[Dprglist] Light Sensors, Moth/Anti-Moth Behaviours, and PID steering

[Murray Altheim]

Hi,

Among my other side projects is a moth/anti-moth behaviour. Rather than a
passive sensor I thought to use one that provides a better very-low to
bright ambient light range. From Pimoroni I ordered a pair of Adafruit
GA1A12S202 Log-scale Analog Light Sensors, which are cheap and tiny 3-pin
boards that produce an analog output. They arrived yesterday and I built
a small holder for them on the front of the robot, wired them up to a pair
of analog inputs, then began writing a Python support class.

But when I thought to look into more details about the sensor I found out
that Adafruit have discontinued these sensors because they don't actually
work so well as described. They intended to replace them with a Vishay
VEML7700 lux sensor wired up as an I2C device, but sadly the designer never
thought to provide a way to alter the fixed I2C address, so one can only
mount a single sensor on a robot without resorting to multiplexing, which
I'd rather avoid. A real oversight. Nice little board otherwise.

Investigating the VEML7700 it turns out that its support library has some
serious bugs at higher light levels, such as what one would encounter
outdoors (!), and that part of the problem may be the sensor itself.

Vishay make an improved sensor called the VEML6070, and Adafruit even make
a breakout board for it (since the sensor is SMT), but, but, but, they
again didn't provide for altering the I2C address.

So for now I'll be sticking with the GA1A12S202 since I have two of them
installed and my robot (for now) is indoors. But since the board has been
discontinued and apparently isn't good for outdoor use I'm looking for
alternatives.

Adafruit list 26 different sensor boards (!), but it's unclear which have
good performance and suitable output across a wide light range, and of the
26 many have been discontinued. I'm not sure if a UV sensor would work as
well for a moth/anti-moth sensor; I'm guessing not so well in a dark room.
There's a TSL2561 that permits altering the I2C address but has also been
discontinued. From what I can see, almost all of the I2C light sensor
boards (e.g., TSL2561) that haven't been discontinued have fixed addresses.

The one exception to this is the BH1750,

   https://www.adafruit.com/product/4681

which has a solder bridge to change the address. I may order a pair of these,
as their lux range looks suitable. It's a bit larger than I'd like (at 25x18mm
it's about 3x bigger than the pair of sensors I'm using now) but seems to be
the only one Adafruit sell that fits the bill. Is it any better than the
little GA1A12S202 boards I have?

Anyone have any experience with this or could recommend a good light sensor?
And preferably small and relatively cheap? Most of these boards are well
under US$10, e.g., $2.50, $4.50.

----

This is all in mind of developing that moth/anti-moth behaviour, where the
sensor is supported by a Python class that has various possible outputs: an
Orientation enum indicating where the brightest light is coming from (an
output of PORT, STBD, or NONE, with a hysteresis setting to permit a center
point); a bias value from -MAX_V to +MAX_V; a pair of raw floating point
values from 0.0 to 3.3 (volts); and a pair of ints from 0 to MAX (which is
calculated as round(v) * 100, in this case 330).

I've got this up and running, supported off two analog input pins on a
Pimoroni IO Expander board (the little Nuvoton MS51 on the front of my
robot), so far so good:

    https://github.com/ifurusato/ros/blob/master/moth_test.py
    https://github.com/ifurusato/ros/blob/master/lib/moth.py

----

With our ongoing discussion of PID loops and the use of a PID controller
for steering a robot, using velocity+rotation rather than left and right
velocity as the steering model. The idea I had was that if one were using
a simple proportional PID for steering, to implement a moth or anti-moth
behaviour it'd be a matter of adding the output from get_bias() to the
rotation value, multiplied by some kind of fixed constant to control how
forcefully the robot reacts to changes in light level. I thought about
having its reaction slow as it got closer and closer to bright light,
until at a certain threshold it would stop (in moth mode). A bit like my
cat. This could be used to charge the batteries from a solar panel.

Is this something that anyone in the group has done before? If so I'd be
interested in hearing of your experience and any advice.

Cheers,

Murray

...........................................................................
Murray Altheim <murray18 at altheim dot com>                       = =  ===
http://www.altheim.com/murray/                                     ===  ===
                                                                    = =  ===
     In the evening
     The rice leaves in the garden
     Rustle in the autumn wind
     That blows through my reed hut.
            -- Minamoto no Tsunenobu