RVR, issue with color sensor

Hi,

My problem is the ability to trigger actions on the color cards supplied with the unit. By this I mean, going over a color that I programmed in the Scratch program and trigger an event like “Speak out the color”.

I laid all the card side by side and had the RVR go over them at speed 50. The only color that triggers the event 100% of the time is the red one, the others maybe 1 out of 10. I made a snapshot of the color sensor recording when it was passing over green, purple, yellow,green and finally red. I can see that they are all detected but only the red as a strong pulse.

There was also a large lag between the red color card and the event. I did a test at speed around 50 and the RVR was way pass the color card, but this may be another issue.

Here are some info:
This info was sent to the support group.
The sensor is clean and without scratch
The white LED underneath is lit when going over all the cards
The lighting in the room is good
All the events are programmed with their specific color using the cards.

I have pictures of the sensor, card setup and color detection pattern snapshot.

If someone had this problem or knows how to resolve it please contact me.

Thanks

Benoit

1 Like

Hello Benoit,

Thank you for posting here, and welcome to the community forum!
Can you tell me more about the setup that you are using?
What device was your RVR connected to?
Do you have a snapshot of your program I could take a look at?

Thank you!

Quentin

1 Like

Hi Quentin,

Device was running from my iPhone and there are 3 pictures attached. The firmware versions are 7.1.422, 7.1.435.
I re did the test and attached 3 more pictures, the pattern and the sensor data and video. On this test the green and blue were not detected. I put my program public " color test card".

1 Like

Thanks for the information.

I cannot see the pictures you added, would you be able to post them again?

Thank you!

Quentin

1 Like

Hi Quentin,

This is a link from a shared album on google, can you see the pictures and did u located my program ?

1 Like

I can now see all 6 images in the shared album. I still do not see the code for the program you are running! Could you add a screenshot of that?

Thanks!

1 Like

Here it is included with the other pictures, I tough you could easily find my program in the public.

Thanks

1 Like

Hi Benoit,

Thank for your sending more pictures.
I just looked at your program, and it looks good.
I have a question about it though.
For each “on Event” block, can you try to set those colors again by clicking on the colored squared? You might want to calibrate again the robot against those cards.

Let me know if it helps!

Quentin

PS: Je suis français :wink:. J’espère qu’on va pouvoir résoudre votre problème rapidement !

1 Like

Quentin,

A chaque test j’ai refait la calibration sans résultat.

J’ai été contacté aujourd’hui par le groupe de support qui est a amélioré la précision du capteur. Nous verrons peut-être cela dans la révision du firmware si attendu en Q1.

Merci

1 Like

I am also having a similar problem with my RVR. Even on re-calibrating the color multiple times, the sensor overlooks it. The situation’s worse in case of low lighting and a moderately high speed.
Eager to know a solution for this @quentin.
Thanks!

1 Like

Hi,

I do not know if you read my answer in french to @quentin. We do have the same issue and I was told by the support group that they are working, among’s others, on this issue. I also supplied them a table showing the standard RGB values vs my sensor values. Hope to see that fix in the next firmware update.

Bye

1 Like

Lol, and then there is me with grass covering all my sensors…

1 Like

Lucky, here it would be snow

1 Like

I meant grass inside RVR, not growing on the ground…

My class is having the same issues with my RVR. My students have been trying to code through a maze and getting very fustrated from each card only working about 1 out of 15 times. If someone knows of a way to help the sensor work better it would be greatly appreciated.

1 Like

Any solutions to this problem?
It seems that the “on color” event is only triggered by a specific RBG value.
That doesn’t work due to lighting conditions, variations in materials, and of course, the accuracy of the color sensor itself.
It would be more useful for “on color” to be triggered by a range of values.

Anyone have any workarounds?

1 Like

I’m having this same problem and agree the best solution would be a way to match RGB with a threshold percentage or tolerance.

2 Likes

For a start, please make sure the color sensor window is clean, with no dirt or fingerprints. Any contamination of the window surface will affect the sensor readings.

The firmware generates a confidence value based on the detected color’s proximity to the nearest color in the active detection palette, which is then compared to a threshold in the app, so no, the RGB values do not need to be an exact match.

For Python SDK users, the commands are all there for fully configuring color detection, but documentation and examples don’t cover the entire interface. That’s on our list for a more thorough example in the future.

Hopefully a good cleaning of the window resolves these issues.

Jim

1 Like

Jim I have this a while back with no success and the lens is not scratch.

From what I read, there is a lot of users with the same problem, color detected randomly. I tough this would have been found and would be corrected in the next software release.

Regards

Envoyé de mon cellulaire

1 Like

Same here. Inconsistent color detection frustrates students.

Maybe there’s a JavaScript workaround?

1 Like
SPHERO Email Marketing -