Table of Contents
In this paper we propose a technique based on off-policy evaluation and, in particular, importance sampling to estimate the impact on performance of modifications to finite-state machines.
The archive linked below contains the starting finite-state machines, the modified finite-state machines for the state pruning scenario and for the parameter variation scenario as well as the estimations produced using our technique and the measured performance. Data archive
The source code of the off-policy evaluation for finite-state machines generated with AuotoMoDe for this paper is available
Video of an experimental run where the swarm is controlled by one of the automatically generated control software instances. To limit its duration, the video runs at 4x normal speed.