Watch an amazing Arduino project evolve over the timespan of 4 years. The builds in this video use Nema17 stepper motors, micro servos, Phototransistors, Arduinos, Raspberry Pis, and electret mics. Music: "Mysterons" by David O'Brien
@@scwfan08 Ya, swiftkey might able to counter this, as there is no tap XD but for the customize size, I dont think it is an issue, as all the data can be collected and then analyze the position of the tap and roughly guess the position of the key.
@David It's actually based on time/phase differences. With a high enough sampling rate, it's easy to extract time or phase differences of 4 microphones. The human brain also mostly uses IAD (Inter-Aural Delay) to detect direction rather than amplitude differences. The amplitude can be the same and direction can still be found so long as group delay or phase differences are detectable.
it’s a good exercise to practice solving problems and learn more when you are a developer like for clowns they enlist to make people laugh through stupid comments
Great job with this project! I would suggest for next iteration to polish algorithm, so instead of reactive correction is to make one more intentional. For example, no mater where you drop the ball, it will always go center and stay there. Unless it's impossible due to ball flaws and environment (air movement etc).
Nice job. I work for a company that makes sensors and vision systems for industrial applications. Typically camera applications are mostly about the lighting and can be a little frustrating as the daily light changes in a room. Again nice job.
Thanks. Yeah, the camera approach was really sensitive to it's environment. At some point in time, I tried to use Infrared LEDs in combination with a camera which is able to pick up IR light in addition to the normal light spectrum. Well, guess what; Sunlight has A LOT of IR in it. Compared to what the sunlight was blasting through my windows (even on cloudy days) the light-intensity of my IR LEDs was next to nothing. That approach didn't work out well. The position data when using cameras is good though. And knowing at any point in time where the ball is, is a strong advantage compared to the mic approach where we only know the position at the time of impact. I am thinking about trying another camera approach.
The amount of effort, tenacity and ingenuity in this whole process baffle me. Bravo, what you’ve done here is amazing, and don’t ever listen to the naysayers. F*ck those haters.
I wonder if you can make it stronger, then make multiple tables juggle the ball. Then put each table on a wheeled base. To make an autonomous volleyball, only on a small scale.
This would make a great random number generator! las Vegas might have an interest in this as a random number generator.Every position of the ball generates a numeric value on the board.
I love the last visual, and can see that it must have taken quite some effort to automate that visual (if you didn't automate it and inspected it by hand then kudos!)
Great work! The ball dances over the plate. Is it because you use the microphones to actually "hear" where the ball is hitting the plate, and then try to adjust to the ball with a trajectory calculation? So the plate adjustment always lags a little bit behind? Cameras are getting better and better, but the contrast problem is always present in bright environments. What about ultrasonic transmitters? They could be used to "see" the ball before it hits the plate, and together with the microphones you could use a Kalman filter to combine the sensor data to get a much faster adjustment, I think.
Yes, there is indeed a 1-bounce delay in the control loop. And I think this is the main reason for the circular pattern the ball tends to be drawing while bouncing on the plate. I have thought about using ultrasonic transmitters in combination with ultrasonic mics to somehow sense the balls position. This is an interesting approach, but in my opinion (and I might be wrong), cameras are the way to go. I am current working on a ball bouncing machine with a single camera to extract the balls position (120 FPS image processing.) It's the newest video on this channel.
@theOnionPrate A lot of trial and error. But there was some method to it as well. First I set the P value in such a way that it was able to get the ball to the other side confidently. P alone is, of course, no good. The ball starts to oscillate and fall off the plate after a short amount of time. At this point, we aren't taking into account the ball's velocity. That's what D is for. So with D we try to stop the oscillations and get the ball to stop moving. D is proportional to the velocity while P is proportional to the distance measured from the center of the plate. And at last, there's I. With I we try to adjust for any tendencies the ball is showing. If it looks like the ball is mainly in the upper left-hand side of the plate, let's try to slowly tilt the plate in such a way that this tendency is canceled out.
This can be used as a way of reflex for Robots, same as we humans are aware of someone join in us from behind when we wait for something. In short term '' Aware Reflex''.or Motor Reflex. And you can use that to generate power.
@@letsgocamping88 True, though that could be kept to a minimum by using simulation based self learning, which would create a different problem; The fact that it would be extremely hard to set up properly.
Love it! So it looks like you've got the microphones incrementing minute adjustments to the stepper motors to maintain ball position? Do you have any sections of the code posted anywhere? I'd love to see how you wrote the controls for this. . . Simply top notch. :) I can't imagine writing anything like this, I'm impressed.
WOW!!! Amazing bro congratulations 🥳 I really felt that when I started with a cardboard hand and now I have a complete iron arm... It worked but I dissembled because I wanted tu fixed a mistake but I’m still don’t fixed it :(... Really amazed of seeing your great work!🤩
it can still be improved. You should add a granular movement action where you adjust the pitch of table according to distance of the ball from optimal position. Your ball is bouncing around center of the table. If it's fully done, ball should stay in the center and keep there forever.
Dude Nice!!! Just a suggestion.... instead of simple camera, think about infrared. Just a disk piece as film for the camera and some IR leds and you are done