Here is some code that works. For each, use gunzip (e.g. guncip
cabot3Jan16.tar.gz), then tar -xvf (e.g. tar -xvf cabot3Jan16.tar).
If you have problems, mail Chris.
- From section 3.1 the BrainScales agent
CABot1.py for BrainScales. It's
a txt file, but you can save it as .py. The only way I've had
this work is to login to Heidelberg. It's about the simplest
agent I can think of.
- Section 3.1 also describes a CABot1.
CABot1 tarred and gzipped.
Use gunzip cabot1Sep9.tar.gz, then tar -xvf cabot1Sep9.tar.
This is the September 9 version. Improved Planning with turn toward
both objects. Not well evaluated.
- Section 3.2 describes a CABot2. There are a few in
May 2017, some agents for the
neurorobotics platform. They run in nest and have been tested
on the version of the NRP I installed in May 2017. Untar the
file, and follow the readme.
The most sophisticated agent takes user commmands (from a text file),
and executes them with a turn toward the pyramid.
- Section 3.3
This is the FLIF simulator and CrystalSpace environment.
It is pretty likely that getting CrystalSpace to run will
be tricky, but the FLIF simulator and associated agent
should be relatively straight forward.
2017, tarred and gzipped.
- Section 3.4
CABot 3 from October 23rd,
2017, tarred and gzipped. This runs on Chris' spinnaker
48 chip board with Spinnaker and Nest. It does the full
range of tasks (explores and go before all four shapes)
pretty well. The components and partial testing are in the
subdirectories, cogmap, parser, plan, stateMachine, timer and vision.
- See
the Neuromorphic Agents that Learn page for
more code.
