Robotic Basketball Hoop V2

A few weeks ago, [Shane Wighton] created a basketball backboard which made it impossible to miss a shot even remotely close to the hoop. As a passive device, though, the backboard had its limits. Shots with tremendous velocity wouldn’t go in, and (like most backboards) it was missing facial recognition software. So he got to work on a second version which solves those issues, and takes a more active role in the game.

This version is flat, and looks largely unassuming until a game begins. The flat backboard is mechanized and includes a camera, so incoming shots can be analyzed in real-time while the backboard is moved into a position to direct the ball into the net. Or, since it does include facial recognition, the backboard can always send the ball away from the hoop, ensuring that [Shane] always wins basketball games no matter how many shots his opponent takes.

If you didn’t get a chance to see the original, we featured that a while back, and it’s truly a wonder when you learn about how much analysis went into creating the shape. The new version is even more impressive, doing all of that math in real time, and we can’t wait to see what [Shane] comes up with next.

12 thoughts on “Robotic Basketball Hoop V2

  1. Holy… This is awesome! What’s he do for a living? How come he’s not a multi-biilionnaire with that kind of talents? (Well, may be he is, but I don’t remember the Forbes magazine mentioning that # such and such on the list is some dude making YT videos.)

    BTW, he didn’t explain how he handled multiple shooters. May be it’s “just” facial recognition and it’s simple to do, but I’m an idiot and I’d love to hear more about the background technical info.

    Anyhow, this is definitely a subscription-worthy channel for me.

    1. An idle thought, how can this be used in a full-sized hoop? I remember reading stories about back in the old days when, even pro teams like the NBA and its precursors, played in gyms where the hoops support structures were attached to the walls and how home fans tried to shake the opposing hoop to make them miss shots. Of course, the mechanism can’t be as obvious as this, but may be if, instead of moving the whole backboard, it’d just “warp” it just slightly so that shots that normally would’ve missed would be good (for the home team) and vice versa (for the visiting team). It can even move the rim ever so slightly at the last moment to make dunks harder to make. (c:

      1. Now that is a good warped imagination – as long as it doesn’t sound like a wobble board. I’d think simply making the outer face in thin sheet and pumping a little air pressure behind it – so it domes around the middle would help make shots much harder and be very hard to see. Reversing the pressure – or perhaps pressuring the outer edges as separate cells should make it deform the other way for more self centering shots.
        Bimetallic might work well too though I think somebody is bound to notice the heat, either way both methods are pretty thin so wouldn’t scream odd.

        Never going to be more that a small % improvement but it really could be done.. And its more subtle than sandpaper in the trousers (Damn Aussie cricketers)

        1. “And its more subtle than sandpaper in the trousers (Damn Aussie cricketers)”

          I watched that doco on Prime! That’s a good one (i.e., the series, not the sandpaper). I remember reading about the scandal when it happened and not really quite understanding what it’s all about. After watching the doco I couldn’t help but wonder, “What the $&#% were they thinking?!? It was an international test match against SA, not a picnic with your buddies!”

          Anyway, just to be clear, I’m not advocating a cheater backboard and hoop in real games. May be in a backyard to play horse or 1o1 against “that guy”. (c;

    2. There are a few options for facial recognition you can just use. OpenCV is the only one I’ve played with at all for computer vision so far (and is really complex, capable, and good for just about any vision task, with lots of complex vision processing options already made), but there are many options out there some of which are by all accounts pretty much load and go for facial recognition and supposedly pretty good.

      How often it detects the right face and how fast it does it could be an issue – but as you tend to hold the ball for a while before shooting I would think there is enough time (and even if it didn’t get it right this time by defaulting to keep the ball out you should still win with ease even if it gets it wrong often – false positives tend to be much lower than false negatives when looking for one very specific face. I do wonder how well it will do at correctly knowing who shot the ball if the players are near each other and happen to look the right direction..

      To handle multiple shooters is then fairly simple when using one of the more premade options – have that software/library set the allow shot in flag only if you are holding it before shot. While flag is set it tries to make you score when not it plays dirty tricks instead.

  2. I love this project, and it’s really amazing how well made it is. Honestly, most of the issues he described from a software perspective could be solved by running all of this on an operating system with significantly lower overhead. If he can get the linear algebra to port over, this seems like a great project for a Jetson Nano or something that can really crunch through those matrix multiplies. From a cursory search it looks like it shouldn’t be too hard to interface the Kinect, so not much would even have to change.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.