pfilter::pfilter Namespace Reference

Classes
class	PFilter
Functions
def	likelihood
def	normalize_likelihood
def	predict
def	resample_uss
def	retFalse
def	retTrue
	Functions implementing a particle filter.
def	set_norm_likelihood
def	set_resample_uss
Variables
dictionary	counts = {}
tuple	new_particles = resample_uss(num_particles, normalized)
tuple	normalized = normalize_likelihood(particles)
int	num_particles = 1000
list	particles = [("4", 4), ("1",1), ("2",2), ("3", 3)]

---

Function Documentation

def pfilter.pfilter.likelihood	(		appearance_model,
			measurement,
			particle_set
	)

Evaluate using appearance model 

Definition at line 75 of file pfilter.py.

def pfilter.pfilter.normalize_likelihood

(

weighted_particles

)

Make all the particle weights sum up to 1 

Definition at line 138 of file pfilter.py.

def pfilter.pfilter.predict	(		motion_model,
			control_input,
			particle_set
	)

Predict using motion model 

Definition at line 63 of file pfilter.py.

def pfilter.pfilter.resample_uss	(		num_samples,
			particles
	)

    Universal stochastic sampler (low variance resampling)
    num_samples - number of samples desired
    particles   - pairs of (state, weight) tuples

Definition at line 86 of file pfilter.py.

def pfilter.pfilter.retFalse

(

args

)

Definition at line 29 of file pfilter.py.

def pfilter.pfilter.retTrue

(

args

)

Functions implementing a particle filter.

Note: To instantiate a particle filter you will need a motion and appearance model. Below are signatures and description of the motion and appearance models: (optional) motion.make_set: (int) -> list state motion.predict: (control, state) -> state appearance.weight: (measurement, state) -> double

Where what is considered a 'state' must agree between the motion and appearance classes.

Optional: The particle filter can be sped up by defining additional functions: * weight_partial - partial application * weight_set - any other optimizations

* predict_partial - partial application

Definition at line 26 of file pfilter.py.

def pfilter.pfilter.set_norm_likelihood

(

weighted_particles

)

Definition at line 150 of file pfilter.py.

def pfilter.pfilter.set_resample_uss	(		num_samples,
			particles
	)

    Universal stochastic sampler (low variance resampling)
    num_samples - number of samples desired
    particles   - pairs of (state, weight) tuples

Definition at line 108 of file pfilter.py.

---

Variable Documentation

dictionary pfilter::pfilter::counts = {}

Definition at line 168 of file pfilter.py.

tuple pfilter::pfilter::new_particles = resample_uss(num_particles, normalized)

Definition at line 165 of file pfilter.py.

tuple pfilter::pfilter::normalized = normalize_likelihood(particles)

Definition at line 161 of file pfilter.py.

int pfilter::pfilter::num_particles = 1000

Definition at line 164 of file pfilter.py.

list pfilter::pfilter::particles = [("4", 4), ("1",1), ("2",2), ("3", 3)]

Definition at line 160 of file pfilter.py.