Hi, welcome to the forum. I am sorry for the long post. I have been meaning to post this information about PIR sensor. This is the best I can do based on what my journey finding an outdoor PIR sensor. I learnt a lot. I hope what I can share is useful for everyone here.
The way PIR sensor work is by sensing heat changes across the sensor. With the help of the lens in front of it, the changes, in a way, is amplified. The sensor basically generate compounding Sine Wave. Most PIR sensor will use band pass filter (passing sine wave from x to y Hz). This is the range that they think a human make.
So, what is the problem in regard to sun. Sun movement due to cloud/shadow may also look like a human movement in addition to the sun light reflected on the car moving. If the car move at highway speed, I bet the shadow will not trigger the detection.
I think the lux information is only useful to determine if there is a movement at low lux value (at night or Seattle cloudy day), then it is considered reliable. At high lux value, I do not think that you can reliably say that it is from a human. I believe this is the extend of the value of the detected lux. I do not think adjusting the sensitivity will result in better detection. The band pass filter will pass the signal regardless the amplitude of the signal.
There is also possibility that Hue will do DCT (discrete cosine transform). By doing this, they could do much more analysis (perhaps AI) of the wave in the frequency domain. However, I doubt it. I do not think that you would get months of usage on battery if you do something like this let alone years. I do not design or tear-down Hue motion sensor. However, I doubt any battery power PIR sensor would do something like this at today's technology.
There is anecdote that by manipulating the lens, you can get better accuracy. However, the physic behind it is hard to believe. Making the lens less sensitive basically narrow the band filter range that manufacture intended for. This may actually cause the sensor less sensitive. I may suffer false negative.
My personal experience solving this problem is using 2 sensors (microwave and PIR). I bought Bosch Tritech motion sensor (they use those 2 in one box). A move has to be detected by the 2 sensor before it is sent. They have Zigbee and wired version. They are really cheap. I use it on my ST hub. This is usable sensor for me. It is not great. You cannot use it as is out of the box. You have to adjust it for the range you want. It is not easy to get it right. Later on, I also find that it eat battery quite a bit. It take 6 AA battery. On busy area, you get less than 3 months of use. They do advertise 6 years of use if I am not mistaken. I believe that DCS make wired version of this type of sensor. I also play around to replicate this technique with my own homemade sensor quite successfully.
You can also use 2 plain PIR pointing to different shadow on the same area. The concept is the same as above. However, I find it very difficult to look at the same area and hope your get different shade of sun. In reality, this is much harder to do than the theory.
Thanks
Iman
