This application generates fish trajectories from self-propelled agents bounded within a waterbody.
The agent's movement is driven by a set of forces acting on them which affect their velocity vectors.
The parameters controlling the forces are determined by the behavioural state, which is updated according to markov process.
The trajectories of the fish agents can be downloaded with labelled parameter states.
The motivation is this simulated data can be used to evaluate and train models classifying behavioural states from trajectories.
Visual feedback in the application is valuable for understanding how the states behave.
The simulation code is open-source.