Our project aims to help researchers who study ecosystems by being able to both simulate the ecosystem and animal behaviour, as well as help them simulate what information regarding the objective truth their sensors could yield.
Our product works by first having the researcher set up an ecosystem. This is done through JSONs, which are editable through our UI. Here, researchers can set up both the environment parameters, such as grid size, type of species and number of entities, as well as set up the species specifications, such as energy consumption, their behaviour in the shape of behaviour trees, etc. Through this UI you can also set up the icon for each species in the map, the colour (even on gradient depending on their energy levels), the overlay, etc. Then the simulation begins and the entities act based on ticks. The biodiversity can be followed through graphs and the different measures they present, such as Simpson index. Then with ML you can rebuild the environment and compare it to the ground truth (the simulation).
This allows researchers to model species behaviour and investigate how this can affect the population and its relation to other species. It also saves time for researchers by not forcing them to do trial and error with the sensors in real life, but to simply run a computer-based simulation.
Richard van Dijk, a LIACs software engineer.
We had weekly meetings on Wednesdays.
We had both a scrum master and a product owner.
The work was allocated in a voluntary way. High trust was put on each individual member to select as much as they were able to based on their skills and the time consumed from their previous and current tasks.
The main challenge was trying to balance the coursework with the client's objectives, as he was quite ambitious. Another challenge we overcame was the fact none of us had tocded in Julia, so it was a first time for all. However, it resulted very helpful as we all were able to learn a new programming language in the end.