Main-Logo

The self-learning watering system

 

 

 

 

Our vision

Smart Gardener is an innovative system that integrates modern computing and the natural beauty of gardening.

Nowadays it is increasingly difficult to find the time to take proper care of your plants. Everytime you leave the house for vacation there is always the unwanted hassle of forcing unwilling neighbours to water them while you are away. Gardening should be something that gives you pleasure, not stress. Our product takes the stress and strain out of gardening.

Our goal is to create a system which will take care of them for you, watering them at the perfect moment, and leaving you free to enjoy them.

We want to solve problems, not create additional ones: that's why Smart Gardener is designed to be easy and intuitive right out of the box.

We also care about the environment. SmartGardener activates itself only when needed, saving water and electricity - and thus saving you money.

Smart Gardener is designed to use self-learning algorithms: in this way it is able to intelligently manage a plant watering system in an unobtrusive and efficient way, without requiring complex setup procedures and expensive accessories.

Its decision-making algorithms use a wide range of information to determine the best course of action: using weather forecasts, current weather, status of plants, and by learning from it’s previous actions, the system is able to make the best decision possible to efficiently and effectively water your plants.

Now you can fully enjoy your garden or balcony, without any of the hassle. SmartGardener: the self-learning watering system.

The project

Development areas

The project revolves around a central controller which is able to fetch data from the internet and a local network of sensors. The system uses available data to determine the most appropriate course of action. This is achieved by using smart-learning algorithms. Two main areas of development can be identified: the first one is concerned with the development of successful software able to control and monitor the irrigation system; the second is the design and implementation of the necessary hardware for the system. This includes a sensor network able to convey data from the plants to the controller, and an irrigation network to deliver water to the plants.

Zero configuration

The Smart Gardener system will be able to adapt to different users: some users might desire a "plug-and-play" system while others might prefer a deeply customizable system. Smart Gardener allows for both approaches by using an adaptive configuration. Out of the box a default mode is implemented to keep plants watered without any work by the user. Nonetheless the system allows customization of all its parameters, so that a more savvy user can take full advantage of the endless possibilities offered by Smart Gardener.

User base

In its current form the system is designed primarily for the domestic user. Its flexible algorithms could, however, extend its usability to most irrigation setups: the controller can drive any motor/pump/solenoid and is thus adaptable to pre-existing systems. Its extendable sensor network means that it is possible to gradually improve an existing Smart Gardener system over time, adapting it to individual needs.

Requirements and hurdles

The system has to be able to:

  • Manage a network of sensors, gathering and analysing its data
  • Control a water delivery system in an unobtrusive way
  • Obtain up-to-date weather forecasts for the current location of the device
  • Implement a zero-configuration environment by employing self-learning algorithms
  • Take into account external events (e.g. the plants have been watered by the user)
  • Compensate possible hardware failures by resorting to a fail-safe configuration
  • Present the data gathered to the user in a natural way
  • Offer the possibility of fine-tuning the system, especially for advanced users
  • Be easily extensible: sensors should be easily added to a pre-existing network
  •  

    The main issues we might have can be broken up into two main areas: software, and hardware.

    The possible software issues we anticipate center around the development of a smart-learning algorithm, the design of an effective user interface, and the implementation of the sensor network data analyzer.

    With respect to hardware the anticipated issues are related to the design of a robust and self-configuring sensor network and an effective way of balancing the irrigation system given that the only parameter that can be modified is the irrigation time.