Question about ToF VL53L8X resolution.

[sejik]

Hello, I'm currently working on a project with a VL53L8X sensor.
When I installed this product on the ceiling, the data of adults passing by is well observed, but the data of children passing by is not. I understand that this is a physical limitation that comes from the 8by8 array.
Is there a software solution to this?

Thank you

[John E KVAM · Accepted solution]

With the sensor mounted on the ceiling it only has to see 2 or 3 feet (60-90cm) to see an adult, but children are a lot shorter and smaller, both of which are leading to your issue. They move faster too. 

The trick is to turn up the integration time. Allow the sensor to gather more photons. If you are already running in continuous mode, we are going to have to try running at 12 frames per second instead of 15. Or even 10 frames. 

When facing someone straight on, people are pretty big. But when looking down on them they are kind of small. And hair is not very reflective. 

Allowing more time will gather more photons. 

Another trick is to infer someone is there by not seeing the floor. But if you have a carpet from most of 3 meters away, you might not see the floor - depending on the type of carpet. If you put a wool carpet under the sensor - you could see it easily, but the plastic blends are much harder. 

Avoid putting the sensor where sunlight can become a factor. The excess ambient light only makes your problem more difficult.

- john

[sejik]

Thank you for your answer, I am currently using it within the range of 5 to 65 ms interaction time for the 15Hz setting.

Let me ask you one more question.
If I attach a sensor to the ceiling and sense a 90cm child, the relevant data will only appear in one or two pixels in the 8 by 8 array. Is the only solution to this case the only way to set parameters related to the recognition and judgment of the algorithm?

- sejik

[sejik]

Thank you for your answer, I am currently using it within the range of 5 to 65 ms interaction time for the 15Hz setting.

Let me ask you one more question.
If I attach a sensor to the ceiling and sense a 90cm child, the relevant data will only appear in one or two pixels in the 8 by 8 array. Is the only solution to this case the only way to set parameters related to the recognition and judgment of the algorithm?

- sejik

[John E KVAM]

The 8x8 array covers a 45x45 degree FoV. So in rough terms a 5x5 degree square for each zone. 

And yes, depending on how far away the ceiling is you are only going to get one or two zones. 

The only real hardware solution to your issue is more resolution. And we are working on it, but it won't be out for another year. 

 So your algo is going to have to deal with an object in only one or two zones 

- john