DIY simulator (camera based)

[SwedishGuy]

Just wanted to share what I have built in my cellar. https://youtu.be/msGIdTG7l8Q

[Bubba22]

When I try to open the link it says the video is private.

[SwedishGuy]

Hm, try again now

[Wigalo]

Looks pretty good Swedish...what are you actually able to measure vs calculate? If I recall, you were also helping Martin a while back... I'm guessing your stuff will be completely go ready before his ever will.

[aeroburner]

I am also interested on this can you actually do another video on this but show the measurements as these were cut off on the recording and would like to see these. But this is very cool are you willing to make an instruction video and material list of what you have done once you have gotten this complete for people to build.

[SwedishGuy]

As I say in the comments, I measure club speed, ball speed, path angle and launch angle with the shown setup (2 cameras used). Always getting straight club path and face angle with that setup since that's not at all measured. Ball spin rate is just guessed from the selected club, and spin angle is straight. So no draws or fades.
I have been thinking about making it open source, and have tried to find other people interested in such a task, but with no luck this far.
It's already complete for others to build, it's just a matter of spending time on making the instructions for people (and selecting platform to host it on). I have even made some 3D prints for housing the lasers and sensors but as you could see from the video there is just a plastic "food box" housing the Arduino :-).
The software design is intended to be modular, so apart from adding a club path detecting module, or ball spin (have code for marked balls), I'm also thinking about maybe replacing the laser trigger with something like Martins PX2 (unsure if it has external trigger).
Think I have working code for interfacing with TGC as well but haven't gotten around to buy that license yet.
The cameras used are Point Grey Chamelon 3, but cheaper Basler Dart cameras should probably work as well.
The laser trigger also works as a standalone club speed device since it has a display for showing club speed (and speed record). On that device is also a LED found which, when connected to full system, will be red when not ready to detect shots, and turns green (ready for shot) when the cameras have found a still ball placed in the teeing area, and the simulator program is ready as well.

[SwedishGuy]

Made another video showing some measurement in more detail, and also showing that it's at least giving some expected reasonable numbers. No real calibration/comparison with other launch monitors have been made. Once I tried to put a plastic sheet in front of the impact screen which I shot holes in, and then used simple Pythagoras to verify the numbers, and it looked pretty accurate then. But hard to do many times....

https://youtu.be/EmgOpmtJ8Ek

[SwedishGuy]

Decided to add a club path camera as well to the setup (an old PointGray Chamelon2 I had lying around). Works pretty well also without reflective dots, and now I get draws and fades as well.
Some results can be seen here https://youtu.be/61hCMGe-FkQ

[jhp016s]

This is pretty cool! Nice work.

[TorchRedRob]

This is awesome!! If you can get ball spin, you'll really have something here. Keep up the good work. I look forward to seeing what's next.

[Wigalo]

That's awesome Swedish... your club cam is already more accurate than Martin's and he's been at it over 7 years .

[SwedishGuy]

I actually have a ball spin module as well if I connect the LX I have with my code and specially marked balls. Some problems though, like the fact that the balls should be marked and the LX placed pretty close to the ball (less flexible where I place the ball and I like the fact that I can place it on different turfs now), but it would also add two more cameras and I might get into issues with too much stuff on the USB bus. Only still pictures though so it should be doable since I have a separate USB card in the PC.
I could do it with only one camera as well, but with less accuracy in spin rate (same accuracy for axis)
But I think I will first try to integrate another program I have laying around which is just using a web cam (Logitech C920) to record the swing. It both presents the live swing with overlaid helping lines, and does recordings. Only thing I need to change is how I trigger the recording, to use the laser trigger signal instead (and of course to have a buffer to include some seconds from before the trigger which happens at ball impact). The camera only gives some 30 fps but quite good quality for the money, and stepping frame by frame still gives some nice feedback on the swing. Switch to a color Point Grey camera with some 150 fps instead for 2-3 times the money and it should be quite good. And 2 of them to get swings from both behind and face on. But then I start to count to 7 cameras all together... Four should be enough but would require some really fast cameras (1000-2000 fps) and onboard processing. I can just hope the progress in self driving cars will leave some interesting and affordable integrated cameras as a spin off

[awisnia]

really, really cool stuff!

[SwedishGuy]

Had some fun yesterday and today and added the Logitech C920 webcam to the solution as well. This means I get a 4 seconds (configurable) video of every swing which I can do slow motion in and draw lines in (both in the live feedback and the recorded). Can also save out to disk if wanted. Works nicely, might put up a video of it on youtube as well eventually.
This actually means I want to have an extra monitor placed on the floor right in front of me since the live feedback is very valuable for training.

[Snaphook]

That swingrecording is very interesting. I would like to have something like this. I guess this software is something you have made yourself?
Would love to see a video.