Hydrogels Can Learn To Play Pong (arstechnica.com) 11
An anonymous reader quotes a report from Ars Technica: Pong will always hold a special place in the history of gaming as one of the earliest arcade video games. Introduced in 1972, it was a table tennis game featuring very simple graphics and gameplay. In fact, it's simple enough that even non-living materials known as hydrogels can "learn" to play the game by "remembering" previous patterns of electrical stimulation, according to a new paper published in the journal Cell Reports Physical Science. "Our research shows that even very simple materials can exhibit complex, adaptive behaviors typically associated with living systems or sophisticated AI," said co-author Yoshikatsu Hayashi, a biomedical engineer at the University of Reading in the UK. "This opens up exciting possibilities for developing new types of 'smart' materials that can learn and adapt to their environment." [...]
The experimental setup was fairly simple. The researchers hooked up electroactive hydrogels to a simulated virtual environment of a Pong game using a custom-built electrode array. The games would start with the ball traveling in a random direction. The hydrogels tracked the ball's position via electrical stimulation and tracked the paddle's position by measuring the distribution of ions in the hydrogels. As the games progressed, the researchers measured how often the hydrogel managed to hit the ball with the paddle. They found that, over time, the hydrogels' accuracy improved, hitting the ball more frequently for longer rallies. They reached their maximum potential for accuracy in about 20 minutes, compared to 10 minutes for the DishBrain. The authors attribute this to the ion movement essentially mapping out a "memory" of all motion over time, exhibiting what appears to be emergent memory functions within the material itself. Perhaps the next step will be to "teach" the hydrogels how to align the paddles in such a way that the rallies go on indefinitely.
The experimental setup was fairly simple. The researchers hooked up electroactive hydrogels to a simulated virtual environment of a Pong game using a custom-built electrode array. The games would start with the ball traveling in a random direction. The hydrogels tracked the ball's position via electrical stimulation and tracked the paddle's position by measuring the distribution of ions in the hydrogels. As the games progressed, the researchers measured how often the hydrogel managed to hit the ball with the paddle. They found that, over time, the hydrogels' accuracy improved, hitting the ball more frequently for longer rallies. They reached their maximum potential for accuracy in about 20 minutes, compared to 10 minutes for the DishBrain. The authors attribute this to the ion movement essentially mapping out a "memory" of all motion over time, exhibiting what appears to be emergent memory functions within the material itself. Perhaps the next step will be to "teach" the hydrogels how to align the paddles in such a way that the rallies go on indefinitely.
They will be coding next (Score:1)
Re: (Score:2)
That's either sales or upper management.
Re: (Score:2)
Slime mold can design rail networks [wired.com]. The California High-Speed Rail Authority is presently on suicide watch.
The articles/papers are hard to follow (for me) (Score:2)
Can someone give an explanation that a computer-inclined person would understand?
How are ball and paddle position represented digitally, and how is that representation used to train the gel? In DishBrain, apparently there are electrodes that provide electrical stimulation, and with this hydrogel setup, there is some mechanical stimulation. But how do ball and paddle position translate to the input to the motor that stimulates the gel?
And how is an output measured (I believe the outputs are binary "move pa
Re: (Score:2)
I posted below, with title "Yawn (commiserating with LJ)".
I hope that helps.
Re: (Score:2)
Yawn (commiserating with LJ) (Score:5, Informative)
I read the article. Then I reread it, then reread portions, some several times, to understand the details. The authors understood what they were doing, but their skills at writing clearly and communicating effectively need help. For larryjoe who wrote the post "The articles/papers are hard to follow", I feel your pain. As you asked in your post, "any kind soul able and willing to explain?", here is the gist. Follow along with figure 3B in the paper at : https://www.cell.com/cell-repo... [cell.com] .
The "Pong Environment" (PE) is the videogame software. In real Pong, the player positions the paddle, let's say on the left side of the screen, then clicks. The PE then calculates the ball trajectory, etc., and the cycle repeats with the ball coming back in on the right side of the screen. In this experiment, the gel replaces the person. The PE calculates where the ball starts on the right screen edge, then "displays" that by stimulating the gel via the electrodes. The electrodes and gel have just 3 positions, quite a low resolution. The gel is polyacrylamide that has been doped with NaCl, and the ions will migrate based on the electrical fields applied, same principle as used in bio labs for gel electrophoresis. The sensing electrodes then read the ion concentrations among the three positions (the technicalities of how they do so is a bit more elaborate, but that is the gist). Based on where the ions have migrated, that is taken as where the player's pong paddle is to be positioned in response. Calculating paddle position from ion concentration measurements is itself a post-ADC digital process, but that routine is a pre-process representing the player playing, and that calculated information is then transferred to the PE to complete the loop. Effectively, the PE says "CALL PlayerResponse (paddle_position)" which acquires analog data, computes the position, then feeds that into the core Pong executive. That is how I understood this after multiple reads.
BUT, it's a bunch of brouhaha over nothing.
It is easy to make materials that can record data then use that data to adjust a response or output. In this case, it is not as if the gel is itself making any decisions or is directly driving the response without intermediaries. It is just recording an event, in this case an electric field. It begs the philosophical and engineering question if there is a distinction between memory and recording. I can use iron filings to record a magnetic field. In the extreme, that concept becomes magnetic media for audio, video, and digital hard drive recording. I can use sand to record sound vibrations or harmonic motions (pendulum and Lissajous). I can use dyes and pigments to record imagery, as on a photograph or an LCD display. Etc., etc., etc. Each of those technologies affords far better resolution and information density than what they did here.
What they did was just an analog recording of the inputs, but fuzzy, very fuzzy, due to a strong RC (resistance-capacitance) factor inherent in ions physically migrating through the gel, overcoming electronic repulsion, etc. Using a sensor which looks like about 5x5 cm, they were able to resolve three inputs well enough to improve predictive responses about 20% over random (from the article, accuracy after "training" went from 50% random to 60%, if I understood what they were saying). I could tape an op amp to a spent battery and toss it at a roulette wheel and get better outcomes.
And that is really all that the gel is doing. It is like a sloppy op amp or capacitor, serving an integrator function in their circuit to accumulate charge after a stimulus. Sure, it is memory, just like computer memories made with capacitors to hold the info, but it is leaky, messy, unstable, slow. And, it is far slower and less resolved than real biological neural tissues.
Okay, it must have been a fun little project for them, but I don't see how it means anything. Must have had some spare grant money to spend.
Before Pong... (1958) (Score:3)
Tennis for Two was first introduced on October 18, 1958. It showed the ball bouncing up and down as it went between the two sides of the screen, rather than the left-right action one sees with Pong.
https://www.bnl.gov/about/history/firstvideo.php
The bioneural gelpacs on Voyager could play pong. (Score:2)
Breast implants that can play Pong... (Score:2)