MicroRaptor Begins

[JonHylands]

Well, my Bioloid kit finally showed up today, so MicroRaptor is on its way to being constructed.

http://www.bioloid.info/tiki/tiki-index ... icroRaptor

Click on "MicroRaptor Construction" to see the first couple pics...

- Jon

[billyzelsnack]

That's going to be quite the beast when it is walking around on its own. Awesome.

[DerekZahn]

Looking good Jon!

[JonHylands]

More progress...

http://www.bioloid.info/tiki/tiki-index ... nstruction

- Jon

[Pev]

wow very cool

[billyzelsnack]

"Both legs are done, and I've temporarily used the CM-5 box to join them (I don't have the real chassis machined yet)."

Why not keep it stock so others can easily build one too?

[JonHylands]

Because this is a research project for AI. It's going to have a wifi link to a PC, which will be controlling it. It will also have a pair of wireless cameras that will be providing stereo vision, and a 5 or 6 axis IMU added in.

Very little about this robot will be stock, other than the servos and brackets from the Bioloid kit.

See http://www.bioloid.info/tiki/tiki-index ... icroRaptor for details of the project in general.

If anyone wants to build one using the CM-5 box, nothing I am doing will prevent that...

- Jon

[JonHylands]

I've updated the MicroRaptor page, adding a couple movies of playing back a couple of simple motion sequences. This is not using the CM-5 or any of the Robotis software - it is using my serial interface, with custom control software written in Squeak Smalltalk.

http://www.bioloid.info/tiki/tiki-index.php?page=MicroRaptor

Click on the "MicroRaptor Movies" link...

- Jon

[JonHylands]

I've added a new page - screenshots. Right now it has a shot of the main GUI I'm developing for operating MicroRaptor.

http://www.bioloid.info/tiki/tiki-index.php?page=MicroRaptor+Screenshots

This GUI allows me to play with the servos in real time, play back sequences of motion, and provides a nice visualization of a sequence.

Next up is sequence capture/editing. The sequence editor will allow me to extract a single servo from a sequence, and edit its positions graphically, and apply smoothing functions as it moves. This will help a lot when building the initial versions of a walking gait.

- Jon

[DerekZahn]

Nice work! I'm starting to code a similar type of app for Bing. Are you planning for a method where you can view a recorded sequence so you can see what happened while testing an idea? That's what I'm coding now, including recording video from a webcam so I can relate debugging output and sensor readings to what actually happened during a trial.

You're making great progress!

[JonHylands]

Hi Derek,

I'm not sure what you mean by view a recorded sequence. The sequence visualizer on that window will show any sequence, whether it is generated or recorded. You'll be able to pull out a single servo from the sequence and view/edit it as an XY graph (time versus position), and be able to drag the data points around on the graph to make adjustments.

The webcam idea is interesting - I'll have to look at what that would take.

- Jon

[JonHylands]

Hi Derek,

I'm not sure what you mean by view a recorded sequence. The sequence visualizer on that window will show any sequence, whether it is generated or recorded. You'll be able to pull out a single servo from the sequence and view/edit it as an XY graph (time versus position), and be able to drag the data points around on the graph to make adjustments.

The webcam idea is interesting - I'll have to look at what that would take.

- Jon

[JonHylands]

Hi Derek,

I'm not sure what you mean by view a recorded sequence. The sequence visualizer on that window will show any sequence, whether it is generated or recorded. You'll be able to pull out a single servo from the sequence and view/edit it as an XY graph (time versus position), and be able to drag the data points around on the graph to make adjustments.

The webcam idea is interesting - I'll have to look at what that would take.

- Jon

[DerekZahn]

Well, here's what I meant.... an app like the one you posted some great screenshots to can be used to show what is going on while the robot is running, but could also be used to do a detailed analysis of what has happened in the past. Because it's something I'm working on at the moment I was curious how you are thinking of showing that data from the past -- how you correlate sensor readings with the internal controller program decisions and so forth -- and how you are thinking of comparing different runs... like if MR falls over in one trial but (with or without a change in coding) does not fall over in another trial.

I'm using the VFW API for my stuff by the way, which although Windows-specific and a bit out of date is fairly easy to deal with.

[JonHylands]

Derek,

Yes, that is all part of the plan. All sensor readings are going to be logged one way or the other - since the brain is running on a PC, I have lots of memory and lots of hard drive space.

I haven't really thought that far ahead though, in terms of what the UI will eventually evolve into. I typically write my user interfaces as I need them, adding features and refactoring things until they provide the functionality I need.

I use Squeak Smalltalk for all my programming, including the "brain" of the robot and all the user interfaces. It has its own built-in graphical user interface paradigm called Morphic, which can be extended and enhanced however needed.

- Jon