Hopefully we can extract the key elements of the dynamics and this can serve as a platform for both desktop and on board simulation. But, I think we may end up with two levels of simulation depth, due to the complexity of the problem.

1. a desktop version (w/full state model)
   • full modeling of many surfaces, dynamic possibilities
   • no sensing limitations
2. on-board (reduced model for speed)
   • Simplified to account for general cases (surfaces, dynamic regions). Some less likely cases may be neglected in model. More limited by sensor info?

• Want desktop model for design/verification
• Want on board system ready for early prototypes
• Adjustable for evolution and different creatures (seems like a very good idea)
• Control integration into simulation seems like great idea (Adams currently has some diffiiculties with this).

-- AMcClung? - 10 Jun 2003