TempAMLSI: Temporal Action Model Learning Based on STRIPS Translation

Abstract

Hand-encoding PDDL domains is generally considered difficult, tedious and error-prone. The difficulty is even greater when temporal domains have to be encoded. Indeed, actions have a duration and their effects are not instantaneous. In this paper, we present TempAMLSI, an algorithm based on the AMLSI approach to learn temporal domains. TempAMLSI is the first approach able to learn temporal domains with single hard envelopes, and TempAMLSI is the first approach able to deal with both partial and noisy observations. We show experimentally that TempAMLSI learns accurate temporal domains, i.e., temporal domains that can be used without human proofreading to solve new planning problems with different forms of action concurrency.

Action Model Learning from Noisy Traces: a Probabilistic Approach

“…In such situations, action model learning approaches that assume correct traces, such as (Yang, Wu, and Jang 2007;Cresswell and Gregory 2011;Cresswell, McCluskey, and West 2013;Stern and Juba 2017;Lamanna et al 2021), are not applicable. There are existing approaches designed for handling noisy traces (Rodrigues, Gérard, and Rouveirol 2011;Mourão et al 2012;Segura-Muros, Pérez, and Fernández-Olivares 2018;Grand, Pellier, and Fiorino 2020); however, we believe that the problem of action model learning from noisy plan traces cannot be considered fully addressed; especially when the noise level is particularly high, where current state-of-the-art approaches do not provide a satisfactory performance.…”

“…The problem of learning symbolic planning domains, aka action model learning, (Aineto, Celorrio, and Onaindia 2019;Grand, Pellier, and Fiorino 2020;Mordoch, Juba, and Stern 2023) consists of inducing a symbolic planning domain from a set of observations of the executions of domain actions. Most of the proposed approaches start from a set of plan traces (Arora et al 2018), i.e., sequences of (partial) descriptions of the states of the environment interleaved by the executed actions.…”