• A neuro-symbolic generative model for Automated Planning problems is proposed. • Our method generates valid, diverse and difficult problems, and reduces human effort. • Problem generation is formulated as a Markov Decision Process. • Generative policies are trained with Deep Reinforcement Learning. • Problem constraints are encoded via a novel semi-declarative. language In the field of Automated Planning there is often the need for a set of planning problems from a particular domain, e.g., to be used as training data for Machine Learning methods or as benchmarks in planning competitions. In most cases, these problems are created either by hand or by a domain-specific generator, putting a burden on the human designers. In this paper, we propose NeSIG ( Neuro-Symbolic Instance Generator ), to the best of our knowledge the first domain-independent method for automatically generating typed-STRIPS planning problems that are valid, diverse and difficult to solve. We formulate problem generation as a Markov Decision Process and train two generative policies with Deep Reinforcement Learning to generate problems with the desired properties. We conduct experiments on five classical domains, comparing our approach against handcrafted, domain-specific instance generators and various ablations. Results show NeSIG is able to automatically generate valid and diverse problems of much greater difficulty (6.8 times more on geometric average) than domain-specific generators, while simultaneously reducing human effort when compared to them. Additionally, it can generalize to problems more than twice the size of those seen during training.

In the field of Automated Planning there is often the need for a set of planning problems from a particular domain, e.g., to be used as training data for Machine Learning or as benchmarks in planning competitions. In most cases, these problems are created either by hand or by a domain-specific generator, putting a burden on the human designers. In this paper we propose NeSIG, to the best of our knowledge the first domain-independent method for automatically generating planning problems that are valid, diverse and difficult to solve. We formulate problem generation as a Markov Decision Process and train two generative policies with Deep Reinforcement Learning to generate problems with the desired properties. We conduct experiments on three classical domains, comparing our approach against handcrafted, domain-specific instance generators and various ablations. Results show NeSIG is able to automatically generate valid and diverse problems of much greater difficulty (15.5 times more on geometric average) than domain-specific generators, while simultaneously reducing human effort when compared to them. Additionally, it can generalize to larger problems than those seen during training.

This article is focused on how a general‐purpose hierarchical planning representation, based on the hierarchical task networks (HTN) paradigm, can be used to support the representation of oncology treatment protocols. The planning algorithm used is a temporally extended HTN planning process capable of interpreting such representation and generating oncology treatment plans that have been proven to support clinical decisions in the area of pediatrics oncology.

Workflow technology was developed in order to help tounderstand, coordinate and orchestrate the processes thatmake an organisation work. It is perfectly suitable for themanagement of an e-learning center, where a course managerhave to allocate the execution of a set of tasks amongthe existing human resources. This paper presents how tomake the most of an e-learning business model stored inXPDL, converting it into a planning representation, usingthe HTN paradigm. The subsequent goal is to obtain anaction plan that distribute the tasks needed for the creation and configuration of an online course to a set of workers.

The application of artificial intelligence planning techniques to real world problems is a very difficult task which usually implies overcoming expressiveness and efficiency handicaps. Expressiveness in order to cope with the great amount of knowledge involved in these problems and also to obtain results directly, or almost directly, applicable in the field of study. Efficiency to achieve a good course through the potentially enormous search spaces and, therefore, a satisfactory response capability. This work describes MACHINE, a manufacturing planning architecture which is able to obtain sequential control programs for manufacturing systems, the steps given to overcome these expressiveness and efficiency handicaps and how it is able to solve some real size problems.