Welcome, Guest. Please login or register.
Did you miss your activation email?

Username: Password:


Pages: [1]   Go Down

Author Topic: Pleo RB Programming environment  (Read 7288 times)

Kaeki

  • Hatchling
  • Posts: 3
  • Male
Pleo RB Programming environment
« on: March 20, 2012, 06:55:41 AM »

Hi there,

I'm a HRI research in UPC, Spain, and we're trying to use Pleo rb for 'social robots' experimentation with children at schools and hospitals' ER units. We had some experience with Ugobe's Pleo, but it seems that Pleo rb is a downgrade from the 'robot programming' side. We're generating a ' Pleo RB programming environment' as easy as possible in order to focus on robot interaction. Hence, I'll ask you about my first impressions:

* Pleo_rb 'internal values' (we're interested on them before and after interaction) can be checked and/or modified by using statsrb_v11, proberb_v1 and changerb_v1 (thanks a lot PleoPet)
* Pleo_rb USB connection is currently a nightmare. It seems that the USB BatteryAdapter (degers) should work. Could you send me any feedback? Works it? However, we cannot obtain pdf instructions. Could you help us?
* Windows7x64 seems that will not work. Will W7x32 works? I mean, is it possible to obtain drivers?
* finally, a lot of doubts about about dino-mite. Will dino-mite_v102 work with the previous configuration?

Yes, we have started to work with PrbDK and Myskit, but we are not so much interested in modifying behaviors, but to study how they change after robot interaction with children. Hence, we are looking for an environment that allows similar initial behaviors for all the children and easy to read/modify behaviors, avoiding multiple SD-card insertions and battery replacements.

Thanks a lot,
Kaeki
Logged


PleoPet

  • Pleontologist
  • * Posts: 162
  • Female
Re: Pleo RB Programming environment
« Reply #1 on: March 21, 2012, 09:01:57 AM »

>but it seems that Pleo rb is a downgrade from the 'robot programming' side.
I agree. The location of the micro-SD card inside the body makes it hard to swap for program testing.
The the location of the USB port, and lack of drivers make tools like Dinomite not work on the Pleo RB.
In addition, the old UGOBE model Pleos have the raw serial port accessible under the body (for direct access to wireless modules or other things) so you can avoid the USB hassles altogether.

---
>but we are not so much interested in modifying behaviors, but to study how they change after robot interaction with children.

Can you describe your setup and program goals in more detail?
When running do your Pleos need to communicate with a separate computer, or can they run stand-alone?
[You can still write programs on the non-RB UGOBE Pleo, and trivially port it to the PleoRB.]

Have you tried doing something similar on the non-RB UGOBE Pleos?

However it sounds like you are NOT interested in writing your own personality, but using the built in personality.
I don't know if that will work without significant customization of the personality (which is difficult) or rewriting the personality (which is also very difficult).

---
Additional comments:
>* Pleo_rb 'internal values' (we're interested on them before and after interaction) can be checked and/or modified by using statsrb_v11, proberb_v1 and changerb_v1 (thanks a lot PleoPet)
That only works if you use the default personality (or one with minor tweeks)

>* Pleo_rb USB connection is currently a nightmare. It seems that the USB BatteryAdapter (degers) should work. Could you send me any feedback? Works it? However, we cannot obtain pdf instructions. Could you help us?
Unfortunately that special build USB adapter didn't happen. http://bobthepleo.com/forums/index.php?topic=2329.0
You can make your own with a few jumper cables (extend the 3 battery leads), but it is not a good way to run the robot normally since it needs an external power supply.

>* Windows7x64 seems that will not work. Will W7x32 works? I mean, is it possible to obtain drivers?
The W7x32 drivers do work partially, but last I tried they didn't actually work as a serial port.

>* finally, a lot of doubts about about dino-mite. Will dino-mite_v102 work with the previous configuration?
DinoMite doesn't work on a PleoRB for a number of reasons. Even if it did, the difficult access to the USB port prevents you from using it for real PleoRBs.
Logged

Kaeki

  • Hatchling
  • Posts: 3
  • Male
Re: Pleo RB Programming environment
« Reply #2 on: March 25, 2012, 01:03:16 PM »

>but we are not so much interested in modifying behaviors, but to study how they change after robot interaction with children.

>>Can you describe your setup and program goals in more detail?

our main hypothesis is that using 'standard' rb_Pleo behaviors, we could associate changes in rb_Pleo's internal variables with user's interaction. In fact, this is the main proposal with pet robots, to evolve social variables according to interaction. Obviously, it would be wonderful to know how standard behaviors' logic is, or, at least, to work with a simplified version (no need for food, for instance). Programming from scratch a completely new behavior is out of our scope since we are more interested on interaction than programming. However it would be very interesting to inhibit some features in order ti simplify variables' relationships.

Another interesting point would be to monitor variables through some wireless element, as it was able with ugobe Pleo. A funny video was made available in youtube with bluetooth some weeks ago, but either, it is a fake or a 'soon improvement' by innvolab, not ready for the current version.


>>When running do your Pleos need to communicate with a separate computer, or can they run stand-alone?
[You can still write programs on the non-RB UGOBE Pleo, and trivially port it to the PleoRB.]

They can run stand-alone, but, obviously, it would be an interesting feature to have wireless communication.

>> Have you tried doing something similar on the non-RB UGOBE Pleos?

No, we made the Xbee hack on one of them and we were expecting that it would be possible to do in the new rb_Pleo. In fact, information from camera would be of interest, but there is not enough memory to save data, neither wireless communication is possible.

>>However it sounds like you are NOT interested in writing your own personality, but using the built in personality.
>>I don't know if that will work without significant customization of the personality (which is difficult) or rewriting the personality (which is also very difficult).

yes, a simplified default personality would be the most interesting one, but we can arrange with the built-in personality. Our initial programming problems are more in the variable-reading side.

---
>>Additional comments:
>>* Pleo_rb 'internal values' (we're interested on them before and after interaction) can be checked and/or modified by using statsrb_v11, proberb_v1 and >>changerb_v1 (thanks a lot PleoPet)
>>That only works if you use the default personality (or one with minor tweeks)

how to program minor tweeks without interrupt the main program? Which kind of small modifications could be done? Do you have some example that we can use to extrapolate? I read you were able to un - scary a Pleo when no-light was present. Is it very difficult to make this kind of modifications or inhibitions in the default behavior?

>>* Pleo_rb USB connection is currently a nightmare. It seems that the USB BatteryAdapter (degers) should work. Could you send me any feedback? Works it? >>However, we cannot obtain pdf instructions. Could you help us?
>>Unfortunately that special build USB adapter didn't happen. http://bobthepleo.com/forums/index.php?topic=2329.0
>>You can make your own with a few jumper cables (extend the 3 battery leads), but it is not a good way to run the robot normally since it needs an external >>power supply.

Thanks for your assessment.

>>* Windows7x64 seems that will not work. Will W7x32 works? I mean, is it possible to obtain drivers?
>>The W7x32 drivers do work partially, but last I tried they didn't actually work as a serial port.

:-(

>* finally, a lot of doubts about about dino-mite. Will dino-mite_v102 work with the previous configuration?
DinoMite doesn't work on a PleoRB for a number of reasons. Even if it did, the difficult access to the USB port prevents you from using it for real PleoRBs.

:-(

Thanks a lot, again, PleoPet for your reply. We are starting to work with statsrb_v11 and changerb_v1.
Logged

PleoPet

  • Pleontologist
  • * Posts: 162
  • Female
Re: Pleo RB Programming environment
« Reply #3 on: April 05, 2012, 11:31:46 AM »

Followup:

> how to program minor tweeks without interrupt the main program? Which kind of small modifications could be done? Do you have some example that we can use to extrapolate? I read you were able to un - scary a Pleo when no-light was present. Is it very difficult to make this kind of modifications or inhibitions in the default behavior?

It is tricky, and tedious since Innvo releases new versions, with only minor changes. I have to analyse these changes at a binary level.
Even though Innvo has released a programming kit (the PDK/PrbDK) with a few trivial source code examples, the default personality (PLEOPM) is "closed source".
Ideally Innvo would release their Pawn source code as part of the PDK.
This would let interested parties improve the personality, build on it, fix bugs, and do value added work like your research (without having to start from scratch).

--
Since we don't have source code, what I've done for my purposes is briefly described here:
I've written a few tools that create a work-alike version, under fair-use.
This creates a mostly compilable work-alike replacement module, in messy Pawn source code. It requires human tweeking to make it an exact replacement.
Once I get an exact replacement with my source code, then I can safely add new logic.
I was planning on a Pleo Plus addon that would make tweeking the default personality easier, but never got around to finishing it.

--
Even so, reusing the default personality is made more difficult because you need to understand how the moods and other variables (properties) in the personality are meant to work. Sometimes it is more difficult because the personality is not finished and still has bugs in it (bugs I've reported but Innvo doesn't appear to have fixed).

Bottom line: it would be ideal if Innvo would release the Pawn source code to the PLEOPM personality. You may want to ask them, especially if you are doing work that would give them good public relations.
If not, I still suggest for human interaction research, you will be better off writing your own personality (with the PDK).
« Last Edit: April 05, 2012, 11:34:55 AM by PleoPet »
Logged
Pages: [1]   Go Up
 

SimplePortal 2.3.5 © 2008-2012, SimplePortal