:  haggisv [ 17 Apr 2021, 09:54 ]
Post subject:  Re: Randomising robots
Hubeer wrote:
Why can't you read spin ? You read spin in a match don't you ?

By 'reading' the spin, I meant looking at the opponent's bat movement just before contact, to read the direction and amount of spin on the ball, which is something that does no exist with a robot.
Yes you're right, you can judge spin by looking at the ball and bounce, but it's a lot harder especially for fast balls, and it's more of a last resort when you were unable to read the spin from the bat movement.
For that reason I don't think random spin is a particularly useful feature... while random speed and location is very useful.
Author:  Hubeer [ 18 Apr 2021, 01:29 ]
Post subject:  Re: Randomising robots
haggisv wrote:
Hubeer wrote:
Why can't you read spin ? You read spin in a match don't you ?

By 'reading' the spin, I meant looking at the opponent's bat movement just before contact, to read the direction and amount of spin on the ball, which is something that does no exist with a robot.
Yes you're right, you can judge spin by looking at the ball and bounce, but it's a lot harder especially for fast balls, and it's more of a last resort when you were unable to read the spin from the bat movement.
For that reason I don't think random spin is a particularly useful feature... while random speed and location is very useful.


Sorry. IMO the most dominant characteristic of table-tennis as distinct from other racket sports is spin (of course ITTF had been trying to massacre that trying to take the spin away with somany rule changes as well as by going to TTX etc but that is another story)

So personally if I was making a robot I would focus on spin variations at least equally as much as speed & placement. Inability by the user to read spin is not a problem but a huge business opportunity to be addressed & solved

That said, even if spin was the least dominant element such as say for a pickleball robot, I would try to get an edge on the market by providing the ability to provide more spin. Whether or not how much it is actually used by the customer is not the issue.

Yes it is harder for faster balls to read spin but the user can start slow but move up to professional level. That is a the fantasy you want to market upon and nothing unethical about it.
Author:  palguay [ 17 May 2021, 16:16 ]
Post subject:  Re: Randomising robots
In the Pongfox table tennis robot (I am the developer of the robot), we have an option of random location as well as random speed. In this option for a stroke say counter the ball is placed randomly and the speed chosen is also random. This helps in movement as well as adjusting your stroke based on the speed of the ball.

We are debating if we should let the user add different strokes (backspin and counter) this might be real challenging but might have benefits like having user watch the ball when it bounces to figure out spin

We have a semi-random option e.g.: first ball service to random location, second backspin to random location and then 3 counters to random location, this is something I enjoy playing as this is pretty realistic
Author:  Hubeer [ 18 May 2021, 00:45 ]
Post subject:  Re: Randomising robots
palguay wrote:
In the Pongfox table tennis robot (I am the developer of the robot), we have an option of random location as well as random speed. In this option for a stroke say counter the ball is placed randomly and the speed chosen is also random. This helps in movement as well as adjusting your stroke based on the speed of the ball.

We are debating if we should let the user add different strokes (backspin and counter) this might be real challenging but might have benefits like having user watch the ball when it bounces to figure out spin

We have a semi-random option e.g.: first ball service to random location, second backspin to random location and then 3 counters to random location, this is something I enjoy playing as this is pretty realistic


As I stated earlier in this thread, the most dominant and defining element of table-tennis as compared to other racket sports is spin not speed (I am not saying speed is not important but is secondary).

But as you said, randomizing spin (with speed) may be a difficult design and may at least two if not more feed heads but hey if you want market dominance you have to raise to the challenge.

Watching the ball to figure out various spins (of course easier said than done) was also addressed by me earlier in this thread. You adjust your return not just based on speed (as you state) only but far more importantly on incoming spin as well.
Author:  iskandar taib [ 18 May 2021, 15:48 ]
Post subject:  Re: Randomising robots
haggisv wrote:
Hubeer wrote:
Why can't you read spin ? You read spin in a match don't you ?

By 'reading' the spin, I meant looking at the opponent's bat movement just before contact, to read the direction and amount of spin on the ball, which is something that does no exist with a robot.
Yes you're right, you can judge spin by looking at the ball and bounce, but it's a lot harder especially for fast balls, and it's more of a last resort when you were unable to read the spin from the bat movement.
For that reason I don't think random spin is a particularly useful feature... while random speed and location is very useful.


Random spin isn't useful, but the ability to program different spins one after the other IS useful and is a feature of several robots out there. I suppose once you've programmed a sequence of spins you'd have to memorize the sequence. You'd be able to program a "rally" beforehand - for instance, the first ball comes back as short backspin to the forehand (this would be the "serve return"), second ball is a block (topspin) to your backhand. And then maybe the third would be a long chop to the backhand, which you would attempt to step around and loop.

By the way, "Hubeer" is just another S-jan sockpuppet account. He knows all this stuff you explained already, he's just trolling.

Iskandar
Author:  Hubeer [ 19 May 2021, 20:08 ]
Post subject:  Re: Randomising robots
iskandar taib wrote:

By the way, "Hubeer" is just another S-jan sockpuppet account. He knows all this stuff you explained already, he's just trolling.

Iskandar


Thank You haggisV for appointing Iskandar Taib as moderator, so he can hunt down posters with legitimate content in a thread and call them trolls.

Thank you for letting Iskandar Taib, a moderator to follow and harass a poster he does not like.

BTW this is exactly how Iskandar Taib spent most of his days and nights in RSTT as well as a self-appointed moderator of an unmoderated group savior of the table-tennis world & internet (forums) by hunting down posters and harassing them with personal attacks, completely ignoring the issues discussed in the thread. ( Iskandar Taib also has a particular obsession with a certain object and he thinks people may have forgotten about it and I am leaving it out of here out of decency but if Iskandar, a moderator wants to take every thread into gutter attacking posters, I would be happy to oblige as well)
Author:  haggisv [ 20 May 2021, 10:17 ]
Post subject:  Re: Randomising robots
Stop it, both iskandar taib and Hubeer!

iskandar taib is not a moderator (don't know where you got that idea from?). Myself and other moderators don't read everything, and just because a post is left, does not mean myself or the other moderators condone it. I suggest you both stop picking on each other, or you'll find your posts deleted and account banned.

Back to robot discussions
Author:  palguay [ 02 Jun 2022, 17:01 ]
Post subject:  Re: Randomising robots
IT has been a while , We have now added this feature to the PongFox Table tennis robot.

You can now select multiple strokes and it comes out randomly, I have played with it and based on the strokes you set you can make out from the trajectory and this seems pretty helpful for my training as I am forced to observe the ball to know the spin.

Here is a post of this Random spin in table tennis robot
Author:  maurice101 [ 04 Jun 2022, 12:12 ]
Post subject:  Re: Randomising robots
TenniRobo can randomize one or more of the ball placement, spin level and interval between shots. I believe they are developing a new model that will give you a visual indication of the direction of spin etc. The whole head moves so you know the direction of the ball travel. Do butterfly robots still use a defector place that makes this hard to see?
Author:  dingyibvs [ 20 Dec 2022, 16:09 ]
Post subject:  Re: Randomising robots
I have an Amicus Prime, while it can't randomize spin it can play programmed shots in random order. I've tried having backspin and topspin balls in random order and trust me, it's useless. The issue is that you need different backswings for different spins, if you don't then you're never gonna be a good player. With the random backspin/topspin, you can't tell the spin until it bounces on the table, and that's too late for you to start your backswing.

Also, yes, the Amicus Prime still uses a deflector. Not being able to tell the direction of the ball is a MUCH more useful property. Against high level players you often don't know the direction of the shot until just before your opponent's racket contacts the ball, and this is how it's like with the Amicus Prime. I do a random FH/BH practice with fairly high speed and spin at 80 balls/min to finish each session. It's helped my FH/BH transitions tremendously.
Page 1 of 1 All times are UTC + 9:30 hours
Powered by phpBB® Forum Software © phpBB Group
http://www.phpbb.com/