Let's Make Robots!

Particle Filter Localization


I am reading and researching about particle filters. I understand doing a full implementation of a SLAM would not be suitable if we are using things like IR sensors, Sonars, etc. So instead of making a SLAM I decided to make a particle filter.

Doing some research, I have found out that lego people have built one. Here is the source code: https://wiki.qut.edu.au/download/attachments/103397109/nxt_particles.c?version=1&modificationDate=1302407038000

Basically, a map is defined. a set of way points are defined. the robot executes the way points, and does localization tru particle filters. there is a video of with and without filters of the same bot moving around. without filter it bumps into a wall and gets lost.

I have also found out about https://github.com/erhs-53-hackers/arduibot where it seems like an implementation of particle filter using c++ / arduino is done. however, there is no documentation on this one. I have read the code and I can see a particle filter is implemented, but I have no idea whether it will work as a project or not.


Any ideas and input greatly appreciated.







Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Not quite sure I'm understanding; is this some sort of software based algorithm to filter out certain sensor data in order to map and localize the robot location?

yes, it uses stoastic methods to get data from simple sensors, and localize the robot location. probabilistically.