Particle Filters
₹199 ₹599 67% off
Create free account to enrollLog inLifetime access to every lesson, the equations and the downloadable Python simulation. Your progress is saved to your account.
Kalman filters assume the belief is a single Gaussian bump. But sometimes the truth could be in two places at once: a robot in a symmetric corridor, a target that could be on either side of a sensor. Particle filters represent belief with thousands of weighted samples that can take any shape.
You will learn how the 1993 “bootstrap filter” made this practical, implement sequential importance resampling (SIR) from scratch and run the famous nonlinear benchmark where the EKF fails and the particle filter succeeds.
What you will be able to do
- Explain Bayes filtering with particles and weights
- Implement predict, weight, resample with systematic resampling
- Measure degeneracy with effective sample size
- Compare a particle filter with an EKF on a bimodal problem
Before you start: Course 1, basic probability (Bayes' rule) and Python loops and arrays.
Lessons
- The big idea: a swarm of guessesFree preview
- How it was invented and its first usesLocked
- Where it is used todayLocked
- The equationsLocked
- Worked demo and tuningLocked
- Build your own, step by stepLocked
- Python simulation labLocked
- Practice problems and quizLocked