TinySoar is an implementation of the Soar artificial intelligence architecture that is intended to run on a memory-constrained device, like a robot. Soar is a “real time” (as opposed to “off-line”) performance runtime that incorporates acting, planning, and learning in a rule-based framework.
TinySoar consists of two primary components. The first is a portable, light-weight runtime that implements the Soar decision cycle as a host for a Soar agent. The second is a Tcl extension that is used to create and debug Soar agents, and then export them into a format that can be compiled into the runtime component.
A sample embedding (in fact, the one that inspired the project!) is also included: alternative firmware for the Lego® Mindstorms RCX that embeds the TinySoar runtime. Using this firmware, you can control a Lego robot with a Soar agent. (Disclaimer: this site is neither endorsed by nor associated with Lego. I'm telling you this because every other Lego Mindstorms site does the same.)
tinysoar-0.3.2 runtime
supports the basic Soar decision cycle. Impasses are also
supported, although chunking is not yet implemented. Included is
the Tcl extension for developing and debugging agents, firmware
for the Lego Mindstorms RCX, minimal documentation, and sample
agent code.