DPRG List  

[DPRG] Is subsumption old hat? .... and who cares whattheacademics and doing?

Subject: [DPRG] Is subsumption old hat? .... and who cares whattheacademics and doing?
From: Mr S szinn_the1 at yahoo.com
Date: Wed Sep 12 15:26:20 CDT 2007

I think there is a small problem with 'what the academics are  doing, and this is of course just MHO.

Academics and people who write for a living have a prestated goal. While that is not always the driving factor for research and books, it does print a lot of books and bring a lot of research grants.

Subsumption is well suited to a pic/AVR etc. type robot with two servo motors and a couple of sensors. What DPA did with J-Bot is pushing the boundaries of what hobbyists can and should be doing. Much of what is available in the robotics market is based on the consolidation of hobbyist practices and methods rather than truly being inventive. The little walker robots are clever, and fun to watch, but only really help in the next step. Their use is currently limited by power/weight ratios and such nuisances.

As the controllers available to hobbyists become more powerful, there will be more work on non-subsumption based architectures. If you want prime examples of the current paradigm, DARPA has been setting that challenge for all of us. Space and underwater robots, UAVs, and robotic extensions to all sorts of things mechanical. Think of a robotic gun mount that keeps all the solders INSIDE the humvee and still manages to keep the .50 cal on target. 

There are huge advances in wireless communications (RFID zigbee, WLANs, and automotive comms) for sensors and for data. If you can overlook the flaming laptops, there are huge advances in power technology. DPA has shown that it is possible to get hold of Inertial navigation sensors andput them to use. 

So, when there is curiosity, and desire, newer technology will be put together on a robot, perhaps one that mows the lawn in an organized manner, or one that vacuums the carpet AND monitors the house for security.. there are lots of things that can be done, but it requires tighter sensor fusion, and more processing and memory capabilities than are currently used by hobbyists... 

So, My vote for the next big thing... hmmmm whatever makes hobby robotics fun, and is affordable AND adds more capabilities in a structured way. See sensor fusion, stateful memory, plus subsumption, and advanced technology improvements on the necessary hardware bits.

More information about the DPRG mailing list