Let's Make Robots!

Subsumption Archecture

Hey all I found this very interesting idea in a book (The Robot Builder's Bonanza) for a way to code a robot. Instead of having a robot that merely goes straight till it hits something you have several "levels" or "layers" of code. The first one being an obstacle avoidance sorta thing, and then on the next level an aimless wander. This wander section of code which would ask the obstacle avoidance level if it was OK to move. If it was OK then the random wander would do its magic but if there was an obstacle then the obstacle avoidance would take over like a sort of reflex. You could have a layer over that for finding doorways that could "suppress" the second layer and get the bot to go in a specific direction (still asking the first obstacle avoidance layer if it is OK to move.

After reading this i was feeling that it sounded similar to the human hormone system (for those who are comfortable with biology)


Here is a link to the wiki article it doesn't go in depth but google is your friend!: http://en.wikipedia.org/wiki/Subsumption_architecture

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

I had heard the term before, but i didn't realize that's what i've been doing (the Fritz model). However, I've only recently begun to question its usefullness. The bot i'm currently working on has very complex movement layer, a less complex direction choosing layer "above" it, and a seemingly simple emotional layer "above" that, and the only thing working correctly is the movement. (Not asking for help, just making an observation)

- as a picaxe user, i didn't see much to offer in the very BS2-centric Robot Builders Bonanza, but maybe I'll have to give it another look. Calculon OUT.

This is based on a book about behavior programming.