Incremental Learning of Discrete Planning Domains from Continuous Perceptions

Luciano Serafini, Paolo Traverso

We propose a framework for learning discrete deterministic planning domains. In this framework, an agent learns the domain by observing the action effects through continuous featur…