DPRG List  

[DPRG] Is subsumption old hat?

Subject: [DPRG] Is subsumption old hat?
From: Randy M. Dumse rmd at newmicros.com
Date: Fri Sep 7 20:09:44 CDT 2007

Pete Miles wrote: Monday, August 27, 2007 2:18 PM
> As more and more sensor input is considered at 
> the same time, not only is a priority considered, but some 
> form of a weighting to the sensor input is applied.  How all 
> this is put together is different from developer to 
> developer, but when you break it all down, it is the same 
> thing, subsumption.

Wow. I don't know if I can go there. Subsumption in Brooks
implementation was pretty clearly a highest priority take all.
Also it was completely devoid of any state information. 

To suggest later voting systems or behaviors blending/merging
outputs were Brooksian Subsumption to me is a sacrilege. He was
very specific and unyeilding on those points.

> Subsumption just isn't the buzzword anymore.  Remember Fuzzy 
> Logic?  It was a buzz word 20 years ago, but isn't really 
> hyped anymore because people naturally apply it without 
> thinking about it. 

I remember Fuzzy logic, and I can say I've never done a Fuzzy
Logic implementation in an actual project. I understand most of
the Fuzzy concepts very well. (I also know I don't understand
defuzzifications.) So I don't think I've adopted Fuzzy Logic
into my programming style and just dropped the buzzword in doing

> Same with subsumption, people used it 
> without thinking about the word Subsumption.
> Subsumption isn't old hat.  It is just a buzzword applied to 
> something that was already being done.

Opinion noted. I don't really agree.

I do agree some portions of the original concept survive, and
are generally refered to as the Reactive Paradigm. But to say
they are Subsumption, no I can't go there.

BTW, Will and I had a very interesting discussion of what the
word Subsumption meant, and I have to say he moved my thinking
with his argument. But I don't think the meaning of the word,
and the specific context in how Brooks used it have survived in
many of the Reactive Paradigm implementations.


More information about the DPRG mailing list