{science.games} How to solve it
The puzzle of a ferryman, a goat, a cabbage, and a wolf, can be solved by a symbolic model verifier. You can write the LTL (linear-time temporal logic) specification and the solution can be seen. It's a plannig problem. I have a weird feeling. Someone asked me what I have done over the past week, and I came up with "hey, I learned how to solve the infamous puzzle using NuSMV." He gave me a look, but that's not why I have that weird feeling. I have a weird feeling if such kinds of solutions can be applied largely in robot path planning. I have a weird positive feeling, but that's just me.
posted by Azad at
9:50 PM
3 Comments:
You may need some kind of extensions to the traditional planners. Most importantly, you need to handle uncertainty in your knowledge of the world. In a robot application, the result of an action is not apparent, and ... !
By
Anonymous, At
7/29/2006 6:09 AM
Actually, using a higher level of abstraction, a robotic task, can also be modeled as a reactive system, which is a bit different than nondeterministic systems (both are finite). This lies within a rather growing field of behaviour-based robotics and I believe uncertainties are included therein.
By
Azad, At
7/31/2006 6:15 AM
More to read on "reactive planning": http://en.wikipedia.org/wiki/Reactive_planning
By
Azad, At
7/31/2006 6:22 AM
Post a Comment
Subscribe to Post Comments [Atom]
<< Home