Nothing Special   »   [go: up one dir, main page]

Uncertainty

Download as pdf or txt
Download as pdf or txt
You are on page 1of 28

Artificial Intelligence

Uncertainty

Pham Viet Cuong


Dept. Control Engineering & Automation, FEEE
Ho Chi Minh City University of Technology
Content
✓ Probability: Brief review
✓ Robot localization problem
✓ Baysian filter
✓ Particle filter

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 2
Probability
✓ Probability that a random variable X has value x

✓ Example:

✓ Abbreviation:

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 3
Probability
✓ Probability density function
❖ Univariate normal distribution

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 4
Probability
✓ Probability density function mean vector covariance matrix
❖ Multivariate normal distributions

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 5
Probability
✓ Probability density function
❖ Probability of the random variable falling within a particular range of
values

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 6
Probability
✓ Joint distribution of 2 random variables
✓ Conditional probability

✓ Bayes rule

p(y) doesn’t depent on x

✓ Conditional independence

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 7
Localization Problem
✓ Indoor

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 8
Localization Problem
✓ Outdoor

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 9
Why localization problem?
✓ Fundamental problem in robotics
✓ Control noise

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 10
Localization Problem
✓ 1D localization ✓ 2D localization

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 11
Robot & Environment
✓ State

✓ Control actions

✓ Sensor measurements

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 12
Robot & Environment
✓ State evolution: state transition probability

Markov process: a process for which


predictions can be made regarding future
outcomes based solely on its present state (as
good as the ones that could be made knowing
the process's full history)

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 13
Robot & Environment
✓ Measurement model:

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 14
Robot & Environment
✓ Measurement model:

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 15
Robot & Environment
✓ Hidden Markov chain:

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 16
Belief
✓ Belief

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 17
Bayesian filter

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 18
Bayes filter
✓ Example

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 19
Bayes filter
✓ Example

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 20
Bayes filter
✓ No closed form solution in general
✓ Two approachs:
❖ Gaussian filters
❖ Nonparametric filters

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 21
Particle filter
✓ Represent the posterior bel(xt) by a set of random state samples

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 22
Particle filter
✓ Approximate the belief bel(Xt) by the set of particles Xt

✓ Particles: samples of a posterior distribution

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 23
Particle filter

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 24
Particle filter

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 25
Particle filter

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 26
Probability

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 27
Probability

Pham Viet Cuong - Dept. Control Eng. & Automation, FEEE, HCMUT 28

You might also like