The task of supporting a human operator to understand generated plans, and to explore the plan space, are important problems in automated planning. In this work, we consider the problem of plan explainability and plan space exploration in underwater autonomous vehicle missions. In this context, concepts that are useful for querying the system, such as distance and duration, will not necessarily map directly onto components of the planning model, such as actions. To overcome this difficulty, we focus on an important substructure of these problems: the multi-agent spatial-temporal (MAST) structure. Using this structure, we define a collection of model extensions, which include additional concepts relevant to the MAST structure. We then consider the problem of user-guided plan space exploration, and identify useful query types in this domain, including user queries based on numeric functions. These queries can make use of the extended model, allowing the user to directly reference the new concepts. In an empirical study, we demonstrate the use of the new structure within queries, and compare the new query types in our target domain, and in benchmark domains with the MAST structure. Finally, we report on a qualitative user study, where we investigate the use of these new structural concepts in underwater autonomous vehicle scenarios. Our study indicates that the extended concepts can be used in user queries and agent responses, enabling the user to better communicate their intent in shaping mission objectives, and supporting explanations with more relevant information.

In autonomous vehicle mission planning, supporting human operators to understand and influence the decision-making process is crucial for building the operator’s trust and establishing effective collaboration. However, it has been observed that human and agent representations will typically not align. As a consequence, concepts that are useful for effective human-agent communication, will not necessarily feature in the agent’s representation. Focusing on specific spatial-temporal concepts, we define automatic model extensions, which can introduce these additional concepts. We report on a qualitative user study, where we investigate the use of these new structural concepts in underwater autonomous vehicle scenarios. Our study indicates that the extended concepts can be used in user queries and agent responses, enabling the user to better communicate their intent in shaping mission objectives, and supporting explanations with more relevant information.

The software submission presents a fully working artificial cognition model, which controls a NAO social robot. The model was specifically designed to control a socio-affective companion robot for use in a medical setting. It was deployed using embedded hardware: a Raspberry Pi 4B and a Jetson Nano Board, and an external RGB-D camera. Based on the ROS operating system, this software package includes components for social signal processing, behaviour selection, affective behaviour rendering, and a web-based user interface. The robot's behaviours are selected by a planning system, which generates the robot's behaviours based on the state of the interaction, the progress of the medical procedure, and the user's affective state. The system has been tested in simulated environments and is currently being used in two clinics to perform a usability test and will subsequently be used to carry out a series of clinical trials

Narrative Generation has attracted significant interest as a novel application of Automated Planning techniques. However, the vast amount of narrative material available opens the way to the use of Deep Learning techniques. In this paper, we explore the feasibility of narrative generation through self-supervised learning, using sequence embedding techniques or auto-encoders to produce narrative sequences. We use datasets of well-formed plots generated by a narrative planning approach, using pre-existing, published, narrative planning domains, to train generative models. Our experiments demonstrate the ability of generative sequence models to produce narrative plots with similar structure to those obtained with planning techniques, but with significant plot novelty in comparison with the training set. Most importantly, generated plots share structural properties associated with narrative quality measures used in Planning-based methods. As plan-based structures account for a higher level of causality and narrative consistency, this suggests that our approach is able to extend a set of narratives with novel sequences that display the same high-level narrative properties. Unlike methods developed to extend sets of textual narratives, ours operates at the level of plot structure. Thus, it has the potential to be used across various media for plots of significant complexity, being initially limited to training and generation operating in the same narrative genre.

AI Planning has been shown to be a useful approach for the generation of narrative in interactive entertainment systems and games. However, the creation of the underlying narrative domain models is challenging: the well documented AI planning modelling bottleneck is further compounded by the need for authors, who tend to be non-technical, to create content. We seek to support authors in this task by allowing natural language (NL) plot synopses to be used as a starting point from which planning domain models can be automatically acquired. We present a solution which analyses input NL text summaries, and builds structured representations from which a pddl model is output (fully automated or author in-the-loop). We introduce a novel sieve-based approach to pronoun resolution that demonstrates consistently high performance across domains. In the paper we focus on authoring of narrative planning models for use in interactive entertainment systems and games. We show that our approach exhibits comprehensive detection of both actions and objects in the system-extracted domain models, in combination with significant improvement in the accuracy of pronoun resolution due to the use of contextual object information. Our results and an expert user assessment show that our approach enables a reduction in authoring effort required to generate baseline narrative domain models from which variants can be built.

Explainable AI is becoming an area of key focus in Artificial Intelligence. Within Automated Planning (AP) the area Explainable Planning (XAIP) focuses on explanations of the planning process. The relative transparency and flexibility of the planning process have been identified as key aspects suggesting that AP is well positioned to make an important contribution in Explainable AI [8]. However, there is still a wide gap between explanations that can be directly extracted from AP models and effective explanations. There are a growing number of frameworks that are considering the problem from both the user side, where it is interesting to understand the form that an explanation might take; as well as the planner side, which must be able to explain various decisions and related properties. However, approaches have focused on single domain settings, where substantial domain specific content is produced, or at a general level, where only abstract planning concepts can be used. We aim to develop an abstraction layer that sits between these and exploits the often overlapping concepts and structures that exist between many planning domains. We propose exploiting domain analysis techniques in order to identify common roles and generic problem structures (GPSs). By attaching the concepts used for explanation to these structures we can exploit the contextual information supported by the structure, and also reduce the burden of constructing explanations in domains where these structures exist. In this work we explore the opportunities for exploiting GPSs in XAIP.

The visualization of procedural knowledge from textual documents using 3D animation may be a way to improve understanding. We are interested in applying this approach to documents relating to patient education for bariatric surgery: a domain with challenging textual documents describing behavior recommendations that contain few procedural steps and leave much commonsense knowledge unspecified. In this work we look at how to automatically capture knowledge from a range of differently phrased recommendations and use that with implicit knowledge about compliance and violation, such that the recommendations can be visualized using 3D animations. Our solution is an end-to-end system that automates this process via: analysis of input recommendations to uncover their conditional structure; the use of commonsense knowledge and deontic logic to generate compliance and violation rules; and mapping of this knowledge to update a default knowledge base, which is used to generate appropriate sequences of visualizations. In this paper we overview this approach and demonstrate its potential.