The first step towards a decent comparison of the different approaches is to have a framework that includes them all. The Non-Markovian Reward Decision Process Planner, NMRDPP, is a platform for the development and experimentation of approaches to NMRDPs. it provides an implementation of the approaches we have described in a common framework, within a single system, and with a common input language. NMRDPP is available on-line, see http://rsise.anu.edu.au/~charlesg/nmrdpp. It is worth noting that Bacchus et al. [2,3] do not report any implementation of their approaches.


Sylvie Thiebaux 2006-01-20