Over the last 6 months, this blog has been sadly neglected. Some bits of work here and there were completed, but none of these bits reached any sort of satisfying conclusion, so writing about them hardly seemed worthwhile. Occasionally, someone does ask what I am working on these days though, so in this blog I will attempt a brief summary.
As the previous blog entry, written at the end of May alluded to, I attempted to create a new type of robotic joint that consisted of electromagnets constructed from soft iron rods and magnet wire, arranged orthogonally to each other, and driven with electric current to move a lightweight, plastic arm with a small permanent magnet fixed to its end. The idea was to have a two degrees of freedom joint, but without any relatively heavy conventional electronic motors to make it work. I had found some theoretical papers on this idea, but no practical implementations. The reason for this soon became apparent: When an electromagnet was activated with current, it would attract the permanent magnet at the end of the plastic rod, but only if the permanent magnet was fairly close to the end of the electromagnet. And once an attractive force was established, the permanent magnet would quickly swing over to the end of the electromagnet, and remain fixed there, even after the current was turned off. At this location, the permanent magnet would have been too far from any other electromagnet to allow it to produce enough attractive force to remove it from its current location. Possibly some different selection of larger diameter soft iron rods for the electromagnet(s) would have worked a bit better, but this would seem to defeat the original goal of making something more lightweight than a conventional electric motor. So the project was cast aside.
Once summer started, the fine weather inspired me to make some sort of robot that could work outdoors in the sunshine. Robotic boats are fun to work on and test, but a bit time consuming to drive to and from water. Why not create a robot that could function in the backyard, and do something useful like pick up dog poop? There is already a robot that purportedly does this: https://www.housebeautiful.com/lifestyle/kids-pets/a31289426/robot-picks-up-dog-poop/ although so far as I can tell, it is not commercially available yet. It seemed like it would be fun to develop something that uses visual AI to identify dog poop, scoop it up, and place it in a disposal bin of some sort. I figured it would make sense to get some sort of mechanical platform up and running as quickly as possible, so that I could then enjoy the fun part of developing visual AI algorithms.
Of course, the mundane assembly work for creating a platform occupied the better part of the summer. I used a piece of plywood for the base, a scavenged wireless transmitter from an old AMOS prototype, an Arduino Uno microcontroller and stepper motor driver from an abandoned compass calibration apparatus, a couple of garage door motors and DC motor driver from a previous attempt at making a paddle robot boat, and a worm gear and stepper motor that was scavenged from a broken 3D printer. The wheels were 3D-printed. Here is the ugly result:
Testing it out in the basement wasn't too bad. It would drive around, and the scooper, although perhaps a bit on the small side for our Labrador retriever, could be driven up and down. On the grass however, the robot failed miserably. The large back wheels tended to spin uselessly in our long, wet grass. Possibly if the robot had more weight, and better wheels (lawnmower wheels?) it could have worked better, but by this point the summer was nearly over, and the initial appeal of the project had faded.
I decided to switch projects again, although for some reason, stuck with a pet-related theme. We have an Apple AirTag for our orange tabby cat, Tiggy. It is very lightweight, and its small coin cell battery is able to power it for about a year before it needs changed. So long as Tiggy is within about 50 m of my phone, or 50 m of someone with an Apple-related product that has location tracking enabled, I can find his location on my phone. The problem however, is that when Tiggy ventures outdoors, he is seldom within range of anyone's Apple product.
I was wondering if it would be possible to have a similarly sized device that has GPS capability, and the ability to transmit the GPS location to a base station, at distances up to 500 m (Tiggy's estimated maximum range). There is a somewhat bulky GPS device available on the market already, that has a cellular transceiver. You need to purchase a monthly plan for it though, and the battery only lasts a few days I think before it needs to be recharged. I was wondering if it might be possible to use a LoRa transceiver instead for wireless communications, with a small GPS chip. If position updates were relatively infrequent, or occurred only at times of interest (e.g. when it is time to bring Tiggy in for the evening) it might be able to last a couple of months I think on a single coin cell. There would be some challenges in power delivery however, which would likely require a super-capacitor of some sort to be able to source enough current for times when the device is active.
I purchased a couple of LoRa evaluation boards, that feature the very small 1SJ chip from Murata. On one of them, I attached a small, flexible antenna, although I needed to re-solder an incredibly small 0-ohm resistor on the board to do it, and I'm afraid this connection is a bit less than ideal, since the wireless power level sent from the device tends to increase by about 5 to 10 dB whenever I push down on the resistor with my thumb. Nevertheless, I was able to hook up one evaluation board on the railing of our deck, with a short stubby antenna, and positioned the other evaluation board with the flexible antenna at various corners of our property, to see what kind of wireless power could be received on the deck. Here are the results:
The amount of wireless power received was disappointingly low, but I think the aforementioned issue with soldering the miniscule 0-ohm resistor could be to blame. I think using a better antenna on the base station, and elevating the base station somehow (maybe on top of our roof) would work a lot better. Trees, buildings, and other obstructions really wreck havoc with the wireless signal, so I think in some cases, there won't be much that can be done. Theoretically, the LoRa transceivers should be able to work over a distance of ~ 10 km, but that assumes both transmitter and receiver are mounted high up, well away from the ground, without any obstructions between them.
So for now, I'll keep at it, try to improve the wireless power throughput, and start looking at how to collect the GPS data. Tiggy is remaining mostly indoors now that it is December and usually too cold outside, so I'll have a few months at least to come up with something he can try out. 😀
