Hi guys,
i’m looking for robust and stable solution to track a bowling ball.
The ideal solution is a tracking system that gives back coordinates of the ball.
I’ve excluded lidar solutions a cause of reflections of the bowling alley.
I believe that the obvious solution are hi speed cameras, but i never used it, so i’m asking suggestion for solution of type or other technologies able to achieve this job.
Ideas?
Cheers.
Have you looked at any of the Zed camera options?
Hi jes… I never worked with zed but watching the specs im not sure that has enough frames catched per seconds to achieve the result.
Bowling balls, expecially in pro’s challenges, are very fast (the world record is around 210 km/h…
) So i believe that the mid range of shots velocity are over 100km/h … And i believe that 60fps are not enough to have a good track but… Maybe i could wrong on this.
Anyway thanks for your reply i’ll try to inspect better zed cameras or i’ll try ro write to company to ask more information about.
Cheers.
Hi @lucesintetica ,
I don’t know which brand of LiDAR you’re used to work with, but reflection of the floor or walls shouldn’t be an issue, you could use a 2D LiDAR parallel to the floor, a few centimeters above (ideally at a distance equal to the ball radius). In my opinion it’s a perfect use case for this sensor.
I’m using LiDARs from SICK brand, and models in the LMS-1xx series should do it.
You can contact me if you need more infos on this.
Best,
i agree with @michelchrome, I’d say LiDAR 2D would perform great for this case. Also easy to calibrate per bowling lane.
We’ve used a 3D LiDAR in a narrow alley with one side of the alley being a giant mirror, without issues.
I’m in the 2D Plane LIDAR camp on this one (I’ve personally worked with SICK, Hokuyo, and Leuze), but I’m thinking about alternative solutions.
Given the (usually) stable lighting conditions, mostly predictable lane structure, and measurements, it feels like you could go as simple as a CV Background subtraction to determine the centroid of the bowling ball. I’d probably go with a 120fps frame camera and use homography to align the captured image back to a 2D coordinate space. Exclude certain areas with ROI restrictions. Difficult part is the pins at the end of the lane.
thanks marshall,
but the road of lidar is more “secure” because i don’t know what kind of environment/s i can find (the project have to be serializable in different conditions and places… maybe 10 location or more different).
computer vision was my first idea… but:
i worked only with leuze rod4 but for a touch led wall but i’m interested to know more about sick products…and how they are interfaced with touchdesigner (just because hokuyo and leuze are native in touchdesigner…
thanks for your reply!
You can certainly position 2D LiDAR at the exact height of the center of the bowling ball, but if I’m understanding correctly wouldn’t you be dealing with potential occlusions by the bowler, or any of the ball return apparatus, the pins themselves (if positioned at the end of the lane) or balls thrown in other lanes (if positioned at the side)?
The only LiDAR solution I’ve seen with temporal resolution above 60fps is the Sick LMS1000. The Zed-X might be a 120fps solution, but it’s not shipping yet and requires the Jetson AGX Origin platform as a host, using a GMSL2 connector.
Thanks for your reply i’ve only to track the ball to create some effects as trails or similar… so the track start when the ball is on the lane (maybe with a bit of offset to take only the ball and not the human… same for the pins… (i could stop the track before the pins with a good calibration).
I don’t need to take the score or similar…is only for visual purpose.
Eventual other data are managed by the main software of the bowling.