A versatile tracking methodology based on shadowing filters is Introduced. This approach is easily adapted to different settings and applicable to irregularly sampled data. It can be implemented to track a single particle, a system of multiple objects, or ng1d bodies. It IS computationally cheap and outperforms alternative established methods. Our tracking technique enables us to minimise measurement errors and reconstruct the full dynamical state space from limited information. Applying the proposed method on GPS data from pigeons in a flock allows us to model the collective dynamics based on deterministic rules resulting from two major mechanisms (global-local).
|Qualification||Doctor of Philosophy|
|Award date||11 Aug 2016|
|Publication status||Unpublished - 2016|