"Static object Removal" concept possible on latest VL5xx sensors?

[BWies.2]

ST,

Another non-IR sensor we have worked with has something called 'static object removal' in its microcode.

When this function is turned on for a given range, it internally histograms the environment so that objects within this range 'get deleted from consideration' after a few seconds- this allows intruders which subsequently enter within this zone to then be detected. Happily, human intruders (elbow tip, hand etc) are not stable enough or of consistent surface regularity/ reflectance to - even if slow & not moving - get 'deleted', so intrusion detection is not lost even though a fixed object left 'in view' will get 'deleted' after a few seconds.

Is there anything like this on the latest ST VL5xxx sensors or would this require substantive external processing?

Bill Wiese

San Jose CA

bill@bwiese.org

[John E KVAM]

I'm definately going to pass that on to our chip designers. We might just give that a shot. But adding features is not fast.

So in the short term consider setting a motion interrupt. (It's in the ULD additional features.)

What happens is the sensor interruts on changes to the scene. So if nothing changes, you don't get the interrupt.

And you get to decide what is significant.

Would that work?

• john

[BWies.2]

Hi John

We're trying to minimize processing/histogram - and just get 10x/sec stream of intruder data distances.

This is prob something you could do inside the loaded microcode.

What we are using now .. an object placed into 'zone' and stays there for a second or two gets subtracted outta the

histogram and there is

no distance reporting of intruder - until something else comes in, like a hand/wrist/elbow tip etc. The distance

reporting of intruder then begins. (Current devices having this feature don't reject hand because even 'still',

it's enough even minor motion for device not to reject - and the surface contours + lower reflectivty of hand

vs objects in range differ enough that intruding hand(/etc.) will not get subtracted out.

'Scene delta' might work - but we'd want distance to intruder reported when occuring - and not distance to the now-stationary

object introduced into the 'scene'. It's OK it if takes a second or two etc. for sensor internal processing to evaluate scene.

We'd just like the histogramming etc processing to be internal and subtract out hard fixed object.

Bill Wiese

San Jose CA

bill@bwiese.org