Particle Filter with Sensor Inputs

Good morning,

I wanted to know if someone could supply me with and example particle filter that uses some sensor input for localisation?

I would really appreciate any help on this.

Kind regards
Stefan

I wanted to know if someone could supply me with and example particle filter that uses some sensor input for localisation?

I wonder of you can explain what you are talking about? An explanation of what kind of sensor are you reading would be a good start.

What localization has to do with reading a sensor, and how a particle filter (whatever that is) will be used to perform localization would be interesting, too.

Hi, Stefan, can supply more information on what you want to do.
Such as what particles, what sensors and what do you want the controller to do with the information from the sensors.

Tom...... :slight_smile:

I wonder of you can explain what you are talking about? An explanation of what kind of sensor are you reading would be a good start.

What localization has to do with reading a sensor, and how a particle filter (whatever that is) will be used to perform localization would be interesting, too.

From Wikipedia: Particle filters or Sequential Monte Carlo methods are a set of on-line posterior density estimation algorithms that estimate the posterior density of the state-space by directly implementing the Bayesian recursion equations.

Hi, Stefan, can supply more information on what you want to do.
Such as what particles, what sensors and what do you want the controller to do with the information from the sensors.

I have the Arduino Robot (http://arduino.cc/en/Main/Robot) and added the Arduino Wifi-Shield (http://arduino.cc/en/Main/ArduinoWiFiShield). The aim of my project is to achieve that the robot can localize itself and navigate in a shopping mall by means of using the existing Wifi Signal Strengths as inputs to a particle filter. So my 'sensor' is the Wifi Shield. I have added an Ultra Sonic sensor, however, that is only to avoid obstacles and does not aid in the localisation process.

So what I am looking for is the code implementation of a particle filter. Please let me know if you require any further information.

Thanks for replys and help.
Stefan

by means of using the existing Wifi Signal Strengths as inputs to a particle filter.

Garbage in is going to equal garbage out.

First step is to assure yourself that you can get signal strength values AND that the values correlate to something meaningful.

Obviously there are Wifi Sginals occuring in the mall of interest and these obviously decay with increasing distance, thus can be used to localize a robot.

Obviously there are Wifi Sginals occuring in the mall of interest and these obviously decay with increasing distance, thus can be used to localize a robot.

That is if you assume that they all transmit the same power, and know where they are.
Using signal strength will not work because not all wifi transmitters are equal, plus path losses will be different.
Only triangulation would be feasible, and you need to know the wifi tx locations.

Tom...... :slight_smile:

Obviously there are Wifi Sginals occuring in the mall of interest

I'll concede that this will likely be true.

and these obviously decay with increasing distance

This is also true. However, they decay for far more reasons that just distance. Basing distance SOLELY on WiFi signal strength is not reasonable.

How will you get, reliably, the signal strength? The Arduino WiFi shield, as far as I remember, doesn't tell you anything about the signal strength until AFTER you connect to the WiFi network that you are (going to) measure the strength of. Clearly, running around in a public mall, you are NOT going to be able to connect to every WiFi network in the place. That I guarantee.

Nor are you going to deal with multipath issues. At the frequencies used radio waves behave much like light. There will be interference zones where the RF (radio) 'waves' will add and subtract, both Time, sign and amplitude. All of which are more related to reflections than absolute Losition relative to the transmitter. In a manner solar to "Ghosts" seen on old analog television receivers... not, young cricket, a linear change in amplitude that would indicate relative distance...

Doc

Just to let all of you know, it does work as planned. The robot can localize itself with an accuracy of maximum 1 m off from the actual position. This happens in a shopping mall, based only on WiFi signal strengths. Hardware used is the Arduino Robot and the WiFi Shield.

Regards.

Hi Stefan,

I am working on a similar project to track people indoors using wifi signal strength as an indicator of distance. I have used kalman to smoothen the fluctuations in the signal and have been able to achieve 2-3 m accuracy. Could you pls share your knowledge of how you used particle filter to achieve 1m accuracy. Thanks so much.

Regards,