TOF sensor distinguishing between object and person

TH3 · ‎2024-06-20

Hi all,

I'm working on a product that uses a TOF sensor (exact type# TBD) to identify if someone(s face) who is sitting in a chair is present within a certain distance and angle (around 20 degrees). If the right conditions are met the product should be switched on. I'm looking for ways to prevent the product from switching on based on false positives that the sensor receives from static objects when no on is actually present, such as wall, chairs etc. Appreciate your thoughts!

John E KVAM · ‎2024-06-20

The sensor is accurate to about +/- 3%, but if you pointed it a wall you would get a consistent answer to within a few mm. After a few of these, you can safely assume you are not looking at a person. Although a person sleeping in a chair might be a bit ambiguous.

I might suggest the VL53L4CX - it has a more-narrow Field of view.

Of the VL53L4CD - a shorter distance version on the L4CX. This device can see people to 1.3 Meters. (The max range on the VL53L0 is 2M, but it's effective range when looking at people is about 1.5M) The VL53L4 is our least expensive and easiest to use.

So, it would depend on your requirement for people being off center.

But no matter which you chose, I'm sure they will work for you.

Let's prove it.

For $56 dollars you can buy the P-Nucleo-53L0A1 evaluation board. Download the code and it will graph the data on your PC. Put the sensor where you think it should be and try it.

If you want to try the L4's, buy the P-Nucleo-53L4A2 (it's the eval kit for the VL53L4CX) Download the Graphical User Interface) and try that one. Then download the code for the VL53L4CD and run that on your VL53L4CX (it runs). It will show the performance of the VL453L4CD.

Then you can be confident your system will work, and you've only invested a hundred dollars.

Our community relies on fruitful exchanges and good quality content. You can thank and reward helpful and positive contributions by marking them as 'Accept as Solution'. When marking a solution, make sure it answers your original question or issue that you raised.

ST Employees that act as moderators have the right to accept the solution, judging by their expertise. This helps other community members identify useful discussions and refrain from raising the same question. If you notice any false behavior or abuse of the action, do not hesitate to 'Report Inappropriate Content'

View solution in original post

Anne BIGOT · ‎2024-06-20

Hello,

Did you have a look to our page : Smart presence detection solution - STMicroelectronics ?
You can find many interesting things about our advanced motion detection algorithm to accurately detect the presence and position of people.

Our community relies on fruitful exchanges and good quality content. You can thank and reward helpful and positive contributions by marking them as 'Accept as Solution'. When marking a solution, make sure it answers your original question or issue that you raised.

ST Employees that act as moderators have the right to accept the solution, judging by their expertise. This helps other community members identify useful discussions and refrain from raising the same question. If you notice any false behavior or abuse of the action, do not hesitate to 'Report Inappropriate Content'

TH3 · ‎2024-06-20

Hello Anne, thanks for your swift reply! The products suggested for smart presence detection rely always on multizone detection, am I right? Since the product we want to make is rather low cost I was hoping to find a solution with a TOF that only does range detection such as the VL53L0X.

Perhaps a suitable way of checking if its not someone but "something" is to check if the returned distance value is very constant for some time and then assume its something like a chair / wall rather than a person (assuming they will always move around a bit). Of course this would require the sensor to be accurate enough for this job. Perhaps you can still comment on this?

John E KVAM · ‎2024-06-20

The sensor is accurate to about +/- 3%, but if you pointed it a wall you would get a consistent answer to within a few mm. After a few of these, you can safely assume you are not looking at a person. Although a person sleeping in a chair might be a bit ambiguous.

I might suggest the VL53L4CX - it has a more-narrow Field of view.

Of the VL53L4CD - a shorter distance version on the L4CX. This device can see people to 1.3 Meters. (The max range on the VL53L0 is 2M, but it's effective range when looking at people is about 1.5M) The VL53L4 is our least expensive and easiest to use.

So, it would depend on your requirement for people being off center.

But no matter which you chose, I'm sure they will work for you.

Let's prove it.

For $56 dollars you can buy the P-Nucleo-53L0A1 evaluation board. Download the code and it will graph the data on your PC. Put the sensor where you think it should be and try it.

If you want to try the L4's, buy the P-Nucleo-53L4A2 (it's the eval kit for the VL53L4CX) Download the Graphical User Interface) and try that one. Then download the code for the VL53L4CD and run that on your VL53L4CX (it runs). It will show the performance of the VL453L4CD.

Then you can be confident your system will work, and you've only invested a hundred dollars.

Our community relies on fruitful exchanges and good quality content. You can thank and reward helpful and positive contributions by marking them as 'Accept as Solution'. When marking a solution, make sure it answers your original question or issue that you raised.

ST Employees that act as moderators have the right to accept the solution, judging by their expertise. This helps other community members identify useful discussions and refrain from raising the same question. If you notice any false behavior or abuse of the action, do not hesitate to 'Report Inappropriate Content'

TH3 · ‎2024-06-20

Thank you John, very helpful! I'll proceed like this 🙂