麻省理工学院：《自制决策制造原则》英文版 Particle filters for Fun and profit

Particle filters for Fun and Profit Nick Roy December 1st, 2003 16410

Particle Filters for Fun and Profit Nick Roy December 1st, 2003 16.410

●。● veto Ix The Minerva Experience

Remove Video link

●● Robot localization o 3 canonical problems Position Tracking Global localization Kidnapped Robot problem o Not the mapping problem. We always assume we have a map To start without a map or model is the Simultaneous Localization and Mapping problem, or SLAM. Also known as Concurrent Mapping and Localization

Robot Localization | 3 canonical problems • Position Tracking • Global Localization • Kidnapped Robot problem | Not the mapping problem. We always assume we have a map. • To start without a map or model is the “Simultaneous Localization and Mapping problem”, or SLAM. Also known as Concurrent Mapping and Localization

●。。 Position tracking i take an action Where am I: now? State space Ⅰ was here

Position Tracking Where am I now? I take an action State Space I was here

●。G| obal localization State space i start with no position information Where am i?

Global Localization State Space I start with no position information. Where am I?

oo Kidnapped Robot Problem i take an get moved action without my knowledge 多 多 Can i tell rm not where i thought I was State space Ⅰ was here I think l am here Where am i now?

Kidnapped Robot Problem I take an I get moved action without my knowledge Can I tell I’m not where I thought I was? State Space I was here I think I am here. Where am I now?

●。● Assumptions without loss of generality o Metric, grid-based map o Pseudo-holonomic robot, operating in the plane o 360 field-of-view range sensor

Assumptions (without loss of generality) | Metric, grid-based map | Pseudo-holonomic robot, operating in the plane | 360° field-of-view range sensor