regularize_default

ssapy.utils.regularize_default(particles, weights, num_particles_out=None, dimension=6)[source][source]

Perform particle regularization. This generates a perturbation from the particles’ original values to prevent singularity issues. From “Bayesian Multiple Target Tracking” (2nd ed.) p. 101-102

Parameters:
  • particles (numpy.ndarray) – Particles to reqularize. Each row is an nD particle

  • weights (numpy.ndarray) – 1D array of particle weight. Should be same length as number of particles.

  • num_particles_out (int | None, optional) – Number of particles to return, defaults to None. If not specified, will return the same number of particles as in the input particles.

  • dimension (int, optional) – Dimension of the parameter space for resampling. Assumes the first dimension columns of particles are the parameters to use. Any remaining columns are carried through without modification (e.g., time columns). Default: 6

Returns:

Deltas from original particles and their weights

Return type:

(numpy.ndarray, numpy.ndarray)