Vending Machine for Birds

There are a few things I want to improve on the VMFB, but don't really have the time to knock out right now. The current design works well enough and has worked reliably for a few months with minimal supervision. 

I'm trying to get to Stage 5 before May. Then I will be more motovated to do the following:

1. port the software to a lower-power solution 
2. make the dispense pathway impossible to jam
3. move the PIR sensor to the camera arm
4. make deposit sensor placement easier
5. make the deposit pathway shorter 

While this project is pretty tailored to geting birds to bring stuff to exchange for food, it can easily be used for other purposes.

• Replace unshelled peanuts with other foods and use this to train or condition other animals like rats, dogs or cats.
• Just use it as a bird feeder, dispensing food at intervals during the day to reduce mess, waste and unwanted visitors.
• Replace the deposit sensor with a button, puzzle, microphone or other sensor to study other forms of reasoning and intelligence.

When I work from home I often eat some fruit and put some scraps on the balcony for the birds. Usually they take them away within a few minutes.

However, the other day I noticed a jackdaw trading a piece of apple for a peanut. For a couple of years they have just eaten the apple scraps, and they know they are food. Interesting that one chose to trade one kind of food for another.

This could be the start of the birds realizing they can deposit pretty much anything to get a peanut. I hope this does not lead them away from depositing litter.

Full video of the apple exchange here. Apple deposit at 0:23.

I was away for a couple of weeks and while the birds ran out of provided deposit material during the first week, they didn't manage to bring anything of thier own after that. Visits declined as well, so I put some caps and butts out for them to resume their activity. I put some on the platform as well to try and entice them back.

So I'm still on Stage 4, will think about how to get to Stage 5 and try again.

10 deposits today. 2nd most deposits in one day since mid October!

The dispenser timed out for the first two deposits, but the jackdaws didn't give up and on the third deposit the dispenser resumed dispensing reliably. Still an active failure mode, but it happens rarely (around once in 100 deposits), recovers by itself and does not seem to discourage the birds - they keep trying until it works.

The adjustments I made to the deposit sensor in December seem to have eliminated the occasional missed deposit. The machine has been running for more than a month on its own. All I have done is refill the peanut hopper once.

They should run out of caps/butts by the end of the week. I'll leave it until the end of the month to see if they manage to bring something on their own.

The jackdaws just ran out of deposit material on the balcony a couple of days ago. There were some bits on the floor that I did not see when checking before. Now they are all gone, and while the jackdaws still come by they don't appear to have connected the dots yet.

I'll provide more caps/butts this weekend to keep them interested and try a 2-week test run with no provided material later this month.

I think the caps and butts I provided on the balcony are nearly gone. You can see the birds spending a little more time looking around the balcony before they locate something they can deposit.

I'm not going to provide any more caps/butts until 01-JAN and hope that one of the birds brings something I didn't provide to deposit by then. If not, I'll stay on Stage 4 for another month or so and try again in FEB.

The birds have been making deposits every day for more than a month, except for 26-NOV when I did some balcony cleaning and spent a lot of time near the machine. I've been leaving the birds caps and butts around the balcony to use, and I think there are 3-4 jackdaws that fully understand what they are for. I'm going to let them run out and not provide them anything for a week or so and see what they do.

The weather has turned cold, windy and rainy - standard Dutch November. This means fewer people out and makes food a little harder to find. I expect this to motivate the birds a bit and hopefully drive them to find their own litter. If they do, that will be a big breakthrough for me - Stage 5. If I can get wild birds to do this I will have a much better shot at finding a municipality/university/etc. that will consider letting me place one of these in a park or square for a year or so.

I do need to get to work on the ESP32 port of the software and look into non-wifi IoT comms options in the meantime - it will be a much more attractive solution if I can power it reliably via solar and a ~20K mAh 5V battery and don't need a nearby wifi network.

After running the vending machine unattended for almost a month, it has proven to be pretty reliable but still has some failure modes that are not completely solved.

Funnel Jam - solution in testing

The worst one is when the funnel beneath the dispenser gets jammed. There is no way to fix this remotely, so it needs to be avoided. After 100+ dispenses it jammed yesterday afternoon, leaving the jackdaws a little confused.

I had been thinking of completely redesigning this to not use a funnel, just drop the peanut onto an angled platform to dispense. This would add some complexity and build difficulty, so yesterday I took a shot at improving the funnel design first. The path is smoother now and I think jams will be even less likely to happen. I built a new funnel assembly last night and swapped out the old one today. Hopefully no more funnel jams.

Missed Deposit - solution known, to do

This is the second worst, and should be pretty easy to fix. I need a better way to mount the deposit sensor (using tape and velcro now) to give it a better view of the deposit chute. Right now it appears to miss about 3% of deposits and I think this is the cause. I will have a look this weekend.

This is not a persistent failure mode, and it happens rarely. I do think it can be easily fixed though.

Ghost Dispense - solution ideas, need to refine and test

Occasionally (about 2% of the time in the last month) a piece of shell or debris from the peanut hopper will register as a dispensed peanut. The best way to avoid this is to remove debris from the peanuts before loading the hopper and cleaning it a bit when refilling. I do have an idea to leave one or two small openings away from the sensors and modify the vibrating platform a bit so debris is dispensed but not seen by the dispense sensor. 

This happens rarely, but with other rewards like dog food pellets it would be a worse problem as the vibrations could shake off crumbs and small pieces easier.

Hopefully before the end of the year I will have eliminated or reduced the frequency of these failure modes.

I've been uploading deposit videos to YouTube since 16-OCT.  23 days running so far. I'm rarely at home and it's been unattended during that time except for a peanut top-up and providing some caps and butts on 27-OCT, and providing more caps and butts on 07-NOV.

I'm still on Stage 4 - providing deposit material near the vending machine. Stage 5 is the goal - where a bird brings something I didn't provide.

In the last 23 days I've had 104 deposits from birds.

• only jackdaws making deposits so far
• between 4-5 deposits per day
• no jams or dispense timeouts so far
• 4 deposits were not registered by the deposit sensor - a failure rate of about 4% which I plan to address

To make the channel easier to deal with I think I will start concatenating deposit videos into daily videos or maybe weekly.

The birds set a new record today, at least since 16-OCT - 11 deposits in one day: 

Sometimes the machine has a lot of lurkers waiting for a smarter bitd to make a peanut come out. Here's a jackdaw that had a plan for that.