Dynamical trajectories that emerge spontaneously for agents moving according to FSM, i.e. maximising their future accessible states. Here we use a standard set of parameters N = 50, Ƭ = 4, ns = 40, Δθ = 15°, ν0 = 10, Δν = 2. The agents are initially placed at random in a 100 × 100 box (all lengths in units of the particle size) and aligned with the x-axis up to a Gaussian-distributed noise on each agents orientation, with standard deviation σ = Δθ. When modelling future trajectories each agent here assumes that all other agents will continue to move ballistically, i.e. at the same orientation at nominal speed ν0.