一种支持个性化协调的服务机器人体系结构

A service robot architecture supporting individualized coordination

本文提出一种支持个性化协调的服务机器人体系结构(individualized coordination architecture,ICA).主要动机是通过自然语言人机对话获取用户的个人特性和其他信息,通过对这些信息进行自动推理和规划,实现利用个人特性的自动问题求解,并满足家庭环境对服务机器人的应用要求.本文着重介绍ICA的主要部件的功能及其相互衔接方式,描述任务规划的机制和实现手段,并通过一个实例说明在一个初步实现的原型系统中从自然语言到任务规划的完整过程.

A service robot architecture supporting individualized coordination is proposed. In this approach, user features and other information will be acquired through human-robot dialogue in natural languages. The robot reasons about and makes plans with the information, so that automatic, individualized problem-solving can be carried out under real world environments. We present the main modules and their interfaces of the architecture, the mechanism of task planning, and a case study which includes the entire working process from natural language processing（NLP） to task planning in an implemented prototype system, expected to work with humans in, say, ordinary home environments. Standard NLP techniques such as syntactic parsing have been implemented in the prototype system, together with semantic analyzing developed by the authors. With these NLP techniques, the user＇s commands and descriptions about the environment expressed in natural languages are transferred into logical forms in segmented discourse representation theory（SDRT） and finally in answer set programming（ASP）, which are ready for the task planning, through a 5-steps procedure. In task planning module, ASP technique is employed, making it possible and feasible for the robot to conduct task planning and reasoning about the actions and changes in a unifying way. With some good coding, according to the authors＇ experiments with a real home robot, the task planning module can work for some typical tasks in real-time. This work provides a basis for investigations into further challenges in individualized coordination.