Thornleigh has stood on Lord Howe Island for over one hundred and thirty years.
In the late 1800’s, when Thornleigh was founded, the world was being rocked by the spread of railroads and telegraphs. Today robotics, miniaturised computing, and big data analysis are reshaping the way society functions.
Railroads never came to little Lord Howe Island. But robots have. Read on to find out what we've created and why we built it.
Right now there are two main ways to measure the amount of water in Thornleigh’s water tanks. Someone can walk up and knock on the side of a tank, or eyeball the floating indicators that poke out of the tank roof.
These methods are inaccurate and unreliable. They don’t produce records, unless someone writes their estimation down. Even then the data are unlikely to be readily accessible for analysis. Most importantly, they are boring. No one really wants to go and knock on a tank every single day.
Automating boring tasks means freeing humans to be more creative, and more happy. Yet humans are very effective workers. Even simple tasks like knocking on a water tank involve layers of complex actions and reasoning. We are self-powering, self-repairing, possess versatile limbs for interacting with our environment, and are able to quickly take instruction. Replacing my tank-knocking dad with a robot was going to be tricky.
There are a wide variety of commercially available water tank measurement products. After looking in to various options, I came to they conclusion that they were generally unappealing. They invariably involved compromises.
The most common problem with commercial solutions is that they don’t provide open interfaces to manipulate the data they record. They are designed to be used in closed, vendor-specific systems that force particular capabilities on the customer. They don’t allow flexibility. You will see what I mean when I describe what our prototypes do.
Each sensor unit sits on the underside of the water tank roof. Every few seconds, they transmit a very high frequency sound wave down into the tank. The sound bounces off anything remotely solid, including water. This core capability led me to give them the name ‘Soprano.’
By measuring the time it takes for the sound wave to return to the unit, we can determine how far away the water surface is. The speed of sound depends on temperature and humidity, so the sensor records those values too. Some simple mathematics yields an approximation of water volume.
Data from the Sopranos are compiled by a server on the farm. Every so often, the data are packaged up and transmitted via satellite to a server in Sydney for duplicate storage and processing. In the event of a failure of that server, any Soprano can take over and perform the role of compiler and transmitter.
The overall theme of these capabilities is one of a flexible, open, redundant, and robust system. It is a system that can now be extended to other robotic capabilities: Valve controllers, weather stations, soil saturation sensors, weed-killing-drones, and so forth. And the real kicker for a small farm? The components for each Soprano cost less than $100.
Data from the Sopranos can be analysed and manipulated both on the island, via our farm servers, or on the mainland via virtual servers in commercial data-centres.
The end result is twofold. First, we can analyse our water use in real time, anywhere on the planet, on any device. Second, other automated farm components can make use of water usage data in real time.
I alluded to the idea of automated valves above. Imagine this example: A valve at the main outlet of a tank notices that water levels are dropping faster than normal, so it infers that there is a leak downstream. It shuts off water flow without any human intervention.
Perhaps all this tinkering will end up being useful for Thornleigh. Regardless, it’s great fun. If you’d like to discuss any of this experimentation, say hi to me on Twitter. To receive news of Thornleigh’s progress, sign up to our newsletter.