David Olmsted
06-17-2008, 03:47 PM
The first major expansion of this software will include the first attempt at a virtual 3D environment having real physics where users can create simple animals (think trilobites and snails), insert a simple brain, and test integrated behavioral responses. Yes, plopping brains into creatures designed from Spore's creature creator software would be awesome but I don't see anything that sophisticated in the near future. :)
Also I want to make neural leakage an adaptive element (a neuro-modulated element) since that neural parameter dominates the control of the Definitional Uncertainty mode. Also the user interface needs improving so it can handle larger circuits. Something similar to a 3-D cut and paste or an enforced modular framework.
The expansion after that will include the ability to use multi-core computers. Right now the software is not threaded and only runs on one processor. Microsoft has a development team doing some great work on code that will dynamically allocate program loops between cores. The software for that should be ready by the time this expansion is ready to ship. Brain circuits are inherently parallel so multi-core computers are coming along at just the right time or perhaps Real Simulated Intelligence is coming along at just the right time to make use of multi-cores. This expansion will have to be based upon the .NET 3.5 framework (the present software is based upon the .NET 2.0 framework).
After the addition of multi-core capability with its added simulation speed will be the time for developing a real world interface for autonomous robots. Also the extension of the brain simulation program to clusters of computers should be straight forward by then as well. Once cluster computing is possible then no limit exists on the size of the brain that can be simulated.
Of course, all the above estimates are subject to change since one can never really predict software development unless on has actually coded similar projects.
Also I want to make neural leakage an adaptive element (a neuro-modulated element) since that neural parameter dominates the control of the Definitional Uncertainty mode. Also the user interface needs improving so it can handle larger circuits. Something similar to a 3-D cut and paste or an enforced modular framework.
The expansion after that will include the ability to use multi-core computers. Right now the software is not threaded and only runs on one processor. Microsoft has a development team doing some great work on code that will dynamically allocate program loops between cores. The software for that should be ready by the time this expansion is ready to ship. Brain circuits are inherently parallel so multi-core computers are coming along at just the right time or perhaps Real Simulated Intelligence is coming along at just the right time to make use of multi-cores. This expansion will have to be based upon the .NET 3.5 framework (the present software is based upon the .NET 2.0 framework).
After the addition of multi-core capability with its added simulation speed will be the time for developing a real world interface for autonomous robots. Also the extension of the brain simulation program to clusters of computers should be straight forward by then as well. Once cluster computing is possible then no limit exists on the size of the brain that can be simulated.
Of course, all the above estimates are subject to change since one can never really predict software development unless on has actually coded similar projects.