Multi-Hypothesis Task Planning: integrating temporal AI planning and semantic world modeling for AUV inspections in unknown environments

Autonomous Underwater Vehicles (AUVs) have become fundamental tools for marine scientists and industries to explore and monitor underwater areas. Nonetheless, the collected data quality is not guaranteed: AUVs passively store the sensor acquisitions that are then analyzed offline after their recovery by human operators in charge of identifying and localizing the so-called Objects of Potential Interest (OPIs). As a consequence of these statements, the marine community has sought robots able to meaningfully perceive and model the surroundings while autonomously selecting and performing the requested tasks, which are the foundations of fully autonomous vehicles. Motivated by such considerations, the novel Multi-Hypothesis Task Planning (MH-TP) architecture has been designed to integrate temporal Artificial Intelligence (AI) based planning with semantic world modeling techniques toward fully autonomous AUV inspections in unknown environments. The proposed architecture extends the standard AI planning methodologies relying on the Problem Domain Definition Language (PDDL) by dynamically varying the PDDL problem with the probabilistic, semantically enriched objects that the World Modeling (WM) itself evaluates. The developed methodology has been validated with realistic simulations made by means of the UUV Simulator, where a dynamic model of FeelHippo AUV was implemented alongside the requested perception and payload devices.