Hallo , ich habe im Forum Deine Kommentare zum LIDAR V3 gelesen. Ich habe da mal eine Frage: Wie möchten die Positionen eines Bowlingballs (1m breit) messen. Der Ball hat einen Durchmesser von 21 cm und wird mit einer Geschwindigkeit von ca. 18 bis 31 Km / h gespielt. Die Bahn ist 18 m lang… Die Position soll gemessen werden bei ca. 4m, bei 9 bis 12 m und bei 18 m…
Kann man das mit dem LIDAR v3 realisieren?
Sorry, irgendwas ist untergegangen - die Breite der Bowlingbahn 1 m…
I’m glad to try and offer assistance, but I do not speak German, unfortunately. Therefore, I’ll be using the automatic translation below as a point of reference:
Well, since you want to measure an object’s distance within 4-18 m, then the LIDAR-Lite v3 should be able to do that. The fact that the ball might be reflective to IR and is round might make it a bit harder for the sensor to detect its position (and therefore being a bit slower / require to change its settings to take more samples). Since your ball will be moving at 18-31 km/h (~5-8.6 m/s), if you can sample it at ~100 Hz you’ll get a resolution in position of at worst about 1/100 of that speed (~5-8.6 cm). Of course, if you can sample faster (depends on ball surface, distance, angle of beam from sensor, environment, etc.) you can get more resolution. If getting better resolution is very important to you, you may want to use the LLV3HP which is better at detecting objects and can measure at faster rates, too.
Your main issue will be the fact that the ball can be in a 1 m width lane while it is itself only 21 cm wide. The beam from the LLV3 (and LLV3HP) is not very wide and therefore cannot detect the object in that entire 1 m range. You would need to have more than one sensor or have the sensor move/rotate to detect a ball in the entire width of the lane.
Hallo, erst einmal vielen Dank für die Antwort. Das Problem mit der Bahnbreite habe ich auch erwartet, da der Winkle des Lidar relativ klein ist im Nahen Bereich… Die Idee, den LIDAR beweglich zu machen ist cool…Ich denke mal drüber nach wie das praktisch zu realisieren wäre und melde mich nochmal…
I figure if you make if movable along the width of the lane (i.e.: left-right from the player’s perspective) you could maybe track the ball width-wise position so the sensor stays on target:
This is of course more complex than simply measuring a distance, but not impossibly so.
Two ideas come to mind:
Have two or more physical sensors on the “Sensor” (from the image above). Kinda like a line following algorithm, the sensor will move on its track to follow the ball. By having more than one sensor, you can now if the ball is exiting the sensor’s beam from the left or the right and compensate appropriately.
Alternatively, you could get a camera on top of the sensor that will check for the ball and move the sensor along its track based on it. This could be done with a cheap camera / fast algorithm since the goal is only to figure out if the ball is to the left, dead center or to the right (and nothing really more detailed than that).
Your English is obviously superior to my German (non-existent). That being said, English is also not my first language… so I totally understand if you have some issues!
Bei Variante a.) fehlt mir der Bezug des Messpunktes (die roten Linien zeigen die Werte die ich brauche) in der Variante b.) habe ich einen definierten Bezugspunkt , weiß aber nicht ob der Laser auf einem Schrittmotor die Geschwindigkeit des Balles mitmachen kann…vor allen Dingen, wenn die Geschwindigkeit ehr unterschiedlich ist…
Considering the speed of the ball, option B would probably be much more difficult to implement since your tracking/rotating system would have to move really quickly to measure the ball’s distance.
It would probably be better (requirements wise) to go with option A where you are measuring in line with the ball’s motion.
That being said, do you only need to know when the ball crosses those three points (and what I would assume would be the time at which point the ball crosses them?) ?
If so, you could probably use something simpler, like simple IR or IR laser-based beams that detect when they are cut by an object. You tie three of them to a microcontroller with a good internal clock speed so you can measure the time when the ball passes each.
Please provide more information about what information you need of the ball and how it will be used so we can make better recommendations.
Example: The scanner is installed in the middle on the ball return design. The distance from the sensor to the outer edge of the lane is approx. 37 cm … The ball runs on the bar 5.
The area the scanner captures at an angle of 2.3 degrees is about 1.7 cm at the outer edge of the ball. The measured distance is then about 40.0 cm. (37 cm + 1 bar a 2,72 cm = 39,72 cm) Speed of the ball about 5 to 8 m / s. To get the desired value, you have to calculate the 40 cm - 37 cm + the radius de balles (10.5 cm) divided by the width of a bar (2.72 cm) = bar 5 … which should be displayed become.
Yes, this can be feasible as long as your sensors scan fast enough to catch the ball. If the ball’s top speed is 31 km/h > ~8.61 m/s and the ball is 21 cm in diameter, you’d want to scan (i.e.: read full measurements) fast enough to not miss it. Let’s assume our target is at least half the ball : 10.5 cm > 0.105 m. 8.61 m / 0.105 m = 82. Therefore, you’d want to take measurements faster than 82 Hz to ensure catching that ball. The sensor should have no issue taking measurements at rates around or above 100 Hz so this should be fine.
I’d also recommend either having a blocking surface or angling the beam slightly so it doesn’t escape far off. This will make sampling faster since the beam will always come back and therefore increase sampling rate when a ball is not present.
In reality, you’ll most likely get multiple samples from each sensor of the ball’s position as it’s beam hits various part of the ball so you’ll prob. want to make an average or perform some other statistical analysis of those results.
It sounds like you have your design figured out. Try it out and let us know how it goes!
Hier noch mal in Deutsch - die Übersetzung ist schrecklich…
Hallo, erst einmal vielen Dank für die Mühe - war super. Wir haben alle Deine Hinweise und Tips diskutiert im Meeting und haben die Probleme besprochen. Im Ergebnis haben wir erkannt, dass eine technisch machbare und flexibele Lösung (theoretisch) eigentlich nur ein 360 Grad scanner ist. Habt Ihr damit auch Erfahrung ?
You could certainly try and use a 360° LIDAR (rotational LIDAR). Just make sure it has a good enough angular resolution and high enough scan rate to make sure it does not miss your ball moving by at 8 m/s!
See the category linked above for the catalog of products we offer of that type.
Thanks, … so the device should not be too expensive to test. We chose the RPLIDAR A2M8 360 ° laser scanner or the YDLIDAR F4 Pro 360 ° laser scanner. Is that okay ? But one thing you have to explain to me: What does sampling rate mean: 6000 scans per second and scanning frequency: 5-12 Hz - which value is the more important?
It should be fine. They are both pretty fast and have angular resolutions that are reasonable (0.45°-1.35° for RB-Rpk-02 and 0.46°-0.50° for RB-Ydl-02).
Assuming you place your sensor to the side of a lane (lets say 30 cm from it), in the middle of the length (~9 m from the 0-18 m range of the ball) and the lane is is 1.2 m wide, then the furthest distance from the sensor the ball can be would be ~9.1241 m:
With the worse angular resolution reported for those sensors (1.35°), you may skip an area at that distance of at most ~0.2150 m or 21.50 cm. Since your balls are 21 cm in diameter, that may be cutting it a bit close but should still work if your sensor is placed so that it his the ball in the middle. If it hits too low or too high then the ball is less wide and it may very well miss it in some of the scans (or all!).
Both values are related to each other. The scans per second refer to how many individual distance measurements the device can perform per second. The scanning frequency refers to how many full scans (360° of coverage) can be done per second.
As you can imagine, the accumulated duration of each individual measurement makes up the duration of one full scan (360°) from which you can derive the scanning frequency.
Individual measurements can vary: the closer the object/more reflective the faster the measurement; alternatively if there is little or no signal returned then it can take longer. Therefore, you can help increase the scanning frequency by decreasing the duration of each individual measurement by having a better environment/setup for your scanner.
I think the best at this point would be to choose a part and try it out and see how it performs in real life (and real lighting conditions). So many things in the test (and use) environment can affect a LIDAR so it is best to try it under conditions that are as closed to reality as possible.
You may also want to check this article about LIDARs for some details about the various characteristics.
I hope this info can help you and your team further develop your idea/project.
First question: we only need a scan range of 0 - 180 degrees for the area of a train. Can you program this?
Second question: With a sampling frequency of 15 Hz, I get a reading every 53 cm - at 18 m, that means about 34 readings per litter.
Third question: At the max. Length of 9 m, the detection range of the laser beam is approx. 22 cm. I definitely get a reading, it’s the only object on the track.
I think it should work like this for a first test. The big problem we see is to hide the measurements of objects outside of the track (walls, players, etc.).
Most rotational LIDAR have a scan range of 270-360°. You can simply ignore data points in the range you are not interested in (discard them).
I’m not sure where you get your 53 cm from. Please note that 15 Hz scanning frequency means 15 scans of the full 360° range of the sensor per second. In each of those scans, hundreds of measurements are taken and they wouldn’t be spaced out that much (53 cm).
Yes, that is true. It should also be noted measurements from a partial beam return may not provide proper measurement (depends on many factors).
Yes, I think so too.
This is why I recommended earlier to either add obstacles to block the beam to a fixed distance (panels around the lane) or slightly angle the sensor so it aims towards the lane edge at the far end of its scanning range (if possible).
Good luck with this and let us know how the test goes!
