Socially assistive robotics aims at providing users with continuous support and personalized assistance, through appropriate social interactions. The design of robots capable of supporting people in heterogeneous tasks, raises several challenges among which the most relevant are the need to realise intelligent and continuous behaviours, robustness and flexibility of services and, furthermore, the ability to adapt to different contexts and needs. Artificial intelligence plays a key role in realizing cognitive capabilities like e.g., learning, context reasoning or planning that are highly needed in socially assistive robots. The integration of several of such capabilities is an open problem. This paper proposes a novel “cognitive approach” integrating ontology-based knowledge reasoning, automated planning and execution technologies. The core idea is to endow assistive robots with intelligent features in order to reason at different levels of abstraction, understand specific health-related needs and decide how to act in order to perform personalized assistive tasks. The paper presents such a cognitive approach pointing out the contribution of different knowledge contexts and perspectives, presents detailed functioning traces to show adaptation and personalization features, and finally discusses an experimental assessment proving the feasibility of the approach.

中文翻译：

---

社交辅助机器人行为适应的整体方法

社交辅助机器人旨在通过适当的社交互动为用户提供持续的支持和个性化帮助。能够支持人们执行各种任务的机器人的设计提出了一些挑战，其中最重要的是需要实现智能和连续的行为，服务的健壮性和灵活性，以及​​适应不同环境和需求的能力。人工智能在实现认知能力（例如学习，情境推理或计划）中扮演着关键角色，社交辅助机器人迫切需要这些能力。几个这样的功能的集成是一个开放的问题。本文提出了一种新颖的“认知方法”，将基于本体的知识推理，自动计划和执行技术相结合。核心思想是赋予辅助机器人以智能功能，以便在不同的抽象级别进行推理，了解与健康相关的特定需求并决定如何采取行动以执行个性化的辅助任务。本文提出了一种认知方法，指出了不同知识环境和观点的贡献，提出了详细的功能轨迹以显示适应性和个性化特征，最后讨论了实验评估，证明了该方法的可行性。