Follow ME my robot

[benwasafish]

i am trying to construct a robot that can follow along behind me.
i  have been considering either using a bluetooth receiver so that i can use the bluetooth signal from my phone.
another thought was a radio transmitter in my pocket.
the only problem is, i have never used anything but IR. so i have no idea how to eve begin to do this. any help?

[airman00]

search the forum and google, its a very popular topic

then when you have an idea of what you want to do , we'll help you out from there
this project sounds very promising if you document it

[benwasafish]

I've searched out the forums and have found people wanting to do similar things, but the only information was based around IR or cameras.
google is not really telling me what i want to know. i found quite a bit of info on bluetooth devices etc.
is it possible to use a bluetooth transceiver to make a robot follow where the signal is strongest? (even if it isn't in line of sight)
also i would have to worry about interference. would it be worth trying to get them to pair?
i ask because i know that bluetooth is used to send and receive data, but i don't really want to do something that complicated.

I'll detail what i want to do for you all:

in my pocket i have something putting out a signal
on the ground somewhere else, there is a robot doing its usual obstacle avoidance, honing in on the signal.

Doesn't SEEM that complex.
the first thing i decided i needed was the whole signal circuitry, as this is where my knowledge is most incomplete.

i found this:
http://www.sagitron.es/data_sheet/Bluetooth/rfsolutions/DS360-2.pdf

it seems to me that i could use this. working on knowing exactly how at the moment.

[Commanderbob]

I don't know if you can tell which way the signal is coming from. You would probably need three receivers to compare the different signal strengths and then calculate the direction. There may be a much easier way to do this, I don't know, just throwing this idea out.
Justin

[benwasafish]

i was planning to have at least two receivers/sensors so that i can triangulate the signal as you said.
i just don't know what i need to do to read the signal.

[Admin]

 
http://www.societyofrobots.com/robot_faq.shtml#robot_track_person

[pomprocker]

I'd be interested in hooking up a webcam and using roborealm. Do you have any tutorials about that on the site here?

[airman00]

i;ve done it and it'll be part of the documentation for chives

[pomprocker]

does that require mounting a laptop on my robot? cause i was hoping to implement it on my $50 robot

[Admin]

I'd be interested in hooking up a webcam and using roborealm. Do you have any tutorials about that on the site here?

http://www.societyofrobots.com/programming_computer_vision_tutorial.shtml

In the middle mass section, I include a roborealm track red module you can download.

[pomprocker]

but how about actually hooking up a usb webcam to a robot? i'm not sure I understand how thats done (without mounting a laptop on the robot)

[Admin]

You'd need a laptop, or palmpilot like device with USB, to use a USB webcam . . .

Your other option would be use use a serial based camera, like the CMUcam, AVRcam, or Blackfin. These directly hook up to a microcontroller and do onboard processing for you . . .

[airman00]

does anyone know how to make a USB webcam into a serial cam?

anyone ever heard of something like this being done?

[ceruleanplains]

i'm using my webcam (USB) over ethernet.  that serial enough? : )

[airman00]

ever connect the webcam to a microcontroller over ethernet?

[ceruleanplains]

no - the submarine ROV has with two comm. lines instead of one - horribly inefficient and more costly, but I was looking for simplicity and implementation time.

1 line goes from the webcam to the laptop directly
1 line goes from the MCU to the laptop directly

I honestly couldn't figure out quickly how to efficiently dump the video stream data onto the MCU serial connection while streaming out sensor data and commands - because I haven't A.) hacked the line to observe the data stream from the camera... B.) I use OpenCV to play with the video and it doesn't fit MCU side so I figured why bother screwing with the I/O on a single COM port to try and filter it and bin it anyway.

If it were an analog camera one could do it using the figure shown on the rollette.com/rovrev2 blueprint but you'd still need to look at the data before you design any vision processing algorithms to assist.

regardless - i suppose it's possible to experiment with the CPU in the loop and use it as a bent communications pipe to the MCU.  that way you can sniff the data and control packets better and limit the I/O rate to the MCU.

[krich]

Looks like the analog camera is simply being sent to the computer.  No processing.  The schematics are a bit grainy, so I might have missed something.

[ceruleanplains]

that's correct the rovrev2 is direct analog to the computer with no processing.  on my application i used the process described above with the USB (digital) webcam on a seperate data line and the serial MCU connection on another.

the question is really if you can put in a link to go through the MCU to use the webcam locally.  I had the additional interest of getting the webcam data through the serial connection as well to save me a comm. line while the thread in general is interested in doing it wirelessly - but the concept is indentical... just w/o the wires.

[Oscillation]

On other websites and robot kits ive seen people use different sensors that you angle towards the front and it follows a moving object, then if it gets to close it drives backwards. However i've only seen this and read about it so i don't know how you'd go about it. It just means you wouldnt need to have some sort of signal in your pocket all the time

[airman00]

On other websites and robot kits ive seen people use different sensors that you angle towards the front and it follows a moving object, then if it gets to close it drives backwards.

thats only ranging and will work with the closest object only , but not a person that is far away.


I suggested RF tracking in another post somewhere over here where I asked about following a person. Search the forum for that because when you find that it will be a goldmine  of information.

[ceruleanplains]

There's a parallax project that someone called Boe-dar http://www.parallax.com/tabid/345/Default.aspx

if you don't mind the user having an active transmitter to have your robot follow him/her it might be an interesting application change on this concept.  otherwise... emitting RF to track a person's radar cross section could get into the world of dangerous.

The magazine Robot also had a previous article on a robot called Follow-me.  it was based on measuring the ambient temperature and then "tracked" targets with temperature within a specific threshold regime.  it had a linear array of IR receivers in different bands (if I remember right) to estimate temperature. 

[airman00]

The magazine Robot also had a previous article on a robot called Follow-me.  it was based on measuring the ambient temperature and then "tracked" targets with temperature within a specific threshold regime.  it had a linear array of IR receivers in different bands (if I remember right) to estimate temperature. 

yep I saw that

you can use the Devantech thermal array sensor

[happytriger2000]

mm..Following robot, I did a similar project back in year 2003 ( it was called Follow-Me Robot) by using IR beacon transmitter(held by user) and 2 receiver on the robot platform, it worked quite well, but not good enought compare to Shadow Caddy.
I posted a thread in another forum about shadowcaddy,
http://www.electro-tech-online.com/general-electronics-chat/95273-any-idea-how-thing-work.html
I think RFId is the tracking device used.