[Robotgroup] Divide and Conquer!

Edwin Wise edwin at simreal.com
Thu Sep 13 13:17:41 PDT 2007


On 13 Sep 2007, at 2:37 PM, Mark Hinkle wrote:
>    For example: you built Odex-1 and you want Odex to lift its leg.  
> How
>    far? For how long?

Yah, depends on how you set your command structure -- something that  
will require some fair thought.  I would see the master brain deal  
with point-to-point issues and high-level pattern generation, where  
the reflex brain would do accel/decel work on the movement curve, low- 
level pattern generation (if there is such), and dynamic  
proprioceptive tasks to determine if things are going well, if there  
is resistance/conflict, and so on.

> What do you do if it cannot lift or lower its leg?

That's when the leg-state signals make themselves known!  Gotta close  
the loop to the big brain; tight fast closed loop at the leg reflex,  
and a higher-level but slower closed loop to the brain. Biology  
blazes the trail here.


>    ? How will you synchronize the legs for
>    "walking"? How fast? What direction? How far?

There's lovely work in gait pattern generation out there... in that  
case you would even want peer-to-peer communication, so it's a leg- 
generated pattern at the core, with high level moderation and peer to  
peer coordination.

Scared yet?


>    That is just the tip of the tip of the iceberg. You can extrapolate
>    from there easily. Distributed motion control is a GREAT idea.
>    Implementation can be tricky, though.

The transport mechanism can actually be pretty simple -- google for  
EtherCAT, it's a fairly elegant protocol that gives distributed,  
synchronized, simultaneous control across multiple devices in a  
subnet, designed for manufacturing, and includes the ability to have  
outputs, inputs, and in theory (based on its CAN roots) peer-to-peer  
communication.

Neat stuff.  I'm something of an expert, now, too.  There are other  
protocols, too, but I just have my hands in this one.

When I did Boris, I got the mechanicals and very basic distributed  
communication/control working, but then just ran out of time and  
energy so I never did do any good brain work on it.  More's the pity.



Edwin!




More information about the Robotgroup mailing list